Evaluation of Various Loss Functions and Optimization Techniques for MRI Brain Tumor Detection

Abstract

Brain tumours can appear in any area of the brain and can afflict people of any age. Recent research has demonstrated that deep learning models such as VGG16, VGG19, Mobile Net, and Dense Net are promising methods for detecting brain cancers using MRI technology. This research was intended to determine the optimal mix of optimisation algorithms and loss functions for brain tumour detection accuracy. The entire cerebral cortex was labelled to show the existence of brain tumours using an open-source dataset. When training deep learning models, different loss functions were used to measure how well they performed. The ADAM optimizer was more accurate and had less loss than the BINARY CROSSENTROPHY optimizer. The proposed method found brain cancer in 97% of MRI scans, so it can be used for both primary diagnosis and clinical decision support systems. The proposed method could make clinical diagnosis and decision-making more accurate. In conclusion, deep learning models, including those with binary cross-entropy and ADAM losses, can classify brain tumour data accurately, and the proposed method is effective and simple to implement.