Meeting Banner
Abstract #1481

Accelerated Fitting for Quantitative Magnetization Transfer in Glioblastoma Multiforme Patients with Uncertainty using Deep Learning

Matt Hemsley1,2, Rachel W Chan2, Liam Lawrence1,2, Sten Myrehaug3,4, Arjun Sahgal2,3,4, and Angus Z Lau1,2
1Medical Biophysics, University of Toronto, Toronto, ON, Canada, 2Department of Physical Sciences, Sunnybrook Research Institute, Toronto, ON, Canada, 3Department of Radiation Oncology, Sunnybrook Health Sciences Centre, Toronto, ON, Canada, 4Department of Radiation Oncology, University of Toronto, Toronto, ON, Canada

qMT has been suggested as a biomarker in Glioblastoma patients. However, reconstruction involves a computationally expensive fitting procedure involving the Bloch-McConnell equations. In this work, the use of neural networks was investigated to perform the fit and to compute uncertainty heatmaps to identify regions of potential error. The dataset consisted of 164 scans from N=41 glioblastoma patients (33=training, 8=testing). Models were evaluated using MAE and correlation in the whole-head volumes and specific ROIs. The model output agreed with a conventional curve-fitting algorithm (r=0.93, and <1% error) with speed up factors of 240000x. Uncertainty predictions were correlated with prediction error (r=0.59).

This abstract and the presentation materials are available to members only; a login is required.

Join Here