Meeting Banner
Abstract #0884

QSMRim-Net: Fusing Radiomic and Convolutional Features for Identification of Chronic Active MS Lesions on Quantitative Susceptibility Maps

Hang Zhang1, Jinwei Zhang2, Melanie Marcille3, Pascal Spincemaille4, Thanh D. Nguyen3, Susan A. Gauthier3, Yi Wang3, and Elizabeth M. Sweeney5
1Electrical & Computer Engineering, Cornell University, New York, NY, United States, 2Biomedical Engineering, Cornell University, New York, NY, United States, 3Department of Radiology, Cornell University, New York, NY, United States, 4Cornell University, New York, NY, United States, 5Department of Population Health Sciences, Cornell University, New York, NY, United States

Synopsis

Chronic active multiple sclerosis (MS) lesions are characterized on Quantitative susceptibility mapping (QSM) by a paramagnetic rim (rim+) at the edge of the lesion. We present QSMRim-Net, a deep neural network that fuses lesion-level radiomic and convolutional image features together for automated identification of rim+ lesions on MRI. On the lesion-level, using five-fold cross validation, the proposed QSMRim-Net detected rim+ lesions with an area under the receiver operating characteristic curve of 0.965 and an area under the precision recall curve of 0.655. QSMRim-Net out-performed other state-of-the-art methods on both metrics.

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

Join Here