Automatic Detection of Microaneurysms in Fundus Images
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
train
README.md
afterclass.sh
afterpeak.sh
lmr.c
output.c
peak.c
prebatch.sh
single.sh

README.md

Microaneurysm Detection in Fundus Images

Objective and Developing Tools

This project is to automatically detect microaneurysms in fundus images, and is built in VisionX system

Implemented stages

The sequence of the following code and scripts correspond to the experiment stages in the reference paper.

  • prebatch.sh
    • Image Preprocessing in batches including inverting green channel and gaussian filtering
    • This part utilizes VisionX image processing commands
  • lmr.c
    • Local Maximum Region Extraction
  • peak.c
    • Cross-Sectional Scanning
    • Peak Detection
    • Property Measurement on the Cross-Section Profile
    • Feature set building
    • Cross-section scanning is implemented by Ling Zeng
  • afterpeak.sh
    • Classification
    • Uses VisionX command vrclasstt here
  • afterclass.sh
    • Nonmaximum Suppression
  • output.c
    • Final Output
    • With one input fundus image, it can output detected MA coordinates in a .csv file (one column y, one column x) and an image annotated with circles.

Reference

Lazar, I., & Hajdu, A. (2013). Retinal microaneurysm detection through local rotating cross-section profile analysis. IEEE transactions on medical imaging, 32(2), 400-407.