Retinal tumor imaging and volume quantification in mouse model using spectral-domain optical coherence tomography

Area of Science:

• Ophthalmology
• Biomedical Imaging
• Machine Learning

Background:

• Retinal tumors, such as retinoblastoma, require accurate imaging and volume measurement for effective study and treatment monitoring.
• Current imaging techniques may lack the resolution or automated analysis capabilities needed for precise quantification in small animal models.

Purpose of the Study:

• To develop and validate an ultra-high resolution spectral-domain optical coherence tomography (SD-OCT) system for small animal retinal imaging.
• To create a novel, machine learning-driven segmentation algorithm for accurate retinal tumor boundary detection and volume calculation.

Main Methods:

• Utilized an ultra-high resolution SD-OCT system specifically designed for small animal retinal imaging.
• Developed a novel segmentation algorithm employing parametric deformable models (active contours) integrated with a machine learning-based Conditional Random Field for region classification.
• Enabled automated segmentation with optional user-guided boundary correction.

Main Results:

• Successfully imaged retinal tumors in a mouse model with ultra-high resolution.
• The novel segmentation algorithm achieved automatic tumor boundary detection and volume calculation.
• The system demonstrated quantitative capabilities for monitoring retinal tumor progression and treatment effects.

Conclusions:

• The developed SD-OCT system and machine learning algorithm provide a powerful tool for quantitative analysis of retinal tumors in mouse models.
• This technology facilitates advanced research into retinoblastoma progression and therapeutic interventions.