Deep learning based classification of multiple sclerosis lesions

Several works of literature have been published in recent years that describe ways for automatically segmenting multiple sclerosis (MS) lesions. Brain lesions that may be seen on MRI scans (magnetic resonance images), which are the primary diagnostic tool for this condition, are one of the defining...

Full description

Saved in:
Bibliographic Details
Main Authors: Mukil, A., Kumar, S., Selvaperumal, S. K., Lakshmanan, Ravi
Format: Conference Proceeding
Language:English
Subjects:
Brain
Deep learning
Image segmentation
Lesions
Magnetic resonance imaging
Medical imaging
Multiple sclerosis
Myelin
Sheaths
Tumors
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Several works of literature have been published in recent years that describe ways for automatically segmenting multiple sclerosis (MS) lesions. Brain lesions that may be seen on MRI scans (magnetic resonance images), which are the primary diagnostic tool for this condition, are one of the defining characteristics of this illness. We demonstrate, via the employment of Integrated Gradients attributions, that the utilisation of brain tissue probability maps as deep network input, rather than raw MR images. When there is a conformational change in the myelin sheath, multiple sclerosis may develop. In many cases, multiple sclerosis may be diagnosed with the use of magnetic resonance imaging. The purpose of this work is to provide an overview of the applications of deep learning in molecular imaging, specifically with regard to the segmentation of tumour lesions, the categorization of tumours, and the prediction of patient survival.
ISSN:0094-243X
1551-7616
DOI:10.1063/5.0229416