| Title: | Development and Validation of a Deep Learning Algorithm for Classification of Psychotic Spectrum Disorders from Retinal Imaging |
| Lead Institution: | Northeastern University (USA) |
| Principal investigator: | Mr Babatunde Aideyan |
Application 75692
WARNING: the interactive features of this website use CSS3, which your browser does not support. To use the full features of this website, please update your browser.
About
The proposed study is a first of its kind investigation that aims to teach a computer to classify individuals with diagnosed psychotic disorders from a healthy group. The computer will learn to evaluate images of the eye, which contain structures associated with a range of diseases. At this time, there are no published studies in this area, though three primary trends in recent research provide background evidence for this investigation: 1) computers are capable of classifying medical images of individuals with or without various diseases; 2) when computers are taught to examine images of the eye, they have been accurate in categorizing certain retinal diseases; 3) though specific patterns are yet to be defined, studies have discovered a number of biological signals in the retina associated with psychotic disorders. Thus, the objective is to demonstrate that a computer can be trained to accurately classify a patient's psychotic disorder diagnosis from an image of their eye. A group of research participants from the UK Biobank who had previously received a diagnosis of any psychotic disorder will be the clinical group, while a subset of participants without any known psychiatric illness will be the healthy group. A computer will be trained to distinguish images of the eye of the clinical group from the healthy group using a specific computational framework that has driven computer vision research in artificial intelligence. Said another way, this study's primary aim is to teach a computer to "see" the differences in images of the eye between individuals with psychotic disorders and those without. While discovery of changes in the eye that indicate various psychiatric illnesses is still a growing line of research, it is expected that the powerful pattern recognition capabilities of modern machine learning technologies will encode trends that have previously eluded researchers. Results from this research may lay the groundwork for technologies that augment diagnosis of psychiatric disorders. If propelled to a validated clinical technology, research in this area may improve the clinical decision-making of researchers and providers supporting psychiatric patients. In 2018, a device that automatically diagnoses diabetic retinopathy from retinal imaging became the first-ever United States Food and Drug Administration (FDA) approved autonomous medical system. Devices like this, as well as forthcoming technologies buttressed by this research, are thought to produce a "synergistic effect" in such a way that clinicians and artificial intelligence (AI) working together produce "better results than either alone."1 Publication
| Pub ID | Title | Author(s) | Year | Journal |
|---|---|---|---|---|
| 18917 | Retinal cytoarchitectural alterations across the psychosis spectrum and their correlates with cognition: a UK biobank nested case-control study | Erik Velez-Perez (+5) | 2025 | Molecular Psychiatry |
Enabling scientific discoveries that improve human health