The Deep Learning Model For The Examination Of Brain Tumor

Radiologists can locate the tumor location using the experimental medical imaging technique known as magnetic reasoning imaging (MRI). However, manually testing the MRI pictures is time-consuming and requires experience. The development of Computer-assisted Diagnosis (CAD), and deep learning in particular, now enable radiologists to diagnose brain tumors more accurately. Deep learning techniques can be developed to produce accurate classification results without requiring manual feature extraction. The research article suggests the deep learning model for classifying multi-class (pituitary, glioma, meningioma, and no-tumor) and binary-class (tumor, no-tumor) brain tumor datasets. Two datasets were used to test the proposed model: Dataset-1 contains 3264 photos, while Dataset-2 contains 3000 images. The model consists of three convolution blocks whose kernel size is 10, 8, and 6, respectively. The model has three max-pooling blocks. The model also contains one flattened block and two fully connected blocks. Finally, the model included softmax and sigmoid activation functions for the multi-class and binary-class brain tumor image classification. The created deep learning model achieved 94.22% classification accuracy with multi-class brain tumor dataset-1. The same model gained 97.78% classification accuracy for binary-class brain tumor dataset-2. The created deep learning model performs similarly to VGG19 and works well with a smaller dataset than CNN.