Related Experiment Video
Updated: Aug 15, 2025
08:22
Application of Optical Coherence Tomography to a Mouse Model of Retinopathy
Published on: January 12, 2022
4.2K
Cyst identification in retinal optical coherence tomography images using hidden Markov model
Niloofarsadat Mousavi1, Maryam Monemian2, Parisa Ghaderi Daneshmand2
1Department of Electrical and Computer Engineering, Isfahan University of Technology, Isfahan, Iran.
Scientific Reports
|January 2, 2023
Summary
This study introduces a novel method using Hidden Markov Models (HMM) for rapid detection of retinal cysts in Optical Coherence Tomography (OCT) B-scans. The approach significantly improves accuracy in identifying these important indicators of retinal disease.
Area of Science:
- Ophthalmology
- Medical Imaging
- Computer Vision
Background:
- Retinal cysts are significant indicators of various retinal diseases.
- Accurate and rapid detection of these cysts is crucial for diagnosis and treatment.
- Optical Coherence Tomography (OCT) is a key imaging technique for visualizing retinal layers.
Purpose of the Study:
- To develop a new, rapid method for detecting cystic B-scans in OCT images.
- To model the presence of cysts using a Hidden Markov Model (HMM).
- To enhance the accuracy of cyst detection in retinal OCT scans.
Main Methods:
- Utilized a Hidden Markov Model (HMM) to represent cyst presence as a hidden state.
- Extracted various image features including Harris, KAZE, HOG, SURF, FAST, Min-Eigen, and deep AlexNet features.
- Employed AlexNet features as observation vectors for HMM parameter estimation.
Main Results:
- Identified deep AlexNet features as having the highest discriminating power for cyst detection.
- Demonstrated improved performance and accuracy of the HMM-based method for detecting cystic OCT B-scans.
- The HMM approach effectively models the sequential nature of cyst presence across B-scans.
Conclusions:
- The proposed HMM-based method offers a robust and accurate approach for rapid cyst detection in OCT B-scans.
- Deep learning features, specifically from AlexNet, are highly effective for this detection task.
- This technique holds promise for improving the diagnostic capabilities in retinal imaging.
