Patient-attentive sequential strategy for perimetry-based visual field acquisition.

Area of Science:

• Ophthalmology
• Neurology
• Medical Technology

Background:

• Perimetry is a standard clinical test for diagnosing eye and neurological conditions.
• Current perimetry methods are time-consuming (7-8 minutes per eye), leading to patient fatigue and reduced data quality.

Purpose of the Study:

• To introduce a novel perimetry testing strategy, PASS (Perimetry Assessment by Sequential Strategy), that accelerates visual field estimation.
• To improve the accuracy and efficiency of 2D visual field reconstruction in clinical settings.

Main Methods:

• Developed PASS, a reinforcement learning-based strategy for perimetry.
• PASS employs an adaptive selection policy to identify optimal locations for visual stimuli presentation.
• The strategy reconstructs 2D visual fields from sparse observations, prioritizing patient-specific data and maximizing reconstruction quality.

Main Results:

• PASS significantly reduces examination time by 30%–70% compared to state-of-the-art methods.
• The strategy achieves more accurate 2D visual field reconstructions.
• PASS demonstrates patient-specific and non-greedy adaptive testing.

Conclusions:

• PASS offers a more efficient and accurate approach to visual field testing.
• The reinforcement learning strategy optimizes perimetry by adaptively selecting test locations.
• This innovation has the potential to improve patient experience and diagnostic capabilities in ophthalmology and neurology.