Path Analysis of the Cancer Mueller Matrix: From Physical Decomposition to High-Dimensional Vector Mapping

Area of Science:

• Biomedical Optics
• Pathology
• Medical Imaging

Background:

• Conventional diagnostic methods struggle with subtle differences in unstained tissue sections.
• Mueller matrix polarimetry (MMP) offers rich polarization data but faces interpretation challenges due to coupled optical effects.

Purpose of the Study:

• To compare low-dimensional physical decomposition and high-dimensional mapping for cancer region identification.
• To develop an enhanced method for differentiating cancerous from normal tissues using MMP data.

Main Methods:

• Established a pixel-level polarization dataset from lung cancer and basal cell carcinoma sections.
• Developed a vectorial metric norm spectrum incorporating multi-order features.
• Implemented a high-dimensional mapping framework for polarization characteristic representation.

Main Results:

• Conventional decomposition methods showed limited discriminative capability and class overlap.
• The proposed high-dimensional mapping framework achieved accurate differentiation between cancerous and normal regions.
• The method demonstrated strong performance in cross-validated evaluations.

Conclusions:

• A progressive strategy from physical decomposition to high-dimensional representation enhances optical-assisted pathological analysis.
• The developed high-dimensional mapping framework offers a robust approach for cancer region identification.
• This technique shows promise for improving the accuracy of cancer diagnosis in unstained tissues.