Application of Ideal Observer for Thresholded Data in Search Task

TL;DR

This work develops an anthropomorphic thresholded visual-search model observer for location-unknown lesion detection in medical imaging. By applying thresholding to data within a two-stage search-and-decision framework and using stage-specific feature sets, the approach enhances discrimination while reducing training data requirements. Key findings show thresholding improves AUC, stabilizes estimates with fewer images, and that distinct feature sets for candidate selection and decision-making yield closer alignment with human performance across imaging conditions. The method offers a practical pathway for training-efficient observers applicable to clinically realistic search tasks and potentially extendable to related domains such as computer vision and security imaging.

Abstract

This study advances task-based image quality assessment by developing an anthropomorphic thresholded visual-search model observer. The model is an ideal observer for thresholded data inspired by the human visual system, allowing selective processing of high-salience features to improve discrimination performance. By filtering out irrelevant variability, the model enhances diagnostic accuracy and computational efficiency. The observer employs a two-stage framework: candidate selection and decision-making. Using thresholded data during candidate selection refines regions of interest, while stage-specific feature processing optimizes performance. Simulations were conducted to evaluate the effects of thresholding on feature maps, candidate localization, and multi-feature scenarios. Results demonstrate that thresholding improves observer performance by excluding low-salience features, particularly in noisy environments. Intermediate thresholds often outperform no thresholding, indicating that retaining only relevant features is more effective than keeping all features. Additionally, the model demonstrates effective training with fewer images while maintaining alignment with human performance. These findings suggest that the proposed novel framework can predict human visual search performance in clinically realistic tasks and provide solutions for model observer training with limited resources. Our novel approach has applications in other areas where human visual search and detection tasks are modeled such as in computer vision, machine learning, defense and security image analysis.

Figures (6)

• Figure 1: Performance comparison of two features. The red plot is the combination of two features with different SNRs; the green plot is the combination of two features with same SNRs.
• Figure 2: Performance of four observers in search task with lumpy-background images.
• Figure 3: Performance comparison for the stage-specific features model (green line) and visual search observers (black line).
• Figure 4: Perfomance comparison between prewhitening VS model, nonprewhitening VS model, thresholded nonprewhitening VS model and human.
• Figure 5: Performance uncertainty comparison between the VS model, the thresholded VS model, and the human.