A Brain Tumor Detection and Classification Model on Roboflow dataset Using Inception RestNet50

The tenth biggest cause of death for both men and women is brain cancer, which claims thousands of lives every day. An early diagnosis and effective treatment of brain tumors may save their life. To identify malignancies of brain tumors, the medical community needs a method that is rapid, automated, effective, and trustworthy. The ideal standard for cancer detection is a biopsy. However, the biopsy procedure is extremely risky and time-consuming. The conventional methods of diagnosing brain cancers, such as "computed tomography (CT") and "magnetic resonance imaging (MRI)", have yielded helpful information regarding the localization and features of tumors. However, due to brain tumor intricacy and variety, medical practitioners find it difficult and time-consuming to evaluate these images. Using the power of machine learning (ML) algorithms, healthcare professionals and researchers have devised automated brain tumor detection systems to overcome these limitations. Recent years have seen a tremendous increase in the success of approaches that rely on Artificial technique (AI) for identifying and classifying tumors. This study develops an ensemble technique for the use of brain MRI to identify and classify brain tumors and their phases (MRI). Tumor detection and classification from an MRI image are performed using a modified Inception ResNet50 pre-trained model and dense layer. The proposed model classifies the input MRI image into four classes namely "glioma tunor (GT), meningioma tumor (MT), no tumor (NT) and pituitary tumor (TP)". As shows in the experiment result that the proposed models for tumor identification and classification achieve accuracies of 99% during training and 98% during testing.