Summary:
UCLA researchers in the Department of Computational Medicine have developed a computer program capable of automatically and accurately diagnosing optical diseases using OCT.
Background:
Optical diseases, such as age-related macular degeneration (AMD), are serious issues that, if left untreated, can result in partial and even complete blindness. Current diagnosis methods are time consuming and expensive as specially trained physicians are needed to perform and analyze the eye exam results. This leads to both longer diagnosis times and incomplete diagnoses as there may have been unexamined data that physicians did not have time to examine. There is a need for an automated method that can accurately review images for clinically-relevant biomarkers and determine a diagnosis to reduce costs and improve patient care.
Innovation:
Researchers at UCLA have developed a computer program that can diagnose ocular diseases using information gathered from optical coherence tomography (OCT). The program has been successfully tested to make an accurate diagnosis, requiring fewer images compared to a trained physician. The system leverages deep learning to capture three-dimension (3D) information about two-dimension images, allowing the system to identify irregularities in OCT and MRI images. By automating the process of biomarker identification, this program reduces workload of medical professions reviewing images and allows them to review images on an as-needed, rather than required, basis. In addition, because human input is not involved, this optimizes the clinical workflow by reducing the workload on physicians and providing additional protection from misdiagnosis.
Patent:
Biomarker Prediction Using Optical Coherence Tomography (US20230045859A1)
Potential Applications:
- Diagnosis of ocular diseases such as age-related macular degeneration (AMD)
- Disease progression monitoring
- Potential use with other imaging devices like MRIs
Advantages:
- Very accurate with less data
- Can be applied for multiple diagnosis
- Helps to optimize clinical workflow
- Automated
Development to Date:
The program has successful been able to determine diagnoses based on provided images.