Overview / Usage

So medical images are very difficult to store and they come is splices with data formats that are difficult to comprehend, so we plan do first combining the splices to generate a 3D image of the lung using the ITK toolkit or Pydicom. The next step involves accurately identifying and training the model for classifying the disease and create a bounding box around it. In the next phase we try to use a Encode Decoder architecture to use Sematic Segmentation and Instance Segmentation to accurately identify and operate on diseased section with atmost accuracy.

Methodology / Approach

ITK and PyDicom to convert the images and get the 3D images. We have the annotation and the coordinates in an excel sheet so using that we creat bounding boxes and then train the model using tensorflow with intel accelarated python.