An Artificial Intelligence Based Deeplearning Technique for Recognition of Multiple Brain Tumors on MRI Imager

Brain tumors were devastating ailment found on all continents. Healthy brain tissue is ravaged by tumors, which can spread to other parts of the body. Thereby, rapid tumor cell proliferation could be fatal Manual diagnosis is not able to effectively track and categorize tumor kinds. Early brain tumor detection becomes more important since it may help determine the patient’s condition and prevent harmful outcomes. Early brain tumor detection using Magnetic Resonance Images (MRI) may increase the lifespan of patients. The most widespread kinds of brain cancers were meningioma, glioblastoma, and astrocytoma. In this work, brain malignancies such as meningioma, glioblastoma, and astrocytoma are identified and categorized by employing Convolutional Neural Network (CNN) classifier in Deep Learning (DL). While DL techniques might not need any manual feature extraction while attaining appropriate classification outcomes, prior techniques that used machine learning (ML) required manual feature extraction. The Otsu multi-level thresholding method categorizes the medical images into segments for tumor identification after preprocessing the images for scaling and noise reduction. The Gray Level Co-Occurrence Matrix (GLCM) textural method of analysis is used to generate the features. For tumor type identification, a pre-trained CNN model is employed. It indicates no tumor if there is none. The tumor kind is classified otherwise. In the end, the accuracy of the demonstrated model was 98.31%. The other statistical metrics, such as SSI, PSNR, sensitivity, specificity, and MSE, were calculated and contrasted with the practices currently in application.