Automated Abnormality Detection in Patient Retinal Function: A Deep Learning-Powered Electroretinogram Analysis...

Area of Science:

• Ophthalmology
• Medical Imaging
• Artificial Intelligence

Background:

• The electroretinogram (ERG) is crucial for evaluating retinal diseases.
• Manual ERG analysis is time-consuming and requires expert interpretation.
• Objective and efficient diagnostic tools are needed for retinal function assessment.

Purpose of the Study:

• To develop and validate a deep learning-based system for automated ERG analysis.
• To detect abnormalities in retinal function using ERG tracings.
• To assist electrophysiologists and ophthalmologists with rapid diagnostic support.

Main Methods:

• A dataset of 5,640 ERG tracings from 470 patients was compiled.
• Diagnoses were provided by experienced electrophysiologists for training and validation.
• A novel deep learning model was implemented for automated ERG waveform analysis.

Main Results:

• The deep learning system achieved an F1-score of 77.59% in detecting retinal function anomalies.
• The system demonstrated a rapid inference time of 1.2 milliseconds.
• The automated analysis proved effective for full-field ERGs.

Conclusions:

• The proposed deep learning system offers an effective solution for automated ERG analysis.
• This tool can significantly reduce diagnostic time and improve efficiency for retinal specialists.
• The system provides reliable automated assessments to aid in diagnosing retinal diseases.