”Through ophthalmic image collection and intelligent analysis of images, the artificial intelligence-based screening model developed by the team has an accuracy rate of 75% for Alzheimer’s disease screening in multiple communities.” February 6, Ningbo, Chinese Academy of Sciences Zhao Yitian, a researcher from the Intelligent Medical Imaging (iMED) team of the Institute of Materials, told the reporter of Science and Technology Daily that through in-depth analysis and mining of the relationship between eye structural changes and neurodegenerative diseases, an early detection scheme for neurodegenerative diseases can potentially be formed.
The onset of neurodegenerative diseases is long, difficult to be noticed on a daily basis, and often irreversible, with long-term effects on human health. At this stage, the diagnosis of such diseases requires the use of expensive detection methods such as magnetic resonance imaging, or through cognitive function scales, genetic testing, and spinal taps to obtain cerebrospinal fluid. However, related methods have vague indications, or are accompanied by trauma and radioactive defects, and are not suitable for large-scale screening of grassroots populations.
In order to explore the relationship between fundus retinal structural changes and Alzheimer’s disease, the iMED team cooperated with West China Hospital of Sichuan University, Zhejiang Provincial People’s Hospital, Peking University Third Hospital, Ningbo University Affiliated People’s Hospital and other medical institutions to collect A large amount of eye and brain data of Alzheimer’s patients, and the fundus images of optical coherence tomography (OCTA) as the main analysis object.
According to the iMED team, optical tomography is an advanced non-invasive imaging technology that can display structures at different depths of the fundus, including the retina and choroid, and can also scan blood flow changes in the fundus structure with high precision to generate OCTA images. It is of great significance to study the changes of fundus blood vessels caused by Alzheimer’s disease.
Through the self-developed intelligent analysis algorithm, the team automatically quantified the fundus structure of Alzheimer’s patients, and performed a cross-sectional statistical analysis of the calculated biological indicators and clinical data. The analysis showed that a variety of quantitative indicators were significantly correlated with the onset of Alzheimer’s disease, including vascular density, vascular fractal dimension, and vascular curvature. This result is consistent with the clinical prior consensus.
Based on this, the team designed an advanced AI model for Alzheimer’s disease detection based on blood flow imaging image information. After only inputting ophthalmic images into the AI model, it can quickly determine whether the subject has Alzheimer’s disease or not. Alzheimer’s disease.
In addition, the team also carried out ophthalmic image analysis and establishment of intelligent diagnosis models for brain diseases such as stroke and Parkinson’s disease. The results found that some ocular biomarkers were statistically correlated with the onset of disease. The rapid and portable screening provides a new idea.
It is reported that the team is currently relying on multi-centers to carry out large-scale population follow-up surveys, collecting sequence data with clinical research significance, and further analyzing the relationship between fundus structural changes and the pathogenesis of related brain diseases.