Early pigment spot segmentation and classification from iris cellular image analysis with explainable deep learning...

Area of Science:

• Ophthalmology
• Computer Science
• Medical Imaging

Background:

• Retinal disorders affect thousands globally, necessitating early diagnosis to prevent blindness.
• Iris spot segmentation is crucial for analyzing iris cellular images, which are often degraded by noise and off-angles.
• Current segmentation methods struggle with noisy, uncooperative iris images and lack efficiency.

Purpose of the Study:

• To develop an explainable deep learning model for precise iris pigment spot segmentation and classification.
• To improve early detection of eye syndromes by analyzing iris cellular images.
• To overcome limitations of traditional, costly, and time-consuming diagnosis processes.

Main Methods:

• An explainable deep learning model integrated with a multiclass support vector machine was developed.
• The model was trained and tested on three benchmark datasets: MILE, UPOL, and Eyes SUB.
• Iris cellular images were analyzed for early pigment spot segmentation and classification.

Main Results:

• The proposed model demonstrated superior performance compared to existing methods in terms of classification errors.
• The model effectively located micro-pigment spots on iris surfaces.
• Experimental results on benchmark datasets confirmed the model's accuracy and effectiveness.

Conclusions:

• The developed explainable deep learning model offers a promising approach for early iris pigment spot analysis.
• This method can aid in the early diagnosis of retinal disorders, potentially preventing blindness.
• The model's effectiveness in segmenting and classifying iris pigment spots enhances diagnostic capabilities.