AI image analysis tools quantify schisis cystic volume in XLRS retinal dysmorphology

Area of Science:

• Ophthalmology
• Medical Imaging
• Artificial Intelligence

Background:

• X-linked retinoschisis (XLRS) is a genetic disorder affecting the retina.
• Accurate quantification of retinoschisis cystic cavities is crucial for monitoring disease progression and evaluating treatment efficacy.
• Manual segmentation of these cavities from Optical Coherence Tomography (OCT) images is time-consuming and subject to inter-observer variability.

Purpose of the Study:

• To assess the feasibility and utility of artificial intelligence (AI)-based deep learning algorithms for automating the quantification of retinoschisis cystic cavity volume in XLRS patients.
• To compare the performance of AI-driven image analysis with traditional human manual segmentation.
• To evaluate the potential of AI-quantified metrics as endpoints in clinical therapeutic trials for XLRS.

Main Methods:

• Review of two studies utilizing AI-based deep learning algorithms for analyzing OCT retinal images in XLRS patients.
• Implementation of AI-analytics for automated segmentation and quantification of retinal schisis cavities.
• Comparative analysis of AI-derived measurements against human manual segmentation of the same OCT images.
• Simulation of a clinical therapy trial to assess AI-quantified schisis volume (ASV) as a structural endpoint compared to central subfield thickness (CST) and central foveal thickness (CFT).

Main Results:

• Both studies successfully developed independent AI algorithms capable of automating and quantifying retinoschisis cavity spaces.
• AI-based analysis demonstrated performance comparable to or exceeding human segmentation accuracy for XLRS structural dysmorphology.
• In a simulated clinical trial, AI-quantified schisis volume (ASV) emerged as a more sensitive structural endpoint than CST or CFT for evaluating therapeutic interventions.

Conclusions:

• Automating the quantification of retinoschisis cavity volume in XLRS patients using AI is feasible.
• AI-based cavity volume measurement presents a viable and potentially improved outcome measure for future XLRS therapeutic trials.
• These findings support the integration of AI in the clinical management and research of X-linked retinoschisis.