Related Experiment Video
Updated: Feb 25, 2026
07:59
Efficient and Consistent Generation of Retinal Pigment Epithelium/Choroid Flatmounts from Human Eyes for Histological Analysis
Published on: October 28, 2022
3.5K
Improving graph-based OCT segmentation for severe pathology in Retinitis Pigmentosa patients
Andrew Lang1, Aaron Carass1,2, Ava K Bittner3
1Dept. of Electrical and Computer Engineering, The Johns Hopkins University, Baltimore, MD 21218, US.
Summary
This study enhances optical coherence tomography (OCT) segmentation for retinitis pigmentosa (RP) by adapting algorithms to account for retinal layer changes. The improved method achieves higher accuracy in segmenting OCT data from RP patients.
Area of Science:
- Ophthalmology
- Medical Imaging
- Computational Biology
Background:
- Macular optical coherence tomography (OCT) segmentation is crucial for diagnosing retinal diseases.
- Standard OCT segmentation algorithms perform poorly on retinitis pigmentosa (RP) due to photoreceptor layer absence.
Purpose of the Study:
- To enhance a graph-based OCT segmentation pipeline for accurate processing of RP patient data.
- To improve the segmentation accuracy of retinal layers in OCT scans of individuals with RP.
Main Methods:
- Adapted a graph-based OCT segmentation pipeline by relaxing layer thickness and smoothness constraints.
- Incorporated a random forest classifier for boundary probability estimation on RP data.
- Introduced an intensity normalization step to address intensity disparities in degenerated retinal layers.
Main Results:
- The enhanced algorithm successfully segmented eight retinal layers in RP data.
- Achieved an average overall boundary error of 4.22 μm, outperforming the original algorithm's 6.02 μm error.
- Demonstrated improved performance on OCT data from nine RP subjects via leave-one-out validation.
Conclusions:
- The developed enhancements enable robust OCT segmentation in retinitis pigmentosa patients.
- The modified pipeline offers a more accurate tool for analyzing retinal structures in RP.
- This advancement aids in the clinical assessment and research of retinitis pigmentosa.
