Brain Tumor Detection APP

The deployment model consist of backend and frontend.
The backend is deployed using Render.com with FastAPI Framework.
The frontend is deployed using Vercel.com with Next.js Framework.

The preprocessing consist of:

  • Grayscaled: convert 3 class image to 1 class
  • Smoothing: Reduce noise of the image using GaussianBlur
  • Enhancing: Increase contrast of the smoothed image using CLAHE

The Feature Extraction consist of:

  • Local Binary Pattern (LBP): Extract brain tumor texture
  • Histogram Oriented Gradient (HOG): Extract the brain tumor shape/structure
  • Gray Level Co-Ocurrence Matrix (GLCM): Extract brain tissue texture
  • Then these three features are stacked vertically

Accuracy (Precision and Recall)

accuracy

Confusion Matrix

confusion matrix