Brain Tumor Detection Using Image Processing

Abstract

This academic paper presents a novel method for identifying brain tumors by utilizing sophisticated (CNN) architectures. This so many hard detect brain tumor cell for a doctor. So, our main goal is to detect brain tumor very easily. So that patient can get treatment at the right time. Transfer learning involves using pre-train models, such as EfficientNet, ResNet-50, MobileNet and InceptionV3. The results of the experiments show encouraging levels of accuracy for the suggested approach. EfficientNet exhibited remarkable performance, achieving impeccable accuracy at a success rate of 93%. Inception v3 attained an accuracy rate of 92%. MobileNet achieved an outstanding accuracy rate of 91%. ResNet-50 demonstrated marginally reduced accuracy levels, achieving 61%. They remained effective in identifying brain tumors, each in their own way. Here I used some image preprocessing technique like Image Scaling, Crop, Blur, Gaussian Noise, Salt Pepper, Color Juttering. This technique help improve accuracy. The system is developed using GoogleColab.