Abstract

The talk will be on three recent biomedical projects at the BiCLab, Florida Tech.

Tomographic image reconstruction: 3D Single Photon Emission Computed Tomography or SPECT images are being reconstructed from 2D camera views from multiple angles (sinograms) using AI models. Conventional iterative reconstruction algorithms are slow and cannot directly address systemic noise, like that from cardiac motion. Our model compensates for noise while reconstructing the image more efficiently.

Alternative mathematical tool for Dynamic Imaging with Positron Emission Tomography (PET) for quantitation of physiological parameters like blood flow rate. Quantitative images are independent of scanners or imaging protocols, and are more useful in diagnosis /prognosis of diseases than the conventional qualitative or "static" imaging. Dynamic imaging is yet to be fully adopted in clinics for challenges that we expect to alleviate.

We will briefly mention use of the same mathematical tool in analyzing bio-sequences to detect simultaneous multiple mutations (Co-mutational Positions or CMPs) present in a database in significant numbers. We applied it to the Receptor Binding Domain sequences of Sars-Cov-2 (approximately 700,000 sequences, each of approx 600 base-pairs long). Our method may find new sets of simultaneous mutations previously unnoticed by the biologists.

Currently a Professor at the Florida Institute of Technology Debasis Mitra is a Physicst and a Computer Scientist. Professional objective of Debasis Mitra is to solve problems in science, engineering and medicine by using computer science and mathematics. The current theme of his research is to squeeze knowledge nuggets from noisy spatio-temporal data, especially in Bio-medical images. A long term vision is to understand the molecular biological processes for improving human health. In the past Mitra made contributions in artificial intelligence studying the qualitative reasoning problems with spatial and temporal constraints.