Integrated Deep Learning Approach for Brain Tumor  Detection and Segmentation in MRI Images

Authors

Bassam A. Jaradat Department of Management Information Systems,Friends University, Kansas, USA.
Khaled Bawaneh School of Business ,Clark Atlanta University, Goergia,USA

Deep learning, Classification, Segmentation, MRI brain tumor.

This study examines the use of deep learning techniques for magnetic resonance imaging (MRI)-based brain

tumor categorization in lower-grade gliomas by applying the manually segmented FLAIR abnormality masks that

were acquired from The Cancer Imaging Archive (TCIA) and the LGG Segmentation Dataset, which include brain

MR images. We propose a classifier model based on ResNet50 and a segmentation model named U-NET, leveraging

knowledge from earlier research investigating the relationship between form data collected by deep learning

algorithms and genetic subtypes of lower-grade gliomas. This model reaches a remarkable accuracy of 94.75% when

trained to identify whether tumors are present or absent. F1-score, precision, and recall measures are included in the

evaluation to give a thorough understanding of the models functionality. These findings highlight the promise of

sophisticated image processing methods for precise and automated brain tumor classification, with ramifications for

improving neuro-oncology clinical processes and diagnostic accuracy.

Download data is not yet available.