John A Shepherd , PhD
Researcher R5Population Sciences in the Pacific Program (Cancer Epidemiology), University of Hawai‘i Cancer Center
Work Email: johnshep@hawaii.edu
Research Interests
Dr. John Shepherd’s research integrates medical imaging, deep learning, and public health, with an emphasis on breast cancer, obesity, diabetes, and osteoporosis. As Chief Scientific Officer at the University of Hawaii Cancer Center, he leads NIH-funded projects, including the Hawaii Pacific Islands Mammography Registry, which explores imaging biomarkers across diverse ethnic groups. Dr. Shepherd's extensive work in deep learning has advanced the application of AI to improve cancer detection and body composition analysis. His lab focuses on reducing health disparities using cutting-edge imaging and machine learning techniques, offering rich opportunities for prospective PhD students.
Recent Publications
- Tian IY, Wong MC, Nguyen WM, Kennedy S, McCarthy C, Kelly NN, Liu YE, Garber AK, Heymsfield SB, Curless B, Shepherd JA. Automated body composition estimation from device-agnostic 3D optical scans in pediatric populations. Clin Nutr. 2023 Sep;42(9):1619-1630. PubMed Central PMCID: PMC10528749.
- Wong MC, Bennett JP, Leong LT, Tian IY, Liu YE, Kelly NN, McCarthy C, Wong JMW, Ebbeling CB, Ludwig DS, Irving BA, Scott MC, Stampley J, Davis B, Johannsen N, Matthews R, Vincellette C, Garber AK, Maskarinec G, Weiss E, Rood J, Varanoske AN, Pasiakos SM, Heymsfield SB, Shepherd JA. Monitoring body composition change for intervention studies with advancing 3D optical imaging technology in comparison to dual-energy X-ray absorptiometry. Am J Clin Nutr. 2023 Apr;117(4):802-813. PubMed Central PMCID: PMC10315406.
- Glaser Y, Shepherd J, Leong L, Wolfgruber T, Lui LY, Sadowski P, Cummings SR. Deep learning predicts all-cause mortality from longitudinal total-body DXA imaging. Commun Med (Lond). 2022;2:102. PubMed Central PMCID: PMC9381587.
- Zhu X, Wolfgruber TK, Leong L, Jensen M, Scott C, Winham S, Sadowski P, Vachon C, Kerlikowske K, Shepherd JA. Deep Learning Predicts Interval and Screening-detected Cancer from Screening Mammograms: A Case-Case-Control Study in 6369 Women. Radiology. 2021 Dec;301(3):550-558. PubMed Central PMCID: PMC8630596.
Recent Grants
- R01HD103885 Shepherd (PI) 8/9/21-5/31/26 NIH/NICHD Quantifying Body Shape in Pediatric Clinical Research The long-term goal of the Shape Up! Keiki is 1) to provide pediatric phenotype descriptors of health derived from detailed body shape scans from high-speed and high depth resolution 3D cameras, and 2) to provide the tools to visualize and quantify body shape in research and clinical practice.
- R01CA257652 Shepherd (PI) 8/9/21- 7/31/26 NIH/NCI Lesion Composition and Quantitative Imaging Analysis on Breast Cancer Diagnosis The aims of this project are 1) develop q3CB lesion signatures for distinguishing breast cancer lesions from benign lesion, 2) compare radiologists' performance without and with the inclusion of q3CB signature, and 3) investigate the utility of q3CB lesion signatures to improve sensitivity and specificity on CADe-identified suspicious lesions in the tasks of assessing malignancy and associating cancer subtypes.
- R01CA263491 Shepherd (PI) 4/1/23-3/31/28 NIH/NCI Hawaii Pacific Islands Mammography Registry (HIPIMR) Despite recent advances in early detection and treatment, breast cancer remains a major cause of morbidity and mortality. The HIPIMR will be the source to identify and validate novel image biomarkers for the diverse ethnic groups of this region with varying risk factor profiles including Native Hawaiians, Japanese, Filipino, Chinese, and other ethnic groups. This study addresses the need for accurate identification of defined clinical and radiomic risk factors among AANHPI populations and their relation to breast cancer risk to improve outcomes for these women.
- R01DK111698 Shepherd (PI) 10/01/16- 12/31/22 NIH/NIDDK Shape Up! Kids The long-term goal of the Shape Up! Kids Study is 1) to provide pediatric phenotype descriptors of health using body shape, and 2) to provide the tools to visualize and quantify body shape in research, clinical practice, and personal health assessment.
- R01DK109008 Shepherd (PI) 05/15/16-04/30/21 NIH/NIDDK Optical Body Composition and Health Assessment The long-term goal of the Optical Body Shape and Health Assessment Study is 1) to provide phenotype descriptors of health using body shape, and 2) to provide the tools to visualize and quantify body shape in research, clinical practice, and personal health assessment.
Current Graduate Students
- Lydia Sollis, PhD in Computer Science at the University of Hawai‘i at Mānoa 2024 – Present
- Nusrat Zaman Zemi, MSc in Electrical and Computer Engineering at the University of Hawai‘i at Mānoa 2023 – Present
- Arianna Bunnell, PhD in Computer Science at the University of Hawai‘i at Mānoa 2022 – Present
- Dustin Valdez, PhD in Nutrition Science at the University of Hawai‘i at Mānoa (Chair) Thesis: Evaluating Portable Breast Cancer Screening Technology for Women in the Pacific 2019 – Present