Quantitative photoacoustic tomography with light fluence compensation based on radiance Monte Carlo model

Abstract

Objective: Photoacoustic tomography (PAT) is a rapidly evolving imaging modality that provides images with high contrast and spatial resolution showing the optical properties of biological tissues. The photoacoustic pressure is proportional to the product of the optical absorption coefficient and the local light fluence. The essential challenge in reconstructing quantitative images representing spatially varying absorption coefficients is the unknown light fluence. In addition, optical attenuation induces spatial variations in the light fluence, and the heterogeneity of the fluence determines the limits of reconstruction quality and depth. Approach: In this work, a reconstruction enhancement scheme is proposed to compensate for the variation of the light fluence in the absorption coefficient recovery. The inverse problem of the radiance Monte Carlo model describing light transport through the tissue is solved by using an alternating optimization strategy. In the iteration, the absorption coefficients and photon weights are alternately updated. Main results: The method provides highly accurate quantitative images of absorption coefficients in simulations, phantoms, and in vivo studies. The results show that the method has great potential for improving the accuracy of absorption coefficient recovery compared to conventional reconstruction methods that ignore light fluence variations. Comparison with state-of-the-art fluence compensation methods shows significant improvements in root mean square error, normalized mean square absolute distance, and structural similarity metrics. Significance: This method achieves high precision quantitative imaging by compensating for non-uniform light fluence without increasing the complexity and operation of the imaging system.

Quantitative Photoacoustic Tomography Using Iteratively Refined Wavefield Reconstruction Inversion: A Simulation Study

“…In fact, some older qPAT methods eschew fluence modeling entirely, and assume that the fluence is uniform throughout the tissue [7,27,29]. Moreover, recently, several research studies have made use of minimization-based methods [38][39][40][101][102][103]. Nonetheless, the primary drawback associated with these newer approaches is their tendency to exhibit escalating computational complexity, ultimately resulting in substantial computation time requirements [104].…”

“…In qPAT, the optical inverse problem needs to be addressed [33,34]. This problem is typically tackled using iterative approaches based on mathematical models, as described in [35][36][37][38][39][40][41]. The model employed in these approaches is based on the radiative transfer equation (RTE), which accurately simulates the propagation of light to calculate optical fluence [37,42,43].…”

Extractor-attention-predictor network for quantitative photoacoustic tomography

“…Evaluation metrics in previous studies may inadequately reflect the overall accuracy of predicted images in some cases. Per-pixel metrics, such as mean absolute error and mean relative error (MRE) [32] , [54] , are inclined to reflect the error levels within those large tissues. The peak signal-to-noise ratio (PSNR), directly computed from the maximum , may result in a biased value when there is a significant disparity in between different tissues [55] .…”

Monte Carlo-Based Optical Simulation of Optical Distribution in Deep Brain Tissues Using Sixteen Optical Sources

“…where E N represents the normalized E (F or A), Ex,y,z named the E at the position (x, y, z), E max donates the maximum value of the E, and E min means the minimum value of the E. [23] Surface…”