Changes in Visual Attention Patterns for Detection Tasks due to Dependencies on Signal and Background Spatial...

Area of Science:

• Medical Imaging
• Radiology
• Human Visual Perception

Background:

• Radiologists' effectiveness in medical image analysis is high, but misdiagnosis remains a challenge.
• Digital breast tomosynthesis (DBT) images present complex backgrounds with varying breast densities and structures.
• Understanding visual attention mechanisms is crucial for improving diagnostic accuracy in medical imaging.

Purpose of the Study:

• To investigate how image and signal properties influence visual attention during signal detection in digital images.
• To explore the relationship between target morphology, background complexity, and detection performance.
• To identify factors contributing to diagnostic errors in medical image interpretation.

Main Methods:

• Utilized simulated tomographic breast images generated from digital breast phantoms (Bakic and XCAT).
• Introduced two lesion types (sphere and spicule) with distinct spatial frequencies into simulated images.
• Conducted an observer study with six human participants, collecting eye-gaze data during lesion detection tasks in DBT slices.

Main Results:

• Detection performance is significantly constrained by later perceptual stages, with decision failures being a major error source.
• Signal detectability is influenced by a combination of target morphology and background complexity, highlighting signal-background interactions.
• Longer fixation durations on spiculated lesions indicate differential engagement of visual attention based on spatial frequency dependencies.

Conclusions:

• Visual attention mechanisms are critical in signal detection tasks within complex imaging environments.
• Both local signal features and global anatomical noise interact to affect signal detectability.
• Findings have implications for improving digital imaging systems and training for enhanced diagnostic accuracy.