Retinal Blood Vessel Segmentation
Motivation :
Automatic segmentation of medical images is an important step to extract useful information that can help doctors make a diagnosis. For example, it can be used to segment retinal vessels so that we can represent their structure and measure their width which in turn can help diagnose retinal diseases.
Data
Available here
Formulation of the problem:
Each pixel must be labeled “1” if it is part of a blood vessel in the image, and “0” if not.
Solution:
Retinal_Blood_Vessel_Segmentation.ipynb
This is a simple implementation neural net with architecture UNet on pytorch, as part of the extension for MedEyeService.
Results :
