Probing the Influence of Specular Reflection and Overexposure on Backscattering Mueller Matrix Polarimetry for Tissue Imaging and Sensing

Abstract

Mueller matrix polarimetry has great potential for tissue detection and clinical diagnosis due to its ability to provide rich microstructural information accurately. However, in practical in vivo tissue imaging based on backscattering Mueller matrix polarimetry, specular reflection is often inevitable, leading to overexposed regions and the following inaccurate polarization information acquisition of tissues. In this study, we probe the influence of specular reflection and overexposure on backscattering Mueller matrix polarimetry for tissue imaging and sensing. We investigate in detail the differentiation of polarization behaviors between the specular reflection and non-specular reflection tissue regions using a 3 × 3 backscattering Mueller matrix measurement. Then, we obtain the vertical projection profiles to further quantify the Mueller matrix elements of porcine liver tissue in different specular reflection regions. Finally, we calculate the polarization feature parameters derived from a 3 × 3 Mueller matrix and analyze their behavior in overexposed regions. Based on the quantitative analysis and comparisons, we obtain a group of polarization feature parameters with strong immunity to the specular reflection process. This study offers a strategy for selecting the polarization parameters during in vivo polarimetric imaging applications, provides valuable references for further eliminating the characterization errors induced by specular reflection, and may contribute to the advancement of quantitative tissue polarimetric imaging and sensing.

Keywords: Mueller matrix; overexposure; polarimetry; tissue imaging and sensing.

Figure 1

Schematic of the experimental setup and sample: (a) backscattering MM imaging setup using a DoFP camera. L₁ and L₂, lenses; P, polarizer; M, screw linear motor. (b) Flowchart of PBPs acquisition based on 3 × 3 MMs. (c) Porcine liver tissue. The SR regions are inside the yellow dashed areas, and azimuth references to the coordinate system.

Figure 2

Grayscale images of porcine liver tissue sample with single acquisition: these sequenced grayscale images are generated by producing three polarization states for the incident light: horizontal linear (H), 45° linear (P), and vertical linear (V), and, respectively, detecting four polarization components of the emergent light: horizontal linear (H), 45° linear (P), vertical linear (V), and 135° linear (M). The first and second letters indicate the input and output polarization states, respectively. For instance, HP denotes horizontal linear input polarization and 45° linear output polarization.

Figure 3

Grayscale images and vertical projection profiles of MM elements of porcine liver tissue: (a) grayscale images of porcine liver tissue, including (a1) regions with NSR, and (a2) regions with SR. The rectangular regions are selected to calculate the vertical projection profiles. The numbers 1, 2, and 3 in (a2) represent the NSR regions, borderline overexposed regions, and completely overexposed regions, respectively. (b) Vertical projection profiles of the corresponding regions of the MM elements. The upper right subfigures show the pseudo-color MM image. All curves are normalized by the M11.

Figure 4

Pseudo-color PBP images of porcine liver tissue: (a) MMPD parameters images with NSR regions; (b) MMPD parameters images with SR regions; (c) MMT parameters images with NSR regions; (d) MMT parameters images with SR regions.

Figure 5

PBPs images and corresponding average change rates of porcine liver tissue: (a) The average change rate for 22 PBPs. ** indicates the variance less than 0.1, * indicates the variance between 0.1 and 1, and no mark indicates the variance larger than 1. The same column color represents a similar polarization property. The blue dashed lines denote the 0.5 and −0.5 scales. (b) Pseudo-color anisotropy-related PBPs images.