Comprehensive Summary
This study, done by Kaya et. al, used Self-AttentionNeXt, which is a deep learning model designed to detect schizophrenia. This was done using optical coherence tomography (OCT) images of the retina. OCT imaging allows for detailed measurement of the retinal nerve fiber layer (RNFL). The retinal changes that occur in patients with Schizophrenia correlate with disease duration and brain tissue loss. Using OCT to detect changes allows a window into the central nervous system abnormalities. The purpose of this study was to use Self-AttentionNeXt which is a convolutional neural network (CNN) model that integrates attention mechanisms, to classify OCT images and distinguish healthy patients from patients with schizophrenia. The results found that the model had 95.87% accuracy and also found that it had high sensitivity and specificity across the different disease categories. The study shows that retinal imaging could be used as a potential biomarker for schizophrenia with the help of AI.
Outcomes and Implications
This study shows how AI can be used as a powerful tool in diagnostic practices. The researchers found that Self-AttentionNeXt helped to bridge the gap between traditional CNNs and transformer models in medical imaging. The limitations of the AI used in this study is that it has not been validated for clinical use yet, and there was an age mismatch that occurred between the control and patient groups. Future directions for this technology could be to develop real-time clinical applications so that it could be used more widely. Also this model could be applied to other neurological disorders, and MRI and cognitive data could be combined with OCT to better diagnose conditions. Using AI to catch disorders, especially those that are degenerative, in their early stages allows for less aggressive treatments to be used.