An Evaluation of Medical Image Analysis using Image Segmentation and Deep Learning Techniques for Brain-Tumour

An aberrant mass of rapidly proliferating brain cells, known as a brain tumour, may develop into various forms of cancer. The segmentation of brain tumours by MRI scans is a challenging but crucial process with several medical analytic applications. Radiologists or clinical specialists are responsible for performing a tedious and time-consuming process that involves identifying, segmenting, and extracting contaminated tumour regions from MRI images. The different anatomical structures of human organs may be imagined via image-processing ideas. The human brain is notoriously difficult for basic imaging techniques to identify abnormal structures. A large number of researchers looked at several rapid and accurate algorithms for brain tumour (BT) detection and classification. Recent advances in DL-based automated systems have allowed for faster and more accurate segmentation of DT. This research makes use of a DenseNet50 model for medical picture classification of brain malignancies that is built on a pre-trained CNN model made possible by DL. The model is available on Kaggle. An optimiser like Adam and a categorical cross-entropy loss function were utilised to train a model. A suggested DenseNet50 model outperformed other comparative advanced DL models and attained an experimental accuracy of 95%. The visual results also highlight the impressive outcomes of the MRI brain tumour segmentation system, which will help doctors automatically detect and treat brain tumours. The results acquired demonstrate the effective identification of tumorous images, which is extremely beneficial for the diagnosis process and future applications.