Details
Joyita Dutta, Ph.D.
Professor of Biomedical Engineering
UMass Amherst
Abstract: Alzheimer’s disease (AD) is a debilitating neurodegenerative disorder and a looming public health challenge. Amyloid-β plaques and tau tangles, the two pathophysiological hallmarks of AD, are believed to play key mechanistic roles in the disease, serve as potential therapeutic targets, and can be imaged in vivo using positron emission tomography (PET). This talk will discuss novel signal processing and artificial intelligence (AI) approaches that hold great promise in AD diagnosis and prognosis. The first part of the talk will focus on AI-based PET image enhancement. The primary factors compromising PET image quality are low spatial resolution and high noise. The talk will present novel deep learning approaches for PET image super-resolution and denoising, demonstrating their applications to AD neuroimaging datasets. The second part will focus on automated staging of tau pathology using self-supervised contrastive learning on tau PET data. Tau tangles exhibit heterogeneous deposition across brain regions, complicating efforts to define clear progression stages. The talk will present a contrastive staging framework that learns feature representations from tau PET without labeled data, identifying distinct stages that align with progression trajectories for disease biomarkers and cognitive measures. The final part of the talk will focus on digital phenotyping of AD based on an individual’s sleep patterns. Sleep disturbances are among the earliest observable signs of AD. The talk will cover AI-based models for predicting cognitive status from sleep monitoring data and for automated sleep staging using wearable devices, including preliminary results from a smartwatch-based human sleep study conducted in our lab.
Bio: Dr. Joyita Dutta is a Professor with Tenure in the Department of Biomedical Engineering at the University of Massachusetts Amherst. She received her B.Tech. (Honors) from the Indian Institute of Technology Kharagpur and M.S. and Ph.D. from the University of Southern California. She directs the Biomedical Imaging and Data Science Laboratory (BIDSLab) at UMass Amherst, which develops signal processing and artificial intelligence (AI) techniques for image, graph, and time-series datasets. Her scientific contributions include the development of a broad range of tools for medical image enhancement and reconstruction with a focus on multimodality information integration. Dr. Dutta was the recipient of the 2016 Tracy Lynn Faber Memorial Award from the Society of Nuclear Medicine and Molecular Imaging (SNMMI) and the 2016 Bruce Hasegawa Young Investigator Medical Imaging Science Award from the IEEE. She received an SNMMI Young Investigator Award, the 2013 SNMMI Mitzi & William Blahd MD Pilot Research Grant, the 2013 American Lung Association Senior Research Training Fellowship, and an NIH K01 Career Development Award. Her research has supported by multiple grants, including NIH R01, R21, and R03 grants held as PI. Dr. Dutta is a founding member of the SNMMI AI Task Force. She was the Program Co-Chair for the 2022 IEEE Medical Imaging Conference (MIC) in Milan, Italy and will serve as the Chair of the 2027 MIC in Pasadena, CA. She served as the President of the SNMMI Physics, Instrumentation and Data Sciences Council (PIDSC) from 2024-2025. Her trainees at BIDSLab have been awarded prestigious extramural training grants from the SNMMI, IEEE, APS, and AAUW. She was named a UMass Translational Innovation Fellow in 2025.