Segmentation of brain tumors

In this project, brain tumors will be segmented by means of images in order to detect them more consistently. For this, several already trained models that perform the segmentation were used.

Dataset

In this project, a database of 115 images with their respective annotations was used.

VGG-U-Net

VGG-SegNEt

VGG-PSPNe

Renet50-U-Net

Renet50-Segnet

Renet50-PSPNet

Downloading models

6 similar models (vgg and ResNet 50) were selected among them for the classification but with different architectures that perform the segmentation (U-Net, PSPNet, and SegNet)

Test images.

ResNet50-SegNet

ResNet50-U-Net

ResNet50-PSPNet

Results

After using these different models, only a few were effective, as are all those that have ResNet 50 as a classifier.

For more details, you can see the report here.