top of page
Search

November Week 1 Summary

  • Writer: Paul Flowers
    Paul Flowers
  • Nov 7, 2017
  • 1 min read

During this week’s meetings, the team was invested in improving the local and global thresholding for blood vessel detection. Here are the results of testing Otsu local and global thresholding on the retinal images.

As you can see, the global thresholding method produces a much clearer image with significantly less noise than the local thresholding. Yet, the local thresholding reveals more blood vessels that aren't shown when the global threshold method is applied.

Next, we applied the mask on the retinal image before implementing the global or local thresholding method to get a more direct view on the field, which makes is easier to detect the blood vessels and eliminate noise. The result of the local and global threshold applied on the field is shown below.

In these images you can see, again, there is less noise in the global thresholding method's results, but there are more blood vessels shown in the local threshold's method. In conclusion, we will continue to try to improve these results in the following week, and implementing new methods if necessary.

 
 
 

Recent Posts

See All
April Week 2 Updates

This week, we setup the workstation and GPU. So far, I’ve installed Windows 10 and device drivers, memory modules, graphics card, and the...

 
 
 
April 1st Week Summary

This week we are configuring the GPU we will be using to complete the remainder of our project. The GPU used in our project is Nvidia...

 
 
 
March 4th Week Summary

This week the team will complete the following task: - Assure everything is submitted and updated for the upcoming CRA-CREU milestone. -...

 
 
 

Address

Petersburg, VA, USA

Follow

  • Facebook
  • Twitter
  • LinkedIn

©2017 by An Automated Image Analysis Framework for Classification of Diabetic Retinopathy. Proudly created with Wix.com

bottom of page