Share 
Follow 
AboutDownloadsForumsIssuesNews
The page contains links to two data sets — OpenArm 1.0 and 2.0 — each of which contain full 3D data of the human arm under multiple conditions, alongside ground-truth and CNN-generated annotations of select tissue structures for select scans. All code and neural network models used in tissue annotation are also provided. See each package for details.
Image not available
OpenArm 2.0
This package contains the second iteration of the OpenArm data set, a multi-subject set of full volumetric scans of the human arm collected using ultrasound and motion capture. Data are factorial under multiple elbow angles and loading conditions, and thus allow for separable analysis of force- and configuration-associated muscle deformation. Improvements from OpenArm 1.0 include improved data collection procedures allowing for more explicit comparison of force conditions across angles, more subjects, and partial annotations for all subjects enabled by neural-network-based segmentation.
OpenArm 2.0
Apr 24, 2019

  View License

Download Links

Apr 23, 2019
0
Any
Data/images/video
This release contains full volumetric data of the arm for ten subjects under 20 force and elbow angle conditions, as well as 12 scans from one additional subject. Ground-truth segmentation data of the biceps brachii and ventral surface of the humerus are included for one full set of subject scans, as well as several scans of other subjects. Partial neural-network-generated segmentation data are included for all subjects and scans. See readme.txt for further information and the associated publication for a full description of data collection methodology.

PLEASE CITE THESE PAPERS

Yonatan Nozik*, Laura A. Hallock*, Daniel Ho, Sai Mandava, Chris Mitchell, Thomas Hui Li, and Ruzena Bajcsy, "OpenArm 2.0: Automated Segmentation of 3D Tissue Structures for Multi-Subject Study of Muscle Deformation Dynamics," in International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), IEEE, 2019. *Equal contribution. (2019) View


Image not available
OpenArm 1.0
This package contains the first iteration of the OpenArm data set, a set of full volumetric scans of the human arm collected using ultrasound and motion capture. Data are factorial under multiple elbow angles and loading conditions, and thus allow for separable analysis of force- and configuration-associated muscle deformation.
OpenArm 1.0
Nov 28, 2018

  View License

Download Links

Nov 28, 2018
0
Any
Data/images/video
This release contains full volumetric data of the arm for three subjects (Sub1, Sub2, and Sub3) under 20 force and elbow angle conditions. A subset of nine Sub1 scans are annotated with the full visible volumes of the humerus and biceps brachii, as well as sections of the ulna, radius, deltoid, brachialis, and brachioradialis. See readme.txt for further information and the associated publication for a full description of data collection methodology.

PLEASE CITE THESE PAPERS

Laura Hallock, Akira Kato, and Ruzena Bajcsy, "Empirical quantification and modeling of muscle deformation: Toward ultrasound-driven assistive device control," in IEEE International Conference on Robotics and Automation (ICRA), IEEE, 2018. (2018) View


Annotation Source Code
This package contains source code and neural network models for both registration and CNN-based automated annotation of the biceps and humerus, first used to generate the OpenArm 2.0 data set.
Segmentation Code & Network Models
Jul 12, 2019

This release contains all code and neural network models used to generate tissue segmentations for the OpenArm 2.0 data set, using both convolutional neural networks and classical image registration. Code and models used in network training, prediction, analysis, and data augmentation are included.  View License

Download Links

Jul 12, 2019
0
Any
Data/images/video
ZIP archives of all neural network models used in segmenting the OpenArm 2.0 data set, including all models used in the associated publication and the best performing model overall at time of release.

Jul 12, 2019
0
Linux
Source code
Link to GitHub repository of all segmentation code.

Jul 12, 2019
119 KB
Linux
Source code
Archive of all segmentation code in GitHub repository at time of release.

PLEASE CITE THESE PAPERS

Yonatan Nozik*, Laura A. Hallock*, Daniel Ho, Sai Mandava, Chris Mitchell, Thomas Hui Li, and Ruzena Bajcsy, "OpenArm 2.0: Automated Segmentation of 3D Tissue Structures for Multi-Subject Study of Muscle Deformation Dynamics," in International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), IEEE, 2019. *Equal contribution. (2019) View


Feedback