SCIS Event

default


Automatic Detection and Quantitative Dynamic Contrast-Enhanced Magnetic Resonance Imaging Scoring of Prostate Cancer Aggressiveness



Date / Time:

Apr. 29, 2016 @ 11:00 am

Where:

Florida International University
ECS: 349

Speaker:

Andres Parra
University of Miami
Miller School of Medicine

Andres Parra


Description

Routine multi-parametric MRI (MP-MRI) acquisition consists of T2-weighted imaging (T2W), Diffusion Weighted imaging (DWI) and Dynamic Contrast Enhanced (DCE) imaging. Although DCE imaging has been used for qualitative assessment of both location and aggressiveness of prostate cancer, methods to consistently estimate quantitative pharmacokinetic models of the DCE signal present severe shortcomings either in compartment (Tofts) and heuristic-based approaches. Recent efforts to quantify and standardize the diagnostic process using MP-MRI (PI-RADS, Prostate Imaging Reporting and Data System) have significantly limited the clinical role of DCE. In this study I present a method based on non-negative matrix factorization (NMF) that allows the identification of a well-perfused region of interest (ROI) in the prostate using DCE-MRI and subsequently allows the computation of several features which significantly correlate with histopathology, both for MRI guided and template (random) biopsies.


Bio

N. Andres Parra is currently an Associate Scientist in the Department of Radiation Oncology at The University of Miami Miller School of Medicine. He holds a PhD in Computer Science from FIU. His research interests are related to the extraction of clinically relevant information from medical imaging that is associated with histopathology and cancer gene expression profiles, as well as the translation of these computational techniques to the clinical setting. Ongoing projects investigate the fusion of multi-parametric MRI for the diagnosis and management of prostate cancer, and the inclusion of MR spectroscopy (metabolic) imaging maps in the radiotherapy treatment planning of Glioblastoma Multiforme.