'Automatic Identification of MRI Phenotypes'
Date
Killian Pohl, PhD
Assistant Professor
Department of Radiology
Bioengineering Graduate Group
University of Pennsylvania
Philadelphia, PA
This talk focuses on automatically detecting disease specific MRI phenotypes to advance understanding, detection, and treatment of diseases in clinical settings. Improved scanning technology produces MR images that capture healthy tissue and pathology at higher resolution and dimension than ever before. The increase in data complexity often obscures the identification of imaging phenotypes. To aid MRI analysis, we create novel methods to extract accurate quantitative image markers. Our algorithms learn the underlying structure across images as well as condense individual scans to a few parameters. This image encoding has already been valuable for different applications, such as studying the long term effects of corrective surgery for Tetralogy of Fallot, predicting the conversion of patients with mild cognitive impairment to Alzheimer, and analyzing the spatial distribution of brain gliomas. Our technology can capture subtleties often missed by volumetric based methods and is an important part of our vision to improve personalized medicine by presenting clinicians with quantitative markers encapsulating the entire medical record.
Please note that Dr. Pohl is an applicant for the Engineer/Mathematician Faculty Position within the Department of Radiology & Biomedical Imaging.
America/Los_Angeles publicType
Time Duration
Killian Pohl, PhD
Assistant Professor
Department of Radiology
Bioengineering Graduate Group
University of Pennsylvania
Philadelphia, PA
This talk focuses on automatically detecting disease specific MRI phenotypes to advance understanding, detection, and treatment of diseases in clinical settings. Improved scanning technology produces MR images that capture healthy tissue and pathology at higher resolution and dimension than ever before. The increase in data complexity often obscures the identification of imaging phenotypes. To aid MRI analysis, we create novel methods to extract accurate quantitative image markers. Our algorithms learn the underlying structure across images as well as condense individual scans to a few parameters. This image encoding has already been valuable for different applications, such as studying the long term effects of corrective surgery for Tetralogy of Fallot, predicting the conversion of patients with mild cognitive impairment to Alzheimer, and analyzing the spatial distribution of brain gliomas. Our technology can capture subtleties often missed by volumetric based methods and is an important part of our vision to improve personalized medicine by presenting clinicians with quantitative markers encapsulating the entire medical record.
Please note that Dr. Pohl is an applicant for the Engineer/Mathematician Faculty Position within the Department of Radiology & Biomedical Imaging.