Systematic validation of time-resolved diffuse optical simulators via non-contact spad-based measurements

Area of Science:

• Biomedical Optics
• Medical Imaging
• Computational Modeling

Background:

• Accurate forward models are crucial for time-domain diffuse optical imaging (TD-DOI) but existing simulators lack validation, especially for non-contact setups.
• Experimental validation of TD-DOI simulators is needed for reliable photon propagation modeling in scattering media.

Purpose of the Study:

• To rigorously evaluate three open-source TD-DOI simulators: MMC, NIRFASTer, and Toast++.
• To compare simulation accuracy, computational speed, and applicability for non-contact TD-DOI using time-resolved experimental data.

Main Methods:

• Compared MMC (Monte Carlo), NIRFASTer, and Toast++ (FEM) using a unified mesh and point-source illumination.
• Applied virtual source approximation for FEM solvers in non-contact oblique illumination scenarios.
• Acquired time-resolved transmission data using a SPAD array from BIP-certified phantoms and quantified simulation accuracy and speed.

Main Results:

• MMC demonstrated superior spatial-domain and time-domain accuracy.
• NIRFASTer and Toast++ showed comparable performance, offering faster computation for applications like image reconstruction.
• Virtual source approximation significantly reduced errors (>34%) in non-contact FEM modeling.

Conclusions:

• This study provides a framework for selecting TD-DOI simulators based on accuracy and speed requirements.
• The validation against SPAD array data under non-contact conditions bridges a gap in biomedical optical simulation standards.
• Facilitates faster prototyping of novel TD-DOI systems by optimizing solver selection and configuration.