The vast majority of existing image analysis algorithms are trained on slide images acquired via expensive Whole-SlideImaging (WSI) scanners. High scanner cost is a key bottleneck preventing large-scale adoption of automated digital pathology solutions in the developing countries. In this work, we investigate the possibility of performing automated image analysis and quantification using images captured from the eyepiece of a microscope via a smartphone.
The app is built using python's flask framework and screenshot of homepage is shown below:
For getting the mitotic cell predictions, we used a Faster R-CNN model. The qunatification workflow of the system is shown below:
Prediction produced by the model on some sample images are shown below. Left is the input input and on the right is the image with predictions.
Quantification results compared with the ICPR high-res dataset are shown in the tables below:
