Brain Tumor Recognition by Machine Learning with Decorrelation Stretch Image Enhancement Based on MRI Images

The brain tumor is an abnormal mass of substance in which the proliferation of the cells is abnormally rapid and uncontrollable. It is common practice to distinguish between malignant and benign tumors during tumor development. Since brain tumors have the potential to be fatal, it is critical to identify and illustrate their presence with MRI pictures. Many processes, including preprocessing, feature extraction, and image augmentation, may be included in the detection strategy. The outcome of these techniques mostly depends on the algorithm's performance and the quality of the images. This work offers a system that analyzes MRI scans to determine if the brain is tumor-free or has a tumor and uses the CNN algorithm to do it. To support our statement, we first improve the image enhancement techniques, which include three different stages: Using the Decorrstretch picture improvement technology, noise is removed. Ultimately, a convolutional neural network will be trained to identify whether the MRI pictures of the human brain are normal or damaged. The method achieved 96.7%, which is good accuracy compared to older techniques. The method's results validate our premise that using image enhancement to detect brain tumors is feasible and can lead to improved performance.