Identify Age-Related Macular Degeneration with AI Deep Learning Model

Identify Age-Related Macular Degeneration with AI Deep Learning Model

A survey aimed to examine which areas might be associated with age-related macular degeneration (AMD) with point-of-interest (POI) visualization of a deep learning model. The study used OptiNet, a deep learning model that was trained to identify the presence of AMD in spectral domain optical coherence tomography (SD-OCT) macular scans.

SD-OCT has become a cornerstone of assessment in ophthalmology, especially when it comes to diagnosing AMD. As a result, the amount of collected images keeps increasing. Arnt-Ole Tvenning, Department of Ophthalmology, St. Olav’s Hospital, University Hospital Trondheim, and investigators sought to tap into the wealth of information these images have the potential to provide.

The OptiNet deep learning artificial intelligence (AI) tool was used because its artificial neural networks were structured with multiple layers of neurons to understand complex tasks in the same way as the brain.

“Deep learning models learn without a priori knowledge, their own characteristics of a given disease, and these characteristics can then be used to classify medical images such as SD-OCT as pathological or normal,” the investigators wrote. .

Deep Learning AI Points of Interest

OptiNet was trained and validated on 2 datasets for this research. The first set included a single analysis of 269 AMD cases and 115 controls. The second set included 337 scans performed on 40 AMD cases (62 eyes) and 46 scans of both eyes of the 23 control cases.

Visualization of hotspots was achieved by calculating feature dependencies across the hierarchy of layers in the deep learning architecture.

Patients were 50 or older in the first dataset and had intermediate AMD. The controls were age-matched and showed no signs of AMD in either eye. The second set of data was from patients in the same age group who had exams performed every 6 months.

The deep learning model was trained to extract features based on specific information from the scan to create data classification from the scans. Each convolution layer learned to identify relevant features, such as lines and edges, as well as the probability of AMD being present.

Classification of data from SD-OCT images

Investigators identified the retinal nerve fibers (82%) and choroid layers (70%) as points of interest in cases classified as AMD. The most frequently applied areas were the retinal pigment epithelium (98%) and the drusen (97%).

OptiNet obtained an area under the receiver operator curves ≥99.7%.

Results from the deep learning model indicated alterations in SD-OCT imaging regions attributed to retinal nerve fibers and choroid layers. According to the researchers, further investigation into the role of the neuroretina and choroid in the development of AMD is needed to better understand whether the findings represent a change in macular tissue with AMD.

“For the first time, the anatomical areas used by a deep learning model to identify AMD are presented with a new visualization method. This method revealed that regions such as the RNFL and the choroid could be modified in the AMD and demonstrate the potential of deep learning as a method not only for the identification but also for the exploration of retinal diseases,” they wrote. “As a result, the methods applied in this study can be extended to include several retinal diseases.”

The study “Deep learning identifies retinal nerve fibers and choroidal layers as markers of age-related macular degeneration in classification of optical coherence tomography volumes in the macular spectral domain” was published in Acta Ophthalmologica.

#Identify #AgeRelated #Macular #Degeneration #Deep #Learning #Model

Leave a Comment

Your email address will not be published. Required fields are marked *