Deep learning for noninvasive management of brain tumors

Document Type

Book Chapter

Publication Title

Augmenting Neurological Disorder Prediction and Rehabilitation Using Artificial Intelligence

Abstract

Computerized detection and diagnosis of cancer, based on medical image analysis, promises to be a good alternative to the conventional biopsy test. Since repeated biopsy in the brain is challenging and lead to severe side effects, the role of noninvasive image guided diagnostic techniques become necessary; with Magnetic Resonance Imaging being the standard diagnostic tool typically used. This paper discusses the development of computer-aided systems based on deep Convolutional Neural Network (CNN) for improved detection, diagnosis, and prognosis of brain cancer. An encoder-decoder type strategy, to produce the final volumetric segmentation of the tumor and its subregions. The power of deep CNN models is explored for the detection and classification of brain tumors from multisequence magnetic resonance images. The suitability of transfer learning is also studied for this task through fine-tuning of pretrained models.

First Page

15

Last Page

34

DOI

10.1016/B978-0-323-90037-9.00015-1

Publication Date

1-1-2022

This document is currently not available here.

Share

COinS