Building a detection model using a convolutional neural network in Keras.
The dataset contains 2 folders: yes and no which contains 253 Brain MRI Images.
The folder yes contains 155 Brain MRI Images that are tumorous and the folder no contains 98 Brain MRI Images that are non-tumorous.
Since the dataset is small I have used an augmented version of the dataset available at https://github.com/MohamedAliHabib/Brain-Tumor-Detection
After data augmentation, now the dataset consists of: 1085 positive and 980 examples, resulting in 2065 example images.
these 2065 examples contains also the 253 original images. They are found in folder named 'augmented data'.
I have used the negative of the input image as the boundary of the tumor has a clear separation from the boundary. So, I thought that this would help in better feature extraction.
Test accuracy 86.45 %.
I stored the test images in data ( unzip the testimages numpy array before using the test set and weights).
I used the weights for epoch 48.