Enabling Accurate Brain Tumor Segmentation with Deep Learning
R. Sangeetha, Pramod Vishwakarma, Satvik Vats, Lekha Rani, J. Logeshwaran
Abstract
The segmentation accuracy of brain tumors has been improved through the application of deep learning. According to current research, deep learning may produce high-quality segmentation of brain tumor areas in Magnetic Resonance Imaging (MRI) data. The diagnosis of brain cancers can be aided by using these segmented sections. Fully convolutional networks (FCNs) and 3D U-nets, two deep learning algorithms, have shown to be more accurate than more conventional methods such level set methods, active contours, and graph cuts. FCNs provide an end-to-end solution for segmentation, while 3D U-nets are designed to capture spatial context and incorporate prior information about tumor shapes, enabling more accurate segmentation results. These techniques have also been utilized to segment gliomas and meningiomas, two different forms of brain tumors. In the upcoming years, it is anticipated that deep learning would improve accuracy and diagnostic precision in the field of brain tumor segmentation