Diabetic Retinopathy can be effectively screened with Aurora fundus camera

Italy: A new research from Italy found that the Aurora fundus camera can be an effective tool for first-line screening of Diabetic Retinopathy. The in-built AI software can be regarded as a trustworthy instrument for automatically detecting the existence of indications of DR and is therefore used as a potential resource in massive screening programs. The study results were published in the journal Acta Diabetologica. 

Diabetic retinopathy, a microvascular complication of diabetes, is the main cause of vision loss and preventable blindness among working-age adults. With increasing cases of Diabetic retinopathy, there is a prime necessity for proper screening. Hence utilizing artificial intelligence and novel technologies handheld portable devices for retinal imaging were developed with good image quality and diagnostic accuracy compared to the table-top fundus cameras. One such device is the Optomed Aurora IQ camera where an automated image analysis can diagnose retinopathy at early stages. Hence researchers conducted a study to validate the in-built AI deep learning algorithm Selena+ of the handheld Aurora fundus camera (Optomed, Oulu, Finland) for a first-line screening in a real-life setting. 

An observational cross-sectional study was carried out on 256 eyes of 256 consecutive patients including both diabetic and non-diabetic patients. Each patient received a 50°, macula-centered, non-mydriatic fundus photography and, after pupil dilation, a complete fundus examination by an experienced retina specialist. All images were analyzed by a skilled operator and by the AI algorithm. Later all the results of the three procedures were compared. 

Key findings:  

• There was a 100% agreement between the operator-based fundus analysis in bio-microscopy and the fundus photographs. 
• In nearly 121 out of 125 subjects (96.8%) of the DR patients the AI algorithm could identify signs of DR and in 122 of the 126 non-diabetic patients (96.8%) it could identify no signs of DR. 
• AI algorithm showed a sensitivity of 96.8% and a specificity of 96.8%.
• The overall concordance coefficient k (95% CI) between AI-based assessment and fundus biomicroscopy was 0.935 (0.891–0.979).

Thus, the study showed an almost perfect agreement between the standard ophthalmoscopic examination and the AI algorithm sensitivity of 96.8% and a specificity of 96.8% and can be an effective first-line DR screening program without the use of mydriatic eyedrops.

Further reading: Lupidi, M., Danieli, L., Fruttini, D. et al. Artificial intelligence in diabetic retinopathy screening: clinical assessment using handheld fundus camera in a real-life setting. Acta Diabetol (2023). https://doi.org/10.1007/s00592-023-02104-0