Intra-Retinal Layers Segmentation
Posted:
Project Aims
- Develop a computer-aided self-diagnostic system to support ophthalmologists in mass screening for retinal abnormalities, facilitating faster and more accurate diagnoses
- Combining CNN with structure tensor aims to classify retinal layer patches using AlexNet through transfer learning, enhancing segmentation accuracy.
- The project seeks to find retinal layer boundaries by applying Gabor filters and introducing a flattening technique to improve boundary detection across curved retinal surfaces.
- Test and validate the proposed methods on OCT images from the Armed Forces Institute of Ophthalmology (AFIO) and Duke University’s publicly available datasets, ensuring reliability in real-world applications.
