Spatial Linear Mixed Effects Modelling for OCT Images: SLME Model

Area of Science:

• Ophthalmology
• Medical Imaging
• Biostatistics

Background:

• Accurate and efficient disease detection, particularly with limited data, is a key research focus.
• Explanatory models are gaining traction for their ability to elucidate data generation processes.
• Diabetic maculopathy significantly impacts retinal thickness, necessitating advanced analytical tools.

Purpose of the Study:

• To develop and validate a spatial explanatory modeling approach for analyzing Optical Coherence Tomography (OCT) retinal imaging data.
• To improve the accuracy and efficiency of disease detection in smaller datasets.
• To understand retinal thickness changes in diabetic maculopathy for better treatment planning.

Main Methods:

• A spatial linear mixed effects (SLME) inference framework was developed.
• The model incorporates spatial topography, mixed effects, and spatial error structures to analyze retinal thickness maps.
• Heidelberg OCT retinal thickness data from 300 diabetes patients and 50 controls were analyzed.

Main Results:

• The proposed SLME model demonstrated superior performance compared to traditional analysis-of-variance methods.
• The SLME model exhibited higher power in detecting differences between disease groups.
• The model effectively identified specific regions where retinal thickness profiles differ between diabetic and healthy eyes.

Conclusions:

• The spatial explanatory modeling approach using SLME enhances the accuracy of statistical inferences in retinal imaging analysis.
• This method is crucial for understanding diabetic maculopathy progression and aids in early treatment planning.
• The model's framework is adaptable for monitoring other diseases and utilizing different imaging modalities.