Targeted spatial sampling using GOANNA improves detection of visual field progression.

Area of Science:

• Ophthalmology
• Medical Technology
• Computational Science

Background:

• Automated visual field testing is crucial for diagnosing and monitoring glaucoma.
• Current methods may lack precision in detecting subtle changes, particularly around scotoma edges.
• The Gradient-Oriented Automated Natural Neighbour Approach (GOANNA) was developed to enhance spatial resolution in visual field testing.

Purpose of the Study:

• To evaluate if the enhanced accuracy and precision of GOANNA in sampling scotoma edges translate to more accurate classification of glaucomatous progression.
• To compare the performance of GOANNA against existing methods like ZEST in simulated glaucoma progression scenarios.

Main Methods:

• Computer simulations were conducted using six procedures, including three variants of GOANNA and two of ZEST.
• Simulations involved 156 sequences of progressing and 156 of stable visual fields under reliable and typical false positive error conditions.
• Performance was assessed using sensitivity, specificity, and Area Under the ROC Curve (AUC) plotted against the number of visual fields acquired.

Main Results:

• GOANNA demonstrated equal or greater AUC than ZEST across all visits, especially when baseline fields were defective or healthy with no false responses.
• Retest versions of GOANNA showed improvement over the original GOANNA after seven visits for initially healthy fields.
• The enhanced precision of GOANNA around scotoma borders was evident in the simulation results.

Conclusions:

• The GOANNA framework facilitates earlier and more accurate detection of glaucomatous progression compared to ZEST.
• Improved precision and accuracy around scotoma borders are key advantages of GOANNA for identifying disease progression.
• GOANNA shows particular promise in the early stages of glaucoma, enhancing diagnostic capabilities.