Hand Gesture Recognition Robot

The

widespread Hand Gesture Recognition dataset HaGRID – HaGRIDv2. We cover 15 new gestures with conversation and control functions, including two-handed ones. Building on the foundational concepts proposed by HaGRID’s au- thors, we implemented the dynamic gesture recognition al- gorithm and further enhanced it by adding three new groups of manipulation gestures. The “no gesture” class was diver- sified by adding samples of natural hand movements, which allowed us to minimize false positives by 6 times. Com- bining extra samples with HaGRID, the received version outperforms the original in pre-training models for gesture- related tasks. Besides, we achieved the best generalization ability among gesture and hand detection datasets. In addi- tion, the second version enhances the quality of the gestures generated by the diffusion model. HaGRIDv2, pre-trained models, and a dynamic gesture recognition algorithm are publicly available.

Ntroduction

Hand gestures, as natural and intuitive expressions, effectively reflect emotions and facilitate communica- tion .

Their Ability To Convey Messages Quickly

makes gestures invaluable for Human-Computer Interac- tion (HCI) . Thus, the development of Hand Gesture Recognition (HGR) systems has the potential to signifi- cantly enhance user interfaces in various domains [27, 65]

Such As Robotic Control , Driver Assistance , And

medicine [14, 63, 68] for touchless interaction. The pro- posed research aims to develop a comprehensive HGR sys- tem for video conferencing and home automation de- vices [5, 7, 12, 17, 64].

The System Should Enhance Par-

ticipants’ communication, enable remote control of device functions [26, 66], allow the manipulation of various ob- Figure 1. The 15 outlined in red new gesture classes added to HaGRID’s 18 ones (“inv” stands for “inverted”).

jects on the screen, and activate different platform fea- tures . Considering the described application area, the system should enable intuitive operation through easy-to- demonstrate functional gestures, offer instant feedback, and efficiently operate on resource-constrained edge devices.

Neural networks have recently become the primary com- ponent of HGR systems [28, 45,46] and action recognition in general [11, 32, 59].

Arge Datasets Aligned With Sys-

tem constraints are required to train a model resilient to real-world conditions. Based on the requirements described above, the appropriate dataset should include a range of ges- tures categorized as manipulative (e.g., clicking, swiping, or zooming the screen), control (e.g., taking screenshots), and conversational (e.g., expressing approval, disapproval, love, anger, regret) . The first category usually implies dy- namic gestures, while the last two involve static ones. Thus, the dataset should encompass static and dynamic gestures (i.e., videos rather than images) to cover all three categories effectively. The HGR model should efficiently provide real- time inference on the CPU while remaining lightweight, as it should be integrated into the resource-constrained device.

Compliance with such requirements significantly restricts the choice of neural network architectures capable of oper-

Arxiv:2412.01508V1 [Cs.Cv] 2 Dec 2024

ating in the temporal dimension. Most existing gesture datasets are limited in several senses: some overlook functional and commonsense ges- tures, and others only cover static or dynamic gestures. The most suitable dataset, HaGRID , includes various con- versational and control gestures and proposes an algorithm for dynamic manipulative gesture recognition. The algo- rithm required static gestures to construct dynamic ones, satisfying the conditions of the lightweight model. The au- thors pursued the goal of creating the HGR system for home automation and video conferencing services. There- fore, all chosen gestures are straightforward and perform a specific function. However, HaGRID lacks support for es- sential device interaction gestures, such as clicking, zoom- ing, taking screenshots, and cursor control. Besides, the extra “no gesture” class contains too homogeneous hands, provoking the model to predict false positives. In this re- gard, we decided to expand the dataset with new gestures to support the development of dynamic gestures, introduce a new “no gesture” class, and enhance the dynamic gesture recognition algorithm.

It was decided to expand the HaGRID dataset by adding new classes. This paper introduces HaGRIDv2, the second version of the HaGRID dataset, designed to enrich the func- tionality of HGR systems for video conferencing and home automation. The contribution of this paper is three-fold:

• Hagridv2 Incorporates 15 New Gesture Classes

(Fig. 1) performing control and conversational func- tions. The cross-dataset evaluation experiments con- firmed HaGRIDv2’s best domain generalization ability among gesture detection datasets (see Tab. 3).

• The “no gesture” class is upgraded relative to Ha- GRID’s by incorporating domain-specific natural hand positions allowed to minimize false positives by 6 times (see Fig. 4).

• We Extend Dynamic Gesture Recognition Algorithm

capabilities, developing swipes, clicks, zooms, drag- and-drops, and other manipulative gestures (see Fig. 5 in the suppl. materials).

The dataset and the extended algorithm for recognizing dynamic gestures are publicly available12 under the modi- fied Creative Commons CC-BY 4.0 license.

HaGRIDv2 was designed to solve the gesture detection task. However, the proposed paper also inspects the ap- plicability of HaGRIDv2 to address other gesture-related tasks: gesture full-frame classification, hand detection, and text-to-image gesture generation (see Fig. 6 in the suppl.

materials for the difference between the tasks). Although

Https://Github.Com/Ai-Forever/Dynamic_Gestures

HaGRIDv2 does not focus on hand detection, it performs well on the standard benchmarks (Tab. 4). Fig. 3 shows that the gesture bucket expansion improved the HaGRID pre- training abilities in the gesture tasks. Also, Fig. 7 in the suppl. materials demonstrates the HaGRIDv2’s capabilities to address the issue of diffusion models producing anatom- ically incorrect gestures.

Related Work

There are a variety of HGR datasets with gestures cat- egorized based on their applications [24, 58]: sign lan- guage , control [9, 21, 23, 50, 57, 61, 67], conversa- tional [4, 10, 21, 23, 38, 41, 42, 50] and manipulative ges-

Tures . The Proposed Research Aims To

develop an HGR system for device control and video con- ferencing, where manipulative, control, and conversational gestures are essential. Therefore, the system should recog- nize both static and dynamic gestures, which are reviewed in this study. Sign language recognition datasets are ex- cluded because their gestures are unsuitable for performing the described functions.

Dynamic gesture datasets are typically annotated for ac- tion recognition, classifying entire video sequences. In con- trast, static gesture recognition can be achieved by solving various tasks, including gesture detection and classification (see Fig. 6 in the suppl. materials for difference), hand ges- ture keypoint estimation, and gesture segmentation. How- ever, classification labels are impractical in multi-person frames, keypoints can stick together when the person is far away, and segmentation masks are excessive and costly.

Additionally, only third-person data are suitable for video conferences and device control. The overview of the related HGR datasets is organized as follows. Sec. 2.1 discusses datasets relevant to the de- vice control task, while Sec. 2.2 examines datasets related

Grid , Lared , Ouhands , Hands ,

and SHAPE datasets are relevant for the required anno- tations. Each dataset has limitations that can affect its suit- ability for developing a reliable HGR system for real-world conditions.

The Lared And The Ouhands

datasets include images captured from close distances, mak- ing them unsuitable for training models operating over larger spaces. Besides, there is no access to the LaRED

And Hands , Constructed With Only 23 And 5 Sub-

jects, respectively, are inappropriate for developing a robust HGR system. In addition, even the most diverse datasets on classes lack gesture variety for control and conversational purposes. So, while the SHAPE dataset lacks enough

Manually

Table 1. The main parameters of the most popular gesture datasets. “+1” in the third column means the dataset contains an extra class “no gesture”. “-” in some columns means information was not found. * — keypoints prepared by using the MediaPipe hand model. control gestures, the HaGRID suffers from a deficiency of conversational ones, containing only “like” and “dislike” emotional gestures. Since emotional gestures are essential for video meetings, such an omission restricts the overall functionality of the HGR system.

Dynamic Gestures. Datasets such as [6,9,34,40,46,55, 57,61,67] are the most relevant for dynamic gesture recog- nition in device interaction and video conferencing. There

Are Only Chairgest , Jester , And Ipn Hand

datasets meeting the described requirements about the ex- istence of functional gestures and third-person view. How- ever, these datasets focus mainly on manipulative dynamic gestures and lack the necessary static conversational and control gestures. This gap highlights their insufficiency in covering the full spectrum of gestures needed for device control and video conferencing.

Although there is no entirely suitable dataset, the Ha- GRID dataset encompasses control and conversational ges- tures while simultaneously allowing the recognition of dy- namic manipulative ones utilized proposed in algo- rithm. Therefore, we decided to make some changes and create a second version of HaGRID, adding new gestures and diversifying the “no gesture” class.

Other Tasks

HaGRIDv2’s design, with one gesture per frame, sup- ports full-frame classification, which is ideal for single- user interactions with personal devices. Additionally, Ha- GRIDv2 includes a wide variety of hand postures, including complex ones and natural hand positions, making it well- suited for hand detection tasks.

Hand Gesture Classification.

All The Static Datasets

for object detection reviewed below can also be used for image classification. In addition to them, there are NTU

Hands (See Tab. 1). However, These Datasets Also

have disadvantages in solving the human-computer inter-

Action Problem. The Ntu Hgr , Senz3D , And

Kinect Leap datasets contain 100-140 samples per ges- ture class, which is not enough to build a sufficiently high- quality model (see ablation study in ).

N Addition,

Senz3D and Kinect Leap scenes are homogeneous, and the gestures are too close to the camera. The SIT-HANDS dataset was made heterogeneous in subjects, lighting condi- tions, and background. However, the dataset has only 2,800 training frames and contains a relatively small variety of gestures, severely limiting the system’s functionality.

There Are Hand , Egohands , Human-Parts ,

TV-Hands , ContactHands , BodyHands . Hands detection datasets may differ in the annotation type: some use horizontal aligned bounding boxes, while oth- ers use oriented bounding boxes. Since most well-known detection models work with horizontally aligned bound- ing boxes, this type is preferred, making HAND , TV- Hands , and ContactHands unsuitable.

Datasets vary in the number of people, subjects, res- olution, and scenes.

Egohands Is Intended For First-

person gesture detection and segmentation across various

Backgrounds, While Bodyhands And Human-Parts

offer 14,000 and 15,000 diverse third-person samples, re- spectively. Due to their heterogeneity, these datasets are valuable benchmarks for hand detection, but they focus only on natural hand postures and do not include specific ges- tures or complex finger positions. By addressing some key limitations in existing datasets, HaGRIDv2 offers improved support for these areas.

The Following Key Updates:

Figure 2. The key statistics of HaGRIDv2. (a) Image resolution distribution showing the scatter of image dimensions; (b-d) Distribution of subjects in the training, validation, and test sets, respectively; (e) Bounding box area distribution; (f) Brightness distribution; (g-i) Age and gender distributions of subjects, received automatically by MiVOLO neural network; (j) Racial distribution of subjects, received automatically by FairFace neural network.

• We added “holy”, “heart” (in two variations, see Fig. 1), “middle finger”, and “gun” as emotional con- versational gestures used during the conversation, and “three3” for extra functions.

• We expanded the range of control gestures by one- handed “thumb index”, “grip”, “point”, “pinkie”, and

“Grabbing”, And Two-Handed “Thumb Index2”, “Time-

out”, “take photo”, “xsign”. • Some of added static gesture classes were designed to enable the extension of the dynamic gesture recogni- tion algorithm by developing such gestures as “drag

And Drop”, “Click”, “Zoom In”, “Zoom Out” And New

variations of swipes. In total, 15 new static gesture classes and four groups of dynamic gestures were developed to cover various device functions (see Fig. 1 and Fig. 5 in the suppl.

materials). • The HaGRIDv2’s “no gesture” class includes a broader range of natural hand positions (e.g., relaxed hands near the face, holding a cup, natural gesticulation), whereas HaGRID is limited to a single relaxed hang- ing hand posture (see Fig. 8 in the suppl.

Such natural movements were specifically selected as the most commonly used gestures during video confer- ence meetings.

Ataset Creating Pipeline

The data creation pipeline almost followed the one pro- posed by the original HaGRID authors to maintain the consistency of HaGRIDv2’s data distribution. Also, such a pipeline allows us to collect heterogeneous samples in large volumes. We used the same crowdsourcing platforms such as Yandex.Toloka3 and ABC Elementary4 and instructions for crowdworkers through mining, validation, and filtration steps, described in Dataset Creating Pipeline in . In the mining stage, crowdworkers capture photos with a speci- fied gesture under controlled conditions. In the validation stage, images are reviewed to ensure they meet the criteria, and only correctly executed photos are retained. Finally, the filtration stage aimed to remove ethically sensitive images, including those featuring individuals under 18.

We Have Decided To Replace Hagrid’S

manual box annotation with an automated one due to its time- and labor-intensive nature. A substantial size of Ha- GRID is enough to train a robust hand detector for auto- mated annotations. We have also implemented crowd mod- eration for quality assurance.

The annotation process was divided into hand detection and gesture labeling. We trained the YOLOv10x detec- tor on the HaGRID, previously reducing all gesture classes to one class “hand”. For images featuring one-handed ges- tures, the hanging hand is identified as the one that is always below the hand with the gesture (see Fig. 9a in the suppl.

materials). For two-handed gestures, we obtain joint boxes by combining two hand boxes, making a box from the upper left edge of the left hand to the lower edge of the right hand (see Fig. 9b in the suppl. Fig. 9c in the suppl.

materials shows the exception of the “xsign” gesture, where two boxes are merged to a square, the side of which is equal to the distance between the extreme points of the boxes.

Ataset Characteristics

The HaGRIDv2 dataset is an extension of the widely used HGR dataset HaGRID. Adding 531,358 samples di-

Https://Elementary.Activebc.Ru/

vided into 15 new gesture classes and the extra “no ges- ture” class to the original HaGRID, the received combina- tion contains over a million primarily Full HD RGB images (see Fig. 1 and Fig. 2d). Since the presented research aims to build a system for home automation devices and confer- ence control, added classes are related to these domains.

Thus, each gesture is intended to perform a specific asso- ciative function, as shown in Tab. 5 in the suppl. materials. A special “no gesture” class encompasses domain-specific hand postures common for mentioned applications, such as hands near the face, relaxed, or holding objects. Added static gestures contributed to an extension of dynamic ges- ture recognition algorithm, proposed in . There is the opportunity to recognize such dynamic gestures as “zoom”, “click”, and others.

Content. New samples were recorded by 28,394 unique crowdworkers, each located in their own scene. Fig. 2e-g shows the subjects’ age, gender, and race distributions, cal- culated for all HaGRIDv2’s 65,977 subjects. We preserve the HaGRID distribution and ensure domain-specific rele- vance by collecting samples in realistic indoor conditions with varying lighting conditions and subject-to-camera dis- tances (see Fig. 10 in suppl. materials). The mean and standard deviation of HaGRIDv2 images’ pixel values for the RGB channels are equal [0.54, 0.5, 0.474] and [0.234, 0.235, 0.231], respectively.

Annotations. Hagridv2 Includes Bounding Box Anno-

tations for all hands. Each image has one or two boxes for one-handed gestures: one for the gesturing hand and one for the non-gesturing hand if it is in the frame (see Fig. 9a in the suppl. materials). Images with two-handed gestures strictly correspond to two boxes for each hand. Moving from the hand detection task to gesture detection, an additional box encompassing both hands is added to two-handed gesture images (see Fig. 9b-c in the suppl. Bounding

Box Annotations Are Proposed In Coco Format With

normalized relative coordinates. Splitting. The dataset was divided into training (76%), validation (9%), and testing (15%) sets, recorded by 57,656, 4,209, and 4,114 subjects, respectively. Note that training, validation, and test sets of original HaGRID are the subsets of corresponding HaGRIDv2 sets. Fig. 2a-c illustrate the subjects’ distribution across three sets with improved het- erogeneity in the test and validation sets compared to the training one. Sets are balanced in age, gender, brightness, and race due to randomness.

In addition, we provide hashed user IDs for researchers to split the dataset on their own. Also, such automatically received meta information as age, gender, and race for each subject, and keypoints for each hand on the images also supplied. Since the dataset is large, we also provided the lightweight version, with all images resized to 512 pixels on the shortest side.

The Original Hagrid Authors Proposed A Dynamic

gesture recognition algorithm that allows recognition based solely on static gestures, eliminating the need for video- based training. This paper presents a novel approach based on such a logic with extended functionality.

Algorithm. Fig. 11 in suppl.

Materials Shows The

pipeline of dynamic gesture recognition. Note that the al- gorithm processes each frame independently during infer- ence without analyzing the entire sequence from start to end.

We Detect Hands In Each Frame With A Lightweight

RFB model, ensuring faster inference. Further, the re- ceived crops are classified into gestures utilizing a single Residual Block . To identify the specific gesture se- quence boundaries accurately, we create a queue of the rec- ognized crops over the last n frames, where n is experimen- tally set to 30. Besides, we implement checks on the dura- tion of each gesture and track the location of its initiation and completion to ensure accurate classification.

New Dynamic Gestures. The Algorithm Supports Four

categories of gestures: swipes, zooms, clicks, and drag-and- drops. Each category includes multiple gesture variations, as illustrated Fig. 5 in the suppl. materials. These variations allow for associating different functionalities with each ges- ture, enhancing the system’s versatility.

Based on such an approach, the system is predictable and lightweight, with 276,292 parameters and 106.14 MFLOPs for the detector and 102,605 parameters and 6.9 MFLOPs for the classifier, and runs efficiently on standard CPU hard- ware. Additionally, it is easily expandable with new custom gestures, requiring only static images for training.

Experimental Setup

The base experiments are divided into three groups: ges- ture detection, gesture classification, and hand detection. We resized the images’ maximum side to 224 and padded the result to a square. The full-frame gesture classification is based on 33 main classes without the ”no gesture” class, as each image contains one of the target gestures. For per- formance evaluation, we used the F1-score and Mean Av- erage Precision (mAP) metrics for classification and detec- tion tasks, respectively. All models were trained on a single Tesla H100 with 80GB for 100 epochs with a batch size of 128. Early stopping was triggered after 10 epochs without the metric increasing by at least 0.01. Tab. 6 in the suppl.

materials describes the detailed training hyperparameters. Hand and Gesture Detection. Three detection architec-

Yolov10X , Were Employed To Ensure That Hagridv2

can train a robust gesture detector. We use the SGD opti- mizer with an initial learning rate of 0.01 for YOLOs and

-

Table 2. Models training results on the HaGRIDv2. F1-score and Mean Average Precision (mAP) were chosen as classification and 0.0001 for SSDLite. The YOLO models employed default augmentations such as mosaic, hsv, and horizontal flips, while SSD trained without any modifications. Using the same setup, the YOLOv10x model was used to train for the hand detection task.

Resnet-152, Mobilenetv3 Small, Mobilenetv3 Large,

and pre-trained on ImageNet ViTB16 utilized in , we also employed ConvNext as a full-frame gesture classi- fier. The AdamW optimizer with a specified initial learning rate, weight decay, and scheduler for each architecture (see Tab. 6 in the suppl. materials) was used.

Results

Tab. 2 presents the evaluation metrics on the HaGRIDv2 test subset. The metrics are remarkably high, demonstrat- ing the dataset’s effectiveness in training robust models. To ensure the model’s ability to work in real-life conditions, we provide a demo of gesture classification and detection models in our repository.

Ross-Dataset Evaluation

Experimental Setup. This section covers cross-dataset evaluation for hand and gesture detection5. We utilized the same setup across all experiments, employing YOLOv10n as a detector, the hyperparameters and augmentations de- scribed in Sec. 5, and mAP as a detection metric.

We Were Limited To The Hands And

OUHANDS datasets in the gesture detection task since we could not access the LaRED and SHAPE datasets. These datasets intersect only in 5 classes — namely “fist,” “one,” 5Cross-dataset evaluation for classification was excluded due to the lack of accessible datasets with sufficient overlapping gestures, making the comparison non-informative.

Test Avg. Map (↓)

Table 3. Cross-dataset evaluation in the gesture detection task. mAP was computed for each pair of datasets and averaged sepa- rately for training and testing. Higher average mAP during train- ing indicates greater model robustness, while lower mAP during testing reflects higher dataset complexity. Diagonal values were excluded from the averages to ensure unbiased comparison and assess generalization.

“palm,” “peace,” and “three” — with HaGRIDv2. Thus, we left only samples with these overlapping gestures, reducing three training, validation, and testing sets. The original Ha- GRID dataset was excluded from these experiments, as its content is entirely subsumed within HaGRIDv2.

Tab. 3 Indicates The Hagridv2’S Complex-

ity, as evidenced by the lowest test average mAP. Further- more, HaGRIDv2 exhibits superior domain generalization, supported by the highest train average mAP. Notably, Ha- GRIDv2 is the only dataset that consistently achieves valu- able metrics across tests on other datasets, underscoring its value in training robust models.

We Compare Hagridv2 With Datasets De-

signed explicitly for hand detection to assess its ability to solve this task. Since HaGRIDv2 was annotated with verti- cal bounding boxes, we only compared it with similarly an- notated BodyHands, Human-Parts, and EgoHands datasets.

We also included the original HaGRID dataset to test the impact of more quantity and variety in gestures in the Ha- GRIDv2. The training, validation, and test sets of HaGRID and HaGRIDv2 were carefully curated to prevent overlap, eliminating the risk of data leakage and ensuring the in-

Test Avg. Map (↓)

Table 4. Similar cross-dataset evaluation as in Tab. 3 for hand detection task. tegrity of the comparison. Results. Although neither version of HaGRID was ini- tially designed for hand detection, and their samples ap- pear simpler than other datasets (as seen in the Fig. 12 in the suppl. materials), Tab. 4 demonstrates that HaGRIDv2 improves the generalization ability on hand detection task.

Additionally, HaGRIDv2 achieves a higher mAP on Ha- GRIDv2 than HaGRID itself, suggesting that increasing the diversity of gesture classes improves the model’s ability to generalize older gestures.

Experimental Setup. We Compared Hagrid And Ha-

GRIDv2 as pre-training datasets to demonstrate that a 2× increase in samples and class diversity in HaGRIDv2 con- sistently yields more reasonable results. ResNet18 for ges- ture classification and YOLOv10 for gesture detection were employed, applying the same hyperparameters and metrics from base experiments in Sec. 5.

Hands And Ouhands Were Utilized To

fine-tune detectors on their training sets with further as- sessment on their test sets. The pre-trained classifiers were fine-tuned on the Kinect Leap, Senz3D, OUHANDS, and

Hands Training Sets, While Htu Hgr And Sit-Hands

were not available due to the invalid link. Results. Fig. 3 shows that models pre-trained on Ha- GRIDv2 consistently outperformed those pre-trained on HaGRID, improving model generalization and proving its indispensability for pre-training.

False Positive Triggering

Experimental Setup. We conducted a check for false positive triggering to assess the impact of diversifying the “no gesture” class (see Fig. 8 in suppl.

We

trained two YOLOv10n detectors: one on the original Ha- GRID dataset and the other on HaGRIDv2. Each trained model was assessed on the “no gesture” samples from the HaGRIDv2 test set to evaluate the amount of false positives.

The Hagridv2-Trained Model Produces 6

times fewer false positive errors than the HaGRID-trained model, which is especially important for production-level HGR systems (see Fig. 4). The mAP metrics were 57 for Figure 3. Impact of pre-training on gesture classification and de- tection across HaGRID and HaGRIDv2. “Original” metrics are sourced from the respective dataset papers; missing values indi- cate metrics not reported by the authors.

Figure 4. Comparing the false positives on the “no gesture” class for HaGRID and HaGRIDv2 datasets. HaGRID and 72.9 for HaGRIDv2, highlighting the superi- ority of the new “no gesture” configuration.

Gesture Generation

Experimental Setup. This section aims to demonstrate that adding new gestures improves the quality of generat- ing images of people showing gestures. We fine-tuned two

Stable Diffusion 2.1 Models Using The Lora (Low-

Rank Adaptation) from the diffusers library on the Ha- GRID and HaGRIDv2 datasets. Each image in the datasets was annotated with an automatically generated description using BLIP-2 , following the format: “{blip caption} showing {gesture name} gesture”. During sample gener- ation, we used prompts like “there is a person showing {gesture name} gesture”. To conduct a fair comparison, we also compared the HaGRIDv2 fine-tuned model with the original Stable DIffusion 2.1 without fine-tuning.

Evaluation. To Compare Hagrid And Hagridv2, We

evaluated the generation quality of 18 gestures from the original HaGRID. For the original diffusion model, we tested 26 gestures from HaGRIDv2, excluding “inverted” and “thumb index” gestures due to their specificity, and kept only one version of each gesture (e.g., keeping “heart” and removing “heart2”). Each model generated three im- ages per gesture, totaling 6 images per gesture. We con- ducted a Subjective Benchmark Scoring (SBS), as shown in Fig. 13 in the suppl. materials. Three crowdworkers com- pared each pair of images, resulting in 162 comparisons for the fine-tuned models and 234 for the original Stable Diffu- sion and HaGRIDv2-tuned models. Workers voted on two criteria: (1) which image had more anatomically accurate hands and (2) which better resembled the reference gesture.

Results. Fig. 7 in the suppl. materials shows that the model fine-tuned on HaGRIDv2 generates more accurate gestures due to the wider variety of classes it was trained on, allowing for easier replication of specific patterns. How- ever, the original HaGRID achieved slightly better anatom- ical accuracy due to its simpler distribution and lack of complex hand postures. Additionally, the comparison with the original Stable Diffusion 2.1 model indicated its limited ability to generate recognizable gestures, with the anatomi- cal accuracy of the hands also being inferior (see Fig. 14 in the suppl. materials).

Ataset Creation. Hagridv2’S Samples Contain Per-

sonal information, so crowdworkers must consent to collect, process, and publish their photos. We comply with Russia’s Federal Law ”On Personal Data” (27.07.2006 N152), en- suring legal data handling. For ethical reasons, images of children were excluded from the dataset. Crowdworkers in- volved in all stages of data collection were compensated at least the minimum wage in Russia. After validation, we justified each rejected photo and allowed crowdworkers to challenge rejections. As the HaGRID dataset is part of Ha- GRIDv2, we ensured that HaGRID adheres to the described ethical requirements by contacting its authors.

Biases. Utilizing only Russian crowdsourcing platforms can lead to an imbalance in the racial diversity of work- ers. We tried to minimize this gap and covered the most frequently identified races – Caucasian, Negroid, and Mon- goloid. Even though there is an imbalance, trained on the HaGRIDv2 neural network can accurately recognize ges- tures from underrepresented racial groups.

There Is The Risk Of Misusing The

datasets with faces to improve surveillance systems, pro- file individuals based on race, create deepfakes, and con- tribute to identity theft. We release the dataset under a pub- lic license for non-commercial use in research purposes, ac- knowledging the potential risk of its misuse for unlawful activities. We used anonymized user hash IDs in the dataset annotation to preserve crowdworkers privacy and enable the ability to split HaGRIDv2 by its users.

Imitations

Dataset. As noted in Sec. 8, the dataset is biased towards the white race, potentially compromising the algorithm’s ro- bustness and performance in diverse scenarios. Addition- ally, the Gaussian age distribution, limited to individuals 18 years and older, may lead to inaccurate predictions for children and seniors. The “no gesture” class expansion is specified to poses typical for device interactions, potentially leading to false positives in other scenarios. Furthermore, predominantly home-based scenes may lead to incorrect op- eration when people wear outdoor clothing and gloves in different weather conditions.

Ynamic Gesture Recognition Algorithm. The Deter-

ministic nature of the algorithm reduces its robustness, as it demands users to perform gestures with exact precision. This inflexibility can result in recognition inconsistencies, especially in real-world scenarios where slight variations in gesture execution are common, ultimately limiting the algo- rithm’s reliability and user experience.

Generative Models. While the HaGRIDv2 dataset aids in training models to generate people displaying gestures (see Sec. 7.3), it has limitations. The dataset’s focus on spe- cific gestures and the limited variety of hands in natural po- sitions restrict the models’ ability to generate anatomically accurate hands in free poses, which may limit its broader applicability. Additionally, since the model was fine-tuned solely on samples from HaGRIDv2’s limited distribution, it tends to generate images that look quite similar, reducing the overall variety.

Onclusion

This paper introduces HaGRIDv2, an enhanced version of HaGRID, which become the largest and most diverse

Ncluding A New “No

gesture” class significantly strengthens the system’s robust- ness and adaptability to real-world conditions, paving the way for more reliable gesture-based interaction technolo- gies. The HaGRIDv2 can also be used for robust hand gen- eration by diffusion models. We also present a dynamic gesture recognition algorithm that identifies various manip- ulative gestures for device control. The dataset, pre-trained models, and the dynamic gesture recognition algorithm will be published in our repository.

Authors:

Peder EZ Larson 1, 2,* , Jenna ML Bernard1, James A Bankson 3, Nikolaj Bøgh 4, Robert A Bok1, Albert P. Chen 5, Charles H Cunningham 6,7, Jeremy Gordon1, Jan-Bernd Hövener 8, Christoffer Laustsen 4, Dirk Mayer 9,10, Mary A McLean11 12, Franz Schilling13, James Slater1, Jean-Luc Vanderheyden5, 14, Cornelius von Morze 15, Daniel B Vigneron1, 2, Duan Xu1, 2, and the HP 13C

94143, Usa.

Denmark. 5 GE Healthcare, Menlo Park, California, USA. 6 Physical Sciences, Sunnybrook Research Institute, Toronto, Ontario, Canada.

8 Section Biomedical Imaging, Molecular Imaging North Competence Center (MOIN CC), Medicine, Baltimore, MD, USA. Cambridge, United Kingdom.

14Jlvmi Consulting Llc, Dousman, Wi, Usa

#See Acknowledgements for a list of all HP 13C MRI Consensus Group Members This work was supported by the ISMRM Hyperpolarized Media MR Study Group, the ISMRM Hyperpolarization Methods & Equipment Study Group, and the Hyperpolarized MRI Technology Resource Center (NIH/NIBIB grant P41EB013598).

Abstract

MRI with hyperpolarized (HP) 13C agents, also known as HP 13C MRI, can measure processes such as localized metabolism that is altered in numerous cancers, liver, heart, kidney diseases, and more. It has been translated into human studies during the past 10 years, with recent rapid growth in studies largely based on increasing availability of hyperpolarized agent preparation methods suitable for use in humans. This paper aims to capture the current successful practices for HP MRI human studies with [1-13C]pyruvate - by far the most commonly used agent, which sits at a key metabolic junction in glycolysis. The paper is divided into four major topic areas: (1) HP 13C-pyruvate preparation, (2) MRI system setup and calibrations, (3) data acquisition and image reconstruction, and (4) data analysis and quantification. In each area, we identified the key components for a successful study, summarized both published studies and current practices, and discuss evidence gaps, strengths, and limitations. This paper is the output of the “HP 13C MRI Consensus Group” as well as the ISMRM Hyperpolarized Media MR and Hyperpolarized Methods & Equipment study groups. It further aims to provide a comprehensive reference for future consensus building as the field continues to advance human studies with this metabolic imaging modality.

Keywords: Hyperpolarized MRI, metabolic imaging, carbon-13, pyruvate, dissolution dynamic

Introduction

MRI with hyperpolarized 13C agents, also known as hyperpolarized (HP) 13C MRI, has shown great potential as a novel imaging modality, particularly for its ability to probe metabolic processes in real time. The first human studies with HP [1-13C]pyruvate were performed in 2011 in prostate cancer patients (1).

Since then, there have been over 60 papers published with imaging results of human subjects from 13 different sites, with applications including prostate cancer, brain tumors, breast cancer, kidney cancer, pancreatic cancer, metastatic disease, liver disease, ischemic heart disease, diabetes and cardiomyopathies. The vast majority of these studies used [1-13C]pyruvate (1–63), where [2-13C]pyruvate (64) and 13C-urea (56) have been demonstrated too.

As clinical HP 13C MRI advances, there is a growing need to build consensus for best practices, which are critical for comparing data across sites, performing multi-site trials,deploying methods to new sites, partnering with vendors, and potentially for obtaining broader regulatory approvals.

In March 2022, we initiated an effort to build consensus within the HP 13C MRI community with this opportunity in mind, and it was greeted with strong enthusiasm. The “HP 13C MRI Consensus Group”, containing over 55 members from 27 sites, identified the area of greatest need and opportunity for consensus building to be HP [1-13C]pyruvate human

●

Pyruvate is the most mature and widely used HP agent and has the most significant translational evidence emphasizing the potential clinical impact.

●

Clinical trials, particularly multi-site trials, have the strongest need for consensus methods to ensure that data can be combined across sites. This work is a Position Paper for which the goal is to describe current successful practices and study methods for HP [1-13C]pyruvate human studies along with justification to support those practices. This is divided into four major topic areas: (1) HP 13C-pyruvate preparation, (2) MRI system setup and calibrations, (3) data acquisition and image reconstruction, and (4) data analysis and quantification (Fig. 1). The current successful practices and study methods include a literature review of published peer-reviewed journal papers showing human HP [1-13C]pyruvate study data, up to September 2022 (1–63), as well as new unpublished information from surveys of HP 13C study sites. Based on this information, we also highlight the evidence gaps, strengths, and limitations of current practices which are summarized at the end of each section.

Figure 1: Illustration of the HP 13C MRI human study process, including the 4 major areas covered in this paper: Hyperpolarized 13C-pyruvate preparation, MRI system setup and calibration, Acquisition and Reconstruction, and Data Analysis and Quantification.

Figure 2: Anatomical targets of HP [1-13C]pyruvate MRI human studies published up to September 2022.

Hyperpolarized 13C-Pyruvate Preparation

This section covers the processes for creating the HP agent, 13C pyruvate, and will include many aspects and considerations that are needed to safely and effectively prepare doses for metabolic imaging studies in human subjects. These include material, personnel, equipment and facility, fluid path preparation, quality control, and release.

It is helpful to understand that the specifications of a dose of 13C pyruvate suitable for in vivo MR HP metabolic imaging were shaped in part by early preclinical studies performed by GE HealthCare summarized in Ref. (65). In short, the safety of the two novel drug components, 13C pyruvate and the electron paramagnetic agent (EPA) AH111501, were demonstrated in those studies. The more precise formulation of the dose suitable for human use was then determined from clinical studies (66) that included two Phase 1 clinical trials in young and elderly healthy volunteers without hyperpolarization of the 13C nuclei and another Phase 1/2a dose escalation and imaging feasibility study with HP 13C pyruvate in 31 prostate cancer patients at the With the exception of the first HP 13C imaging clinical trial, which utilized a prototype device in a cleanroom (1), all HP 13C studies performed in humans to date have utilized the SPINlab polarizer (manufactured by GE HealthCare). Consequently all doses of the HP 13C pyruvate delivered by SPINlab have been produced using the “SPINlab Pharmacy Kit” that serves as the container-closure system for the various drug components (13C pyruvic acid and EPA mixture, dissolution medium, and neutralization and dilution medium) during sample polarization, dissolution and quality control (QC) processes. Thus many aspects of the HP sample preparation considerations discussed below are related to the SPINlab instrument and the consumables designed to be used with it (67).

General Considerations

While more than 860 patients or healthy subjects having been injected with HP 13C pyruvate as of January 2022 without reports of any serious adverse events (68), HP 13C pyruvate injection remains an investigational MR contrast agent and can only be administered by those with Investigational New Drug (IND) exemption from the Food and Drug Administration (FDA) in the USA, a Clinical Trial Application (CTA) in Canada, approval from National Research Ethics Committee Services in the UK, or approval from the relevant local regulatory body. Thus, methods and processes involved to produce a dose should have patient safety as the first priority. Since utilizing dissolution dynamic nuclear polarization (dissolution-DNP) for human use is still a relatively new development, there are no existing published regulatory guidelines specifically for this method.

There are two major production styles that determine how various sites approach the agent preparation. In the US, the most common approach is to rely on a sterilizing filter (“Terminal Sterilization”) to ensure sterility of the final product, akin to PET tracer production, where a starting molecule with a radioisotope is processed using various other ingredients to make the final, desired and injectable contrast agent within a necessarily short amount of time (69). For these sites, sterilization of the components and accessories upstream of this filter are not required, although many of them were manufactured and tested following Good Manufacturing Practice (GMP) or Good Laboratory Practice (GLP) requirements. The filling process is usually performed under an ISO 5 laminar flow hood, but a clean room or an isolator is not required.

This approach is typically accompanied by testing the integrity of the sterilizing filter prior to release of the dose for injection. Typically, post release endotoxin and sterility tests are performed using an aliquot reserved from each released dose.

In the UK and EU, the most common approach is to more-closely follow sterile pharmaceutical compounding guidelines (70), where all components and ingredients are required to be sterile or manufactured under GMP guidelines and are assembled and filled within a clean room environment or an isolator system (“Sterile Preparation”). Typically a batch of Pharmacy Kits for HP 13C pyruvate injection are prepared together. The sterility of the final dose is also ensured by batch validation testing, in addition to the sterility of the ingredients and the sterile compounding process. The endotoxin and sterility testing are performed for the process validation but are not performed for each injected dose.

Some institutions fill and assemble the Pharmacy Kit required for a specific study on the same day or the day prior to polarization, dissolution, and patient administration, but others have also demonstrated the feasibility of preparing a batch of kits, keeping them in a -20ºC freezer and using them over a period of a few months.

Beyond the obvious requirements that the process and the facility has to ultimately produce a dose that is safe to inject into a human, regulatory authorities will also focus on the question “Are you in control of your processes?”. To be in control of your process requires an in-depth and broad understanding of all processes involved in pre, post, and during the production process.

Personnel

It is typical and may be required to have licensed personnel involved in the production process depending on local regulations.Typically a pharmacist, radiopharmacist or other similarly qualified person (QP), in charge of the facility where the Pharmacy Kit filling and preparation is taking place, is responsible for the overall process and the release of the injectable dose.

Qualified cleanroom technicians are often involved in the Pharmacy Kit filling under the supervision of the pharmacist or QP. As is required for pharmaceutical compounding or PET tracer production, training requirements and training records for all personnel need to be maintained and available for audit by the FDA or equivalent.

Equipment And Facility

The facility and all equipment need to have standard operating procedures (SOPs) that describe how equipment is used, maintained, and calibrated to comply with relevant legislation. Currently, almost all the filling of the Pharmacy Kit takes place within a compounding laminar flow hood or isolator (typically ISO 5). At some sites, the filling is conducted within a cleanroom, while at others, it is conducted in a dedicated non-cleanroom space, reflecting differences in cleanroom approach and specifications between regulators worldwide (71). Some equipment or facilities, such as the compounding hood or cleanroom, may require external certified laboratories for testing.

Material Handling

Material handling guidelines (69,70) require SOPs detailing a system to track all of the materials involved in the HP production process for a particular patient dose, similar to current good manufacturing practice (cGMP) requirements for material handling for drug compounding. This includes acceptance standards, storage conditions, amount used in the patient dose for each ingredient and materials used in the assembly of the fluid path and Pharmacy Kit. Currently some users choose to open and inspect and sometimes modify the Pharmacy Kits upon arrival, but some users keep them in the sealed packaging until they are required for dose preparation.

Pharmacy Kit Filling And Assembling

As required by an IND or its equivalent, the preparation of the doses of HP 13C agent are detailed in the Chemistry, Manufacturing, and Control (CMC) section of an applicable regulatory submission; an example of this has been made available (72). It describes the processes of filling the Pharmacy Kit with the different components that make up the final drug product, and of assembling the final kit for either storage or immediate use in the polarizer. Special attention should be given to the laser welding process in order to satisfy installation qualification (IQ) and operational qualification (OQ). Typically, the final developed process is validated by process qualification (PQ) runs, during which 3 or more Pharmacy Kits are filled and used and the final HP 13C products are tested for endotoxin and sterility and to confirm that they meet the dose specifications for injections (usually including pyruvate concentration, residual EPA concentration, pH, liquid state polarization level and dose temperature). The data from 3 consecutive PQ runs are submitted as part of the IND submission (or its equivalent), and are often also reviewed by the Institutional Review Board (IRB) where the studies are conducted.

Quality Control And Dose Release

The quality control (QC) and dose release can be separated into two aspects: one is the QC and release of the filled Pharmacy Kit, and second is the QC and release of the HP 13C agent for injection, after polarization and dissolution. For institutions filling a batch of kits and storing them to use over a period of time, typically the batch can be released based on initial validation, environmental monitoring data from the day of kit production, and if filters are used during preparation of any of the components, filter integrity testing. But in some cases one or more kits are used for validation before the batch of kits are released for future use. For institutions that fill only the kits required for specific studies shortly before the experiment, the filled kits often do not go through separate release tests before they are used.

The quality control of the HP 13C pyruvate solution post dissolution is primarily performed to ensure that the agent meets the dose specifications (Table 1) before it is administered to the subject. These specifications target both safety (pH, residual EPA, temperature) and efficacy (pyruvate concentration, polarization, volume). Typically, the pyruvate concentration, residual EPA concentration, pH, dose temperature, dose volume, and liquid state polarization are measured by the QC accessory associated with the SPINlab polarizer. Some users perform a secondary measurement for one of the parameters, such as pH, using a different instrument or pH paper. For sites that do not go through a separate release testing process for batch filled kits, the integrity of the sterilization assurance filter, a part of the Pharmacy Kit, is typically tested as a part of the dose release. It is also common for these users to preserve an aliquot of the final HP 13C pyruvate solution for post-release endotoxin and sterility testing. This testing cannot be completed fast enough to test an individual dose prior to injection, but this is why other processes such as PQ runs and validation testing are done to minimize the chance a subject could be injected with a contaminated dose.

The Final Dose Release And Injection

should be done under the supervision of a licensed professional, based on local regulations.

Some Key Challenges

Many of the challenges associated with HP 13C pyruvate preparation can be attributed to the conditions required for the dissolution-DNP method of high magnetic field (~3-7 T) and very low temperature (~1 K) during polarization, with pressurized and superheated water necessary for the rapid dissolution event. These extreme conditions are quite challenging for the design of the container-closure and fluid path system. In particular, the cryogenic temperature in the polarizer requires special attention to any moisture or ambient (moist) air introduced into that portion of the fluid path, which can form an ice block at ~1 K. This ice can lead to flow restriction during the dissolution event and reduce the strength of the laser welded bond between the cryovial and its cap. This can ultimately produce failures in the dissolution step, including variations in final pyruvate concentration and pH that may fail to meet QC release criteria as well as fluid path ruptures that provide no available dose and result in polarizer down-time.

The polarization of the HP 13C pyruvate sample decays quickly over the span of a few minutes after dissolution, and thus the process of dissolution, QC for release, and injection should be completed as fast as possible to preserve the high polarization level achieved. Any delays in the preparation process, such as transportation time or equipment malfunction, can significantly reduce the final polarization and result in lower quality imaging data.

Current Practices

A summary of data collected from all sites performing clinical trials with HP 13C-pyruvate is shown in Fig. 3 and Table 1, including the specification of the final dose and how the quality control and release of the final dose are performed. There is a split in the Production Style, described in the General Considerations section above, with 8/13 sites using Sterile Preparation versus 5/13 using Terminal Sterilization. While many of the dose specifications show notable differences in acceptable ranges, all of these variations listed in tables have been successfully and safely been used to perform HP 13C pyruvate studies in humans. Their differences depend on the institutions’ preferences, resources and their particular regulatory situation. There is high similarity in pyruvate ranges, temperature ranges, EPA limits, and volume limits. There is modest variability in pH ranges and large variability in the endotoxin test limit. There is a 3-fold difference in acceptable polarization levels, which are measured to ensure a futile dose is not injected since the polarization is directly proportional to SNR. This reflects the decision by several sites to believe that useful data can be still be obtained with suboptimal polarizations.

Figure 3: Hyperpolarized agent preparation methods reported by sites currently performing HP

In House

Table 1: HP 13C-pyruvate preparation parameters, methods, and dose specifications used for quality control testing and release as well as validation. These were obtained from a survey of all sites performing clinical trials with HP [1-13C]pyruvate. The parameters used for product release are noted in bold text, otherwise these parameters are measured for batch validation or other QC measurements. The endotoxin and sterility testing are performed during process validation of the batch and/or post-injection, and largely depends on the agent production approach.

Summary

The overall safety record of HP 13C-pyruvate has been very strong, and the SPINlab hyperpolarizer has proven to provide high polarizations at human sized doses while meeting numerous QC and release criteria. A weakness remains the failure modes of the SPINlab Phamacy Kits (e.g. ice blocks, path ruptures), which are placed under extreme requirements particularly during dissolution. The preparation process still requires a high degree of expertise.

Therefore, there is a significant need to improve the reliability, robustness, and ease of operation for generating HP 13C-pyruvate doses for human studies. Furthermore, there is a divide between manufacturing and sterile compounding style preparation as well as other site-specific practices, resulting in variations in SOPs and justification required to relevant regulatory bodies. There have also been no comparisons between these approaches. It is also unclear what release criteria and QC parameters are truly required to ensure patient safety.

However, all of the reported methods are acceptable and approved by the appropriate regulatory authorities, and have led to the rapid expansion of successful human studies in recent years.

Mri System Setup And Calibrations

This section covers the MRI system setup, including the imaging system, RF coils, phantoms, and prescan calibration methods.

Imaging System

The main prerequisite for a given MRI scanner to be capable of supporting studies with HP 13C is its “broadband” capability to transmit and receive radiofrequency (RF) signal at the frequency of 13C, which is around 4 times lower than 1H. This does not come as a default on clinical MR devices. The transmit power of the broadband amplifier should also be sufficient to support the intended flip angle and RF pulse shape with the employed transmission RF coil(s) for 13C. Most studies to date use relatively low flip angles (< 90 degrees) for HP 13C in order to preserve polarization for time-resolved imaging. The capability to receive 13C signal on multiple channels is also desirable to increase SNR, as discussed further in the “RF coils” section.

The choice of magnetic field strength is primarily dependent on the metabolites’ frequency separation due to chemical shift dispersion and 1H imaging. High field strengths do not enhance hyperpolarized 13C signal as they do for 1H because the signal strength in a HP experiment relies on manipulating the population of quantum energy states outside of the MRI scanner.

However, the injected HP 13C-pyruvate and its metabolic products have greater frequency separation at higher fields, and it may thus be easier to separate and quantify these resonances at higher fields. This comes at the cost of a reduction in the achievable T2* and often reduced T1. As the initial polarization is independent of the imaging field strength it has been proposed that the increased T2* at 1.5T can potentially be exploited to increase SNR by adapting the acquisition bandwidth or reduce off-resonance imaging effects in cases when the decay of the transverse magnetization is dominated by T2* (73). In practice, 3T has been used in all published human 13C-pyruvate studies surveyed (Supporting Table S1), and comprises the majority of scanners currently in use for human studies (Table 3). A field strength of 3T is well-suited for 1H MRI anatomical reference and correlative imaging.

Stronger and more rapidly slewing magnetic field gradients support more rapid spatial encoding, particularly for metabolite-specific single-shot imaging using echo-planar imaging (EPI) or spiral imaging (See “Acquisition and Reconstruction”). Although the spatial resolution acquired for HP 13C imaging is typically much coarser than for 1H MRI, the factor of ~4 in gyromagnetic ratio leads to the same reduction factor in performance of the gradient system, so 13C experiments are potentially more limited by gradient hardware performance. To date, all human studies have used the commercially-available integrated gradient systems provided in clinical MRI scanners.

Optimization of scanner design has understandably focused on minimization of artifacts in 1H MRI, where devices such as room lights, the gradient amplifiers, and the motors driving the patient bed are checked to ensure that they do not produce RF interference at the 1H frequency, but artifacts may arise at other frequencies. Eddy current compensation is also not always appropriately adjusted for nuclei at other frequencies (74). In order to optimize for 13C, many sites have performed checks on phantoms for RF interference, gradient artifacts, and eddy currents (74), including the use of post-hoc gradient impulse response function characterisation and correction, and some vendors have fixed these issues as well.

Rf Coils

For HP 13C imaging studies in humans, RF coils for both 1H and 13C nuclei are needed, with 1H MRI providing an anatomical reference for registration and optional additional multiparametric MRI readouts. At the Larmor frequency of 13C nuclei, the relative contributions from coil noise compared to sample noise increase compared to 1H (73,75), although sample noise still is likely the dominant contributor for human-sized coils at 32.1MHz - the resonance frequency of 13C nuclei at 3T.

The key requirement for human 13C-pyruvate RF coils are that the coil geometry and sensitive volume must cover the volume of interest in the subject. Table 2 and Figure 4 shows coil configurations that have been used and optimized for applications in different anatomic regions.

Volume resonators are most commonly used for transmit, as they surround the subject to

Provide B1 Transmit Across The Fov (B1

+). While 1H relies on a large birdcage (“body”) coil built into the scanner, 13C transmit coils must be placed inside the bore. This takes up valuable space within the magnet, and also has led to the use of designs with relatively inhomogeneous

B1

+. Many human studies have used Helmholz pair resonators for transmit, including the “clamshell coil”, which has a notably inhomogeneous B1

+ Profile But Has Been Used Because Of

relatively easy integration into the scanner bore. B1

+ Variation Results In Variations In The Flip

angles that control the use of the hyperpolarized magnetization and creates errors in common HP metrics (9,76). The exception are head coils, where birdcage designs with highly

Homogeneous B1

+ can be placed around the head while easily fitting inside the bore. As with 1H MRI, higher SNR can typically be achieved by smaller receive coil elements, such as surface coils or phased arrays, and the majority of 13C receive coils used have layouts similar to 1H phased arrays.

RF coil quality control is important to ensure proper functioning of the coils to provide consistent imaging quality, especially with limited natural abundance 13C signal in vivo. It typically involves 1) a physical integrity check of the coil cables and connectors and 2) phantom SNR tests to check the coil’s performance and to monitor it over time (see Phantoms below). An useful reference for RF coil quality control is outlined in the MRI accreditation program of the American College of Radiology (77) and can be adapted for 13C coils.

Notably, configurations for brain and prostate studies used dual-tuned 1H/13C coil designs, which greatly simplify workflow and registration of 1H and 13C images, as no switching of coils is needed.

(1)

Table 2: RF coil configurations reported for human HP [1-13C]pyruvate studies.

Tx = Transmit

coil, RX = receive coil. The commonly used “clamshell” TX coil is a Helmholz pair design. For 1H RF configurations, all used the Body coil for TX unless otherwise noted, and “repositioned” indicates the 13C coil was removed for 1H imaging. One representative reference is listed for each configuration. The RF coil configurations reported in the reviewed papers are shown in Supporting Table S1.

Figure 4: Examples of RF coil configurations used for human HP [1-13C]pyruvate brain studies. (A,B) 13C Clamshell TX (Helmholz pair) and 2× 4-channel paddle RX arrays. (C) 13C Birdcage volume TX and 32-channel RX array (RX array slides into TX coil). (D) 13C Birdcage volume TX and 24-channel RX array, combined with a 1H 8-channel RX array. Image reproduced with permission from Ref (16).

Phantoms

Since hyperpolarized magnetization is non-renewable, phantoms containing 13C nuclei are important to: 1) test the multi-nuclear capabilities of the imaging system, including all parts of the signal excitation and receive chain; 2) perform calibration measurements before a scan with hyperpolarized nuclei; and 3) perform necessary pre-scan adjustments (see “Prescan Calibration” section). The phantoms currently in use are listed in Table 3. Their composition must provide sufficient 13C signal, with additional considerations of conductivity, stability, chemical shift(s) present, potential for dynamic imaging, and cost. The phantom geometries are typically either compact, in order to be used alongside the subject during a HP scan, or large enough to mimic the inner volume of a RF coil for system testing.

One popular compact design contains enriched 13C-urea at high concentration, typically 8 M, which provides a single resonance, placed inside a small container ~1 mL. The most common recipe mixes 13C-urea in a 90% water/10% glycerol solution, with glycerol used to increase the urea solubility and doping with a Gd-based contrast agent to shorten T1 which increases the potential SNR per unit time. For example, when Dotarem is added at a 3:1000 volume ratio the 13C-urea T1 is around 500 ms and T2 is around 100 ms. However, when testing pulse sequences influenced by T1 and T2, doping should be used carefully. This phantom is suitable for frequency calibration, transmit gain calibration, sequence testing, and as a fiducial marker when placed next to a patient. However, enriched 13C-urea has a relatively high cost compared to natural abundance compounds.

For larger volumes (>100 ml), the phantoms most often used contain undiluted ethylene glycol, glycerol, or dimethyl silicone. These compounds have sufficiently high carbon concentrations to provide sufficient 13C signal even with the 1.1% natural abundance of 13C. These larger phantoms matching the inner volume of an RF coil are useful for coil testing, including transmit

+) And Receive (B1

-) coil profile mapping, as well as to mimic acquisitions using in vivo FOV requirements. In this case, size and conductivity should match the expected subject size in order to mimic coil loading and get a realistic estimation of B1+. Large-volume natural abundance urea phantoms have also been used by some sites, but suffer from higher conductivity compared to biological tissues. Typically, it is easier to increase the conductivity and hence coil loading of the non-conductive phantom by adding NaCl to match physiological loading (16,78).

Dynamic phantoms that aim to mimic metabolite kinetics have also been developed (79–81), and have the potential to more closely mimic the HP experiment, but so far these are not widely used.

Prescan Calibration

Prior to performing an MRI acquisition, the so-called prescan procedure is used to set the shim parameters to maximize B0 homogeneity over the field of view (FOV) or a specific region of interest (ROI), the scanner center frequency (CF), the RF transmit gain, and the receiver gain.

While this calibration procedure is usually automated for 1H, the lack of sufficient natural abundance 13C signal prevents use of automated methods. (Although natural abundance 13C lipid signal has been detected, there are so far no reports on using this signal for prescan.) Table 3 shows current practices across sites.

Maximizing B0 homogeneity is independent of the nucleus and is therefore performed prior to 13C imaging using the 1H water signal and existing shimming tools, such as by a standard automated process (“Auto Shimming”) or using high order shimming routines. Similarly, the 13C CF can be calculated from the 1H CF using a predetermined scaling factor that depends on the target chemical shift (82). Another common approach used is to have a small, high-concentration 13C phantom, e.g. 8M 13C-urea, integrated in the RF coil or placed next to the scan subject (1). The reference frequency can also be based on real-time measurements after the HP injection but prior to imaging (83). Both the CF and B0 shimming are critical when using spectrally-selective RF pulses, as inmetabolite-specific imaging methods, where the desired excitation bandwidths are typically very narrow and frequency offsets can lead to a failure mode that is only apparent after injection.

The calibration of the RF transmit power is typically performed on a small, high-concentration 13C phantom placed near the region of interest during the scan or on a large 13C phantom of similar size and coil loading as the subject, prior to the subject scan. Reference power is often done by sweeping the power in a pulse-acquire sequence (53,62), or the Bloch-Siegert method (52,84). When using a small phantom, the location of the phantom, B1

+ Inhomogeneity As Well

as any shielding effects, e.g., when the phantom is integrated into a coil (1), may degrade the accuracy. Other methods include real-time Bloch-Siegert method measurements after the HP injection (83), and using the stronger natural abundance 23Na signal that is close enough to the 13C resonance frequency to be detected by 13C coils (82).

The receiver gain is predetermined, either systematically based on independent phantom measurements and assuming the dose and polarization of the HP compound is known prior to injection, or based on past HP imaging studies.

Power [Kw]

Phantom(s) - during study Phantom(s) - before study 13C Frequency

8

13C-bicarbonate doped with dimethyl silicone, various

Power [Kw]

Phantom(s) - during study Phantom(s) - before study 13C Frequency

Maximum Values

Table 3: Summary of the imaging systems, phantoms, and prescan procedures used at sites currently performing HP 13C-pyruvate human studies. These were obtained from a survey of all sites performing clinical trials with HP [1-13C]pyruvate. *Previously performed studies with a Siemens 3T Tim Trio. The imaging systems, phantoms, and prescan procedures reported in the reviewed papers are shown in Supporting Table S1.

Summary

Commercially available 3T MRI systems are by far the most commonly used for human HP 13C-pyruvate studies, although a systematic investigation of the impact of B0 has only recently been investigated (73). The multi-nuclear RF transmit and receive chain has proven sufficient for current acquisition strategies, although many sites have observed artifacts due to RF interference, gradient interference, and residual eddy currents when operating at the 13C frequency. A variety of 13C RF coils, tailored for numerous anatomical targets, have been successfully demonstrated, with the main limitation that most transmit coils take up a lot of additional space inside the bore and provide relatively inhomogeneous B1

+ Profiles. The

phantoms used have converged into generally 2 categories - small phantoms containing 13C-enriched compounds that can be used during the study and human-sized phantoms containing compounds with high carbon concentrations but without 13C enrichment that are used to test and calibrate the coils. There are no standardized compositions or geometry, and dynamic phantoms that recapitulate in vivo kinetics would be desirable but are still an emerging area. Prescan calibration procedures were not well defined in most publications, so we surveyed individual sites to determine current practices. Calibration procedures for the B0 field (13C CF and shimming) for most sites take advantage of 1H signal and methods, while methods

For Calibration Of B1

+ is more variable across sites, likely a reflection of remaining challenges in how to perform this calibration. Standardization of both phantoms and calibration procedures would synergistically improve the robustness and reproducibility of HP 13C studies.

Acquisition And Reconstruction

Data acquisition strategies in human HP [1-13C]pyruvate MRI studies must account for multiple chemical shifts, efficiently utilize the non-renewable HP magnetization, and acquire data quickly relative to metabolism and relaxation decay processes. These studies require spectral encoding to separate metabolites, necessitating pulse sequences that efficiently encode up to 5D data (3 spatial + 1 spectral + 1 temporal dimension). RF pulses must efficiently sample without immediately saturating the non-renewable HP magnetization, and sequences must acquire data quickly and be robust to both experimental and physiologic variation (e.g. B1

+ Inhomogeneity,

variation in perfusion) to ensure reproducibility and minimize scan-to-scan variability. This section covers current successful practices for data acquisition in human [1-13C]pyruvate studies, and accompanying 1H imaging, from different anatomic regions, including scan parameters and image reconstruction.

Acquisition And Reconstruction Methods

The acquisition methods used in human [1-13C]pyruvate studies can be classified into 3 categories: 1) MR spectroscopy or MR spectroscopic imaging (“MRS/I”), 2) chemical shift encoding methods, and 3) metabolite-specific imaging (Fig. 5).

Mrs/I Methods Specifically

resolve a spectrum that can be analyzed to extract expected as well as unexpected resonances, making this approach very robust. It was used in many initial studies (1).

Chemical Shift

encoding methods, most commonly the Iterative Decomposition of water and fat with Echo Asymmetry and Least-squares estimation (IDEAL) method, use imaging sequences acquired with multiple TEs and rely on a model-based separation of expected chemical shifts (85).

Metabolite-specific imaging methods use specialized RF pulses that are spatially and spectrally selective to excite individual metabolites which are then typically imaged with fast k-space trajectories such as echo planar imaging (EPI) or spirals (86).

Their Application To Different

organ systems is described below. The image reconstruction methods used in human [1-13C]pyruvate studies have typically been conventional methods (e.g. FFT, non-uniform FFT, or equivalent). The incorporation of accelerated imaging and advanced reconstruction methods including parallel imaging (4,57,87) and compressed sensing (7) has also been applied in human studies for improved spatial resolution, temporal resolution and coverage, but have the potential for additional artifacts as well as SNR losses due to ill-conditioning of the reconstruction (e.g. g-factor).

The Majority Of

published studies do not use accelerated imaging indicating the resolution and coverage achievable without acceleration is currently adequate for successful data collection. Performing coil combination, even with fully sampled data has also been shown to have specific challenges for HP human images: using naive sum-of-squares methods suffer from high noise amplification in the relatively low SNR regime of HP [1-13C]pyruvate (compared to 1H), motivating several HP 13C-specific methods that include data-driven coil sensitivity estimation which have shown obvious improvements over sum-of-squares (11).

More recently denoising techniques have been applied as post-processing of human HP data(41,42,44). The techniques applied are based on spatial-temporal singular value decomposition for unsupervised estimation of signal and noise components. They have shown improvements in apparent SNR in the brain and liver, while care must be taken to choose parameters such as the rank threshold to avoid oversmoothing and overfitting to the estimated signal components.

Prostate Studies

Prostate cancer was the first human application of HP [1-13C]pyruvate (1), and data was acquired with MRS/I methods: 1D dynamic MRS, single-slice 2D dynamic echo-planar spectroscopic imaging (EPSI), and single time point 3D EPSI. Advances in imaging strategies led to the development and application of new acquisition schemes, including undersampled 3D EPSI with compressed-sensing (7), model-based chemical shift encoding methods that use a priori information (47,59), and metabolite-specific EPI (10), all of which can provide volumetric whole-organ coverage and dynamic acquisitions.

The pyruvate bolus arrival in the prostate can vary by ± 10 s between patients, necessitating dynamic imaging to reliably and consistently capture the pyruvate bolus (18). For this reason, all currently ongoing studies acquire dynamic data. While MRS/I, chemical shift encoding, and metabolite-specific imaging can all achieve dynamic imaging, chemical shift encoding and metabolite-specific imaging provide greater dynamic and volumetric coverage (85). For scan prescriptions, the FOV is designed to provide full prostate coverage and typically to match the orientation of the anatomic imaging used for registration. Flip angles used in current studies are constant through time, as quantification with a variable-through-time flip scheme is highly sensitive to bolus timing (8) and errors in the RF transmit (B1 +) field (76).

Heart Studies

Data acquisition methods for 13C imaging in the heart must be designed to meet the demands of significant cardiac motion and blood flow. To cope with the periodic cardiac motion, most human heart studies to date used gating to the diastolic window, the longest cardiac cycle interval, which has reduced motion (2,22,28,30,35,36,38,45,52). The duration of the diastolic window limits the available data sampling time, making cardiac acquisitions the most time-constrained of the HP 13C MRI applications. The most common acquisition approach is metabolite-specific imaging with spiral k-space trajectories (2). Their single-shot imaging capability makes these methods particularly robust to motion effects. Furthermore, spiral k-space trajectories provide rapid k-space coverage and relatively benign flow and motion artifacts. The majority of studies have used 2D multi-slice acquisitions, but 3D encoding has also been used successfully (35).

Brain Studies

For HP 13C MRI of the human brain, the majority of studies have also used 2D (slice selective) acquisitions (10–12,14,16,28,33,40,41,44,51,53,60), with a trend toward volumetric coverage using 2D multi-slice metabolite-specific imaging. 3D metabolite-specific imaging of the whole brain, with phase encoding of the slice direction (34,57), has been shown to provide similar SNR efficiency (88) compared with multislice imaging. A number of studies have employed MRS/I (5,6,29,31–33,50,55) resulting in a spectrum from each voxel, which has the advantage of not requiring a priori information about which peaks to encode. This was important in early brain studies when it was not known which peaks would be detectable. Chemical shift encoding, using a set of images with different echo times and an iterative reconstruction of the individual resonances (i.e. the IDEAL approach (85)), has also been used (12,49,54), with the drawback that coverage in the slice direction was limited due to the time required to acquire multiple echo time images.

Abdomen And Breast Studies

The fundamental approaches to data acquisition and reconstruction in the abdomen and breast are largely similar to the aforementioned applications, but demand attention to particular challenges associated with these anatomic regions, especially relating to respiratory motion.

Although it has been shown that a basic 2D MRSI approach based on phase encoding and FID readout can be successfully applied for HP 13C imaging in breast (15) and kidney (13), major advantages in terms of spatiotemporal resolution and coverage have been realized using tailored approaches based on metabolite-specific imaging (43,62) and chemical shift encoding (43), which have facilitated multi-slice or 3D dynamic acquisitions over large FOVs in the abdomen (4,37,46).

The significant respiratory motion encountered in these regions can directly blur 13C images, and has further favored these rapid acquisition strategies. Motion also degrades B0 homogeneity, which can shift frequency-selective excitation profiles and introduce artifacts into rapid imaging readouts. This makes accurate determination of the acquisition center frequency and shimming essential in these regions which often cover large FOVs. (See “Prescan Calibration” section for more information). In some studies, breath-holding was used to minimize motion effects and enforce frame-to-frame data consistency (42). A pragmatic and reasonably effective approach for dealing with respiratory motion during 13C data acquisition is an initial breath-hold (as long as can be tolerated), followed by free-breathing (46,62).

1H Imaging

Collection of 1H imaging data is essential both for prescribing the 13C acquisition and for interpretation of the resulting 13C data. Multi-planar 1H scouts are acquired prior to 13C acquisition to enable graphical prescription of the 13C imaging region. All human HP 13C-pyruvate imaging studies acquire conventional MRI scans (e.g. T1- and T2-weighted volumes) for anatomic reference, aiming to cover at least the full 13C FOV. Acquiring these anatomic scans as close as possible to the time of 13C imaging (immediately before or after) minimizes potential misregistration between the data sets. Depending on the application, other advanced 1H sequences are also acquired (e.g. diffusion-weighted imaging for cancer imaging).

When contrast-enhanced data is acquired, it is done after 13C imaging, as paramagnetic contrast agents will accelerate 13C relaxation.

Reported Study Parameters

Figures 5 and 6, and Supporting Table S2 shows the reported acquisition study parameters for human HP [1-13C]pyruvate studies published as of September 2022. Figure 5 shows a mixture of MRS/I, metabolite-specific imaging, and chemical shift encoding methods have been successfully used, where spectroscopy-based methods have become less prevalent in recent studies. Figure 6 shows the acquisition timing, including the important start time and interval/temporal resolution, is quite variable across studies.

Figure 5: Acquisition methods used in published HP [1-13C]pyruvate human studies published up to September 2022, classified into: MR spectroscopy and spectroscopy imaging (MRS/I); chemical shift encoding methods, such as IDEAL, that use multiple TEs and model-based reconstructions; and metabolite-specific imaging methods that use spectrally-selective excitation to image a single resonance at a time.

Figure 6: Temporal acquisition characteristics reported in HP [1-13C]pyruvate human studies published up to September 2022. (a) Reported referencing of acquisition start times.

(B)

Acquisition start times reported when using dynamic imaging and when timing was reported relative to the end of the injection. (c) Temporal resolutions. “Not Applicable” indicates dynamic imaging was not used.

Summary

Three general categories of acquisition strategies have been used successfully for human HP 13C-pyruvate studies: MRS/I, model-based chemical shift encoding (e.g. IDEAL) methods, and metabolite-specific imaging methods. These have enabled successful studies in the prostate, heart, brain, abdomen, and breast. Recent studies increasingly have used the imaging-based strategies of metabolite-specific imaging and chemical shift encoding which are the fastest methods, although a heads-to–head comparison between techniques has not been performed.

Metabolite-specific imaging is quite popular because of its speed and compatibility with single-shot imaging, but is sensitive to B0 field variations and thus requires careful calibrations. Nearly all studies surveyed acquired data dynamically, allowing measurement of the bolus and metabolite kinetics. The exact timings and associated flip angles vary quite widely across reported studies, with no consensus yet as to how to choose these parameters. Image reconstruction is typically done directly using Fourier Transform methods, and accelerated imaging strategies are uncommon.

Data Analysis And Quantification

This section covers the analysis of data from human HP [1-13C]pyruvate studies, including modeling and metrics, visualization, as well as considerations for how to store data and metadata. Depending on study design, the analysis may need to give quantitative or semi-quantitative output reflecting a biological process or may just reflect a contrast between different regions of interest for quantitative evaluation.

Metrics

Figure 7: HP [1-13C]pyruvate raw data (A) have typically been quantified using four categories of metrics depending on the acquisition. Data acquired as a single time point are often quantified using normalized metabolite images or metabolite ratios (B). Dynamic data can be quantified using normalized metabolite images or metabolite ratios (B), or with metabolite timings such as time-to-peak (TTP) or pharmacokinetic (PK) models (C). The latter two require the data to be time-resolved. [1-13C]alanine and 13C-bicarbonate are analyzed similarly to [1-13C]lactate but omitted here for display.

Metabolite images are commonly used as summary metrics for HP MRI data, often including some form of normalization as well as summed over time as an area under the time curve (AUC) (17). These are analogous to the visual evaluation that is most used for routine clinical work (89,90). In these metabolite images, we expect that the [1-13C]pyruvate AUC signal is predominantly weighted towards perfusion and uptake, while [1-13C]lactate, [1-13C]alanine and 13C-bicarbonate AUCs represent metabolic conversion. The strength of this approach lies in its simplicity and relatively few underlying assumptions. Limitations to the use of single-metabolite images or AUCs include sensitivity to inhomogeneous coil profiles (57,87,91), the acquisition strategy and acquisition parameters, pyruvate polarization and concentration level, and signal relaxation rates (92). Further, the reader must be careful to interpret all the images in conjunction to better understand the underlying biology; for example, increased [1-13C]lactate in the presence of decreased [1-13C]pyruvate delivery can have a very different meaning compared to increased [1-13C]lactate with increased [1-13C]pyruvate delivery.

In an attempt to address variations in coil sensitivity, polarization level, and pyruvate delivery, AUC images are often computed by normalizing to a specified parameter, such as the maximum pyruvate or average lactate signals, or presented as a ratio such as lactate/pyruvate or divided by “total Carbon” - the sum total of HP 13C signal observed across all metabolites. The AUC ratios between metabolites and pyruvate are proportional to the corresponding forward kinetic rates (81,93), but are not directly comparable to rate constants when magnetization loss rates (e.g. relaxation and losses due to signal excitation) differ between studies. Similarly, the ratios between the produced metabolites (e.g. bicarbonate/lactate) can reflect the balance between downstream metabolic pathways (12,55). Care must be taken to consider how AUC images are calculated and normalized before comparing values between studies.

To further quantify the interpretation, pharmacokinetic (PK) modeling approaches were developed to compute the apparent kinetics of pyruvate-to-metabolite exchange (92,94–99). These yield semi-quantitative to quantitative apparent rate constants, given in s-1. Some models require a vascular input function, while others avoid this requirement (95). PK models can explicitly account for acquisition-specific details such as excitation angle and repetition time, and thus may reduce the effects of these details on quantification. An input-less model, provided in the Hyperpolarized-MRI-Toolbox (https://github.com/LarsonLab/hyperpolarized-mri-toolbox) (100) and thus frequently employed for human data, has been shown to fit well and robustly to prostate and brain data (8,20). PK models are quantitative in nature, arguably provide more relevant biological information (8,20), and appear to be reproducible across sites (51). However, rate constants derived from PK models are still apparent rates, and likely do not reflect a single biological characteristic.

Some additional considerations include whether complex or magnitude data is used, as the noise behaviors will impact the analysis differently. Additionally, cut-off thresholds or other criteria may be used to identify and avoid voxels with insufficient SNR before analysis to improve robustness (20,41).

Regardless of the analysis approach, the underlying biology is not always clearly represented by the data; instead, the metrics may be influenced by perfusion, barrier permeability, intercellular shuttles, enzyme activities, co-substrate concentrations, or combinations thereof, depending on the organ and disease of interest (19,43,94,101–103). This may be addressed by incorporating complementary information. As an example, HP 13C pyruvate data is influenced by perfusion, and thus addition of perfusion MRI could be important for interpretation (98,104,105).

All the methods outlined above have been explored in clinical studies, described in Supporting Table 3 and summarized in Figure 8. As of September 2022, approximately 52% of studies involving human subjects report rate constants derived from a PK model with a few different models reported. A nearly equal fraction (51%) of the studies report AUC ratio values.

Approximately 66% of these studies report metabolite-specific images or AUC values. About 40% report SNR values; this metric is particularly frequent in manuscripts that describe technical developments for clinical HP MRI. Approximately 16% of these studies summarize model-free metrics, and 10% report measurements from a single timepoint. Most studies report a combination of quantities.

Figure 8: Reported metrics used for analysis in HP [1-13C]pyruvate human studies published up to September 2022.

Visualization

A wide variety of approaches have been used for visualizing data from human HP 13C-MRI studies. The challenges and practical considerations are: 1) choosing the appropriate metrics to display, 2) how to encode the parameters (e.g. the colormap), and 3) choosing how to provide anatomical context and other multi-parametric data. The choice of visualization also depends on the goal which could be for diagnostic interpretation, but also quality control, reproducibility among readers and publication.

Metrics

The choice of HP 13C metrics is described in detail above. At this stage in HP 13C development where there is no standardized metric, often a combination of metabolite images and ratios or PK model parameters are shown.

Parameter Encoding

The mapping function chosen should provide an adequate, often quantitative, impression of the parameter mapped. There is a consensus in the visualization field that perceptually uniform maps are best suited to visualize continuous parameters, like the greyscale typically used by radiologists as well as other monochrome (black to blue) and color ranges (fire-type, rainbow-type) (106,107). Multi-color heatmaps have been the most frequently employed method for HP 13C data, while greyscale has infrequently been used but it ensures there is no coloring-based bias as well as facilitating later reuse (Fig. 9a). Among the color schemes employed in the clinical HP 13C literature, fire-type scheme seems to be the most common [similar to “Plasma” or “Inferno” in matplotlib.org]. Next most commonly employed is the rainbow-type scheme [similar to “Rainbow” in matplotlib.org].

Anatomical Context

HP MRI faces the challenge that it does not necessarily depict the anatomical features, similar to PET, and thus requires an anatomical reference. Most often, a grayscale anatomical image is overlaid with a HP colormap (Fig. 9c,d). This approach is very intuitive, but can skew perception as the grey-scale anatomical reference may affect the brightness of the HP data (e.g. signal in the skull). This bias does not occur when showing adjacent maps (Fig. 9a, b). Here, anatomical outlines may help to provide reference (Fig. 9b).

Related Journal Articles & DOI Links

Selected peer-reviewed publications relevant to 12 Lead ECG Acquisition. Click the DOI to access the full paper (may require institutional access).

Why Choose Us?

Bangalore guidance for robotics, Spectre and autonomous systems projects.