2025
Deep Learning Enables Large-Scale Shape and Appearance Modeling in Total-Body DXA Imaging
Bunnell A, Cataldi D, Glaser Y, Wolfgruber T, Heymsfield SB, et al.
Lecture notes in computer science
University of Hawaiʻi Cancer Center
Bridging computational imaging and cancer research
The Shepherd Research Lab develops quantitative imaging, body-composition, and AI-assisted methods for earlier cancer detection and better metabolic-risk prediction — grounded in real studies, serving the community.
Aloha
At SRL, we are a group of research scientists innovating the ways we detect cancer and metabolic risk. Along with how this benefits all areas of health, we are especially proud of the benefits our studies can bring to our local community on Oʻahu.
The lab is led by Dr. John A. Shepherd at the University of Hawaiʻi Cancer Center, with studies spanning breast cancer detection through breast density, metabolic risk prediction through body-composition analysis, and deep learning for AI-assisted diagnosis.
Research Areas
Three linked research threads. Imaging grounds the methods, body analysis bridges the biology, and AI lifts the diagnostics.
Cancer
Breast cancer detection and risk assessment through quantitative imaging and breast density analysis.
Body Composition
3D body scans and body-composition analysis to predict metabolic health status and chronic disease risk.
AI for Health
Deep learning for AI-assisted diagnosis, tissue classification, and novel biomarker discovery.
Visit the Lab
Body Composition Lab
The lab operates the Body Composition Lab (BCEPEM) — a core facility at the UH Cancer Center offering DXA, Bod Pod, 3D optical scans, BIA, and muscle dynamometry to researchers and the community.
Latest News
All news →Featured Research
Three-Compartment Breast Lesion Detection
Leong LT, Malkov S, Drukker K, Niell BL, Sadowski P, Wolfgruber T, et al. Dual-energy three-compartment breast imaging for compositional biomarkers to improve detection of malignant lesions. Communications Medicine. 2021;1(1):29.
Adding three-compartment breast (3CB) features to computer-aided detection yields an area under the ROC curve of 0.81 — compared to 0.69 for CAD alone. The integrated discrimination improvement demonstrates substantial benefit from adding 3CB features.
Latest Publications
All publications →2025
High dietary total energy intake and energy density are associated with girls’ breast density during puberty
Carrasco C, Garmendia ML, Martínez-Arroyo A, Shepherd J, Pereira A
European Journal of Nutrition
2025
Characterizing cancer patterns in Okinawan vs. mainland Japanese Americans: The Multiethnic Cohort Study
Streicher SA, Guillermo C, Park S, Chiang CWK, Shepherd J, et al.
medRxiv