Reliable Automated Diagnosis of Soft Tissue Tumor using Machine Learning

Abstract

Soft tissue tumours are hard to diagnose because they look a lot like other brain conditions on Magnetic Resonance Imaging(MRI)scans, which can lead to delayed or wrong treatment. To solve this problem, we suggest a diagnostic system based on machine learning that combines preprocessing methods with a Convolutional Neural Network (CNN) to accurately classify tumours. The model sorts MRI images into four groups: glioma, meningioma, pituitary tumour, and no tumour. It uses Grad-CAM heatmaps to show and emphasise tumour areas, which makes the results easier to understand and more trustworthy for doctors. To make models work better on small datasets, data preprocessing includes resizing, normalisation, and augmentation. Experimental results show that the CNN can accurately find tumour areas and classify them correctly, even with limited data. The suggested method lessens the need for subjective manual interpretation, which lowers the chance of misdiagnosis.