Robust domain selection for functional data via interval-wise testing and effect size mapping.

Area of Science:

• Statistics
• Functional Data Analysis
• Biomedical Signal Processing

Background:

• Domain selection in functional data analysis identifies key sub-intervals with specific features.
• Quantitative ultrasound (QUS) signal analysis requires methods to pinpoint functional differences across groups.

Purpose of the Study:

• To propose a robust domain selection method for identifying sub-intervals with distinct location parameter behaviors among different groups.
• To enhance QUS signal analysis by detecting practically interpretable domains and handling data imperfections.

Main Methods:

• Extending the interval testing approach to consider multiple functional features simultaneously.
• Utilizing functional M-estimators for robust inference, addressing outliers and missing data.
• Developing an effect size heatmap for visualizing dynamic group behaviors across the domain.

Main Results:

• The proposed robust domain selection method effectively identifies relevant sub-intervals.
• Functional M-estimators provide robust inference in the presence of outliers and missing data.
• Effect size heatmaps offer comprehensive insights into dynamic functional behaviors for clinical interpretation.

Conclusions:

• The novel robust domain selection method is effective for identifying functionally distinct sub-intervals.
• The method, particularly useful for QUS analysis, improves the understanding of group-specific behaviors.
• The approach facilitates the selection of practically meaningful domains for clinical decision-making.