Databases

Largest longitudinal knee imaging study with 26M+ medical images

26,626,000 images from 4,796 participants. 3T bilateral knee MRI, knee/hip/hand/pelvis radiographs, thigh MRI, and full-limb MRI. Longitudinal data 2004-2015+.

1,228 CTs with segmentations of 104 structures including 59 bones

CT scans with ground-truth segmentations covering all vertebrae, ribs, pelvis, femur, tibia, fibula, patella, humerus, scapula, clavicle, and more. Also 616 MR subjects.

40,561 musculoskeletal radiographs from 12,173 patients

Upper extremity X-rays across 7 types: elbow, finger, forearm, hand, humerus, shoulder, wrist. Labeled normal/abnormal by radiologists.

4,083 fracture X-ray images with expert annotations

Hand, shoulder, leg, hip X-rays with 922 fracture instances. COCO, VGG, YOLO, Pascal VOC annotation formats. Expert-annotated by radiologists.

14,236 pediatric hand radiographs with bone age labels

Hand X-rays labeled with skeletal maturity (0-228 months). Standard benchmark for bone age assessment algorithms.

374 CTs with voxel-level vertebral segmentations

Multi-detector CT scans from 355 patients with vertebrae segmentations and labeling, including enumeration abnormalities.

1,005 CT volumes with 11,100+ labeled vertebrae

Spine CT data covering normal, fracture, scoliosis and other conditions. Large-scale vertebral segmentation dataset.

Cancer imaging archive with bone metastasis and osteosarcoma data

Metastatic spine disease CT (55 subjects), osteosarcoma data, plus thousands of whole-body CTs with skeletal structures.

Complete male and female cadaver datasets with CT, MRI, and cryosection

Full-body CT (1mm axial), MRI, and color cryosection images. Derived 3D models include 28 bones, 76 muscles, 16 cartilages per subject.

15,243 high-resolution full-body CT scans of decedents

Bone-optimized and soft-tissue CT reconstructions. 1mm slice thickness, ~12,000 slices per individual. 69 metadata fields.

DXA scans from ~100,000 participants

Whole-body, proximal femur, spine L4-T4, and knee DXA scans using GE-Lunar iDXA. Bone area, mineral content, density, lean/fat mass.

Full-body postmortem CT scans for shape model construction

Anonymized postmortem CTs from forensic institutes. Core dataset for building statistical shape models of any bone.

In vivo joint loading data from instrumented implants (Charité Berlin)

30+ years of measured hip, knee, shoulder, spine forces during daily activities. Includes load-time diagrams, synchronized videos, and numerical force data.

75,732 bilateral walking trials from 2,084 patients

Ground reaction force data from patients with musculoskeletal impairments and 211 healthy controls. Largest gait GRF dataset.

GRF and COP data from 350 healthy individuals

Ages 11-64, up to 8 sessions with 8+ trials each. Raw and processed CSV with MATLAB/Python scripts.

Complete lower extremity musculoskeletal geometry

Segmented CT/MRI with 55 muscle actuators, 12 ligaments, bony landmarks. 12 segments, 11 joints, 21 DOF, 166 muscle-tendon elements per leg.

US population-representative DXA bone density data

Proximal femur and lumbar spine BMD from thousands of participants. Reference data for T-score and Z-score calculations.

Validated musculoskeletal models with bone geometries

Lower extremity, spine, upper extremity, full-body models. Includes bone geometries, joint definitions, muscle paths and parameters.

108,623+ 3D skeletal models from museum collections

CT/MRI volumetric data, 3D surface models, photogrammetry. Primarily biological skeletal material. Hosts oVert project data.

13,000+ CT scans of vertebrate specimens

Covers >50% of genera of amphibians, reptiles, fishes, mammals. Raw CT and 3D mesh models from US natural history museums.

Biomedical 3D models including skeletal anatomy

Skeletal anatomy models segmented from patient CT/MRI. STL format for 3D printing and visualization.

Dictionary-type anatomy database with 3D structure data

Anatomical concepts as segmented regions of a whole-body adult male model. OBJ and STL formats.

100,000+ 3D medical shapes across 23+ datasets

Bones, organs, vessels, surgical instruments. Python API (MedShapeNetCore) for bulk download.

Community-contributed bone STL models from CT scans

Mandible, pelvis, spine, skull, foot models. Includes democratiz3D DICOM-to-STL conversion tool.

Peer-reviewed 3D models of vertebrate specimens

Osteological models from CT data with segmented bones, teeth, and cavities. Citable, ISSN 2274-0422.

Pre-built tibia-fibula statistical shape model (30 specimens)

Cortical and trabecular bone separated. Includes SSM + code + surface meshes. Three example applications included.

Pelvis, femur, tibia SSMs from 333 pediatric CTs

Ages 4-18 years. PCA weights with partial least squares regression. Predicts shape from age, height, weight.

30 subjects, complete lower body bone surfaces

Pelvis, femur, tibia, fibula, patella, foot bones as PLY meshes. MATLAB/Python example code. Derived from VSD cadaver CTs.

PCA-based statistical shape model of the four fingers

10 training specimens, pose-neutralized. Open for community contribution.

9,376 3D femur + cartilage shapes for ML benchmarks

Segmentation masks + 3D surface meshes. Neural shape model benchmarks, osteoarthritis B-Score.

Bone geometries from 35 healthy volunteers (MR scans)

Femur, tibia, fibula meshes (left and right). 22M, 13F. Height 155-193cm, weight 45-108kg.

WHO fracture risk assessment tool for 80+ countries

10-year probability of major osteoporotic and hip fracture from clinical risk factors and optional BMD.

Clinical standards for bone densitometry practice

Official positions on DXA reporting, reference databases, T-score calculation, body composition analysis.

Curated master list of biomechanics datasets, tools, and resources

Links to dozens of gait, bone, musculoskeletal datasets and processing tools. Best single starting point.