A convolutional neural network (CNN) based project for prediciton of cancer with inputs as DCM files (3d Data)
Accuracy on the first pass : 0.59
The dataset had around 1.5k+ of datasets which is realtively very low for working on a CNN model.
Apart from it the data was unbalanced as 1.2k approx data of non cancer percents and the rest for the cancerous one giving approx 74% for no-cancer as a model
Tensorflow, Numpy, Pandas, Matplotlib, Pydicom, OS (optional for preprocessing), Math and CV2.
The dataset can be downloaded from the torrent file available in the repository.