Detecting Glaucoma using Self Organizing Maps
This was my project under Dr.K.M.M Rao while in BITS-Pilani
image_path = input image path
image_mask_path = input image mask path
output_save_path = out image path
disk_mask_path = disk mask image generated for running Self Organizing Maps
Glaucoma.exe T <image_path> <image_mask_path> <output_save_path> <disk_mask_path>
Trained data will be saved in som_data.txt
set the threshold for SOM in function somThresh()
cv::Mat somThresh(cv::Mat &inp,cv::Mat &mask){
...
...
if ((active>=<min_threshold>)&&(active<=max_threshold))
{
...
...
Glaucoma.exe O <image_path> <image_mask_path> <output_save_path> <disk_mask_path>
Create image of same size as input image and set Region of interest (ROI) pixels to '255' and rest to '0'.
Left fundus image, Right mask with ROI set to '255'
Program will out a single number which is cup-to-disk ratio. This ratio can be used to determine glaucoma. Ratio of 0.8 is a good indicator of glaucoma.
Green channel -> Gaussian matched filter -> OTSU filter -> Morphological closing -> Sobel filter & Thresholding -> Hough circles -> Self-organising maps
Optic disk and cup marked.
This project is licensed under the BSD 2-clause "Simplified" License - see the LICENSE file for details
https://www.academia.edu/35315442/A_Project_Report_On_Biomedical_Imaging_for_Eye_Care