Fully Automated Segmentation of Fluid/Cyst Regions in Optical Coherence Tomography Images With Diabetic Macular Edema...

Area of Science:

• Medical Imaging
• Computer Vision
• Ophthalmology

Background:

• Diabetic macular edema (DME) causes vision loss.
• Accurate segmentation of fluid and cysts in OCT images is crucial for DME assessment.
• Existing segmentation methods have limitations in accuracy and automation.

Purpose of the Study:

• To develop a fully automated algorithm for segmenting fluid-associated and cyst regions in OCT retina images.
• To improve the accuracy and efficiency of DME assessment through advanced image processing techniques.

Main Methods:

• A novel neutrosophic transformation is applied to OCT images.
• A graph-based shortest path method is used for segmentation.
• A new cost function and automated cluster number estimation are incorporated for fluid/cyst segmentation.

Main Results:

• The algorithm achieves improved performance on Duke and Optima datasets compared to previous methods.
• Dice coefficient improvements of 8% (Duke) and 6% (Optima) were observed.
• Precision improved by 5% (Duke) and 23% (Optima), with sensitivity gains of 22% (Optima).

Conclusions:

• The proposed automated algorithm effectively segments fluid and cyst regions in OCT images for DME.
• The novel neutrosophic and graph-based approach offers superior performance over existing methods.
• This algorithm has the potential to enhance clinical assessment and management of diabetic macular edema.