Multimodal Phasor Analysis for Digital Pathology: Quantitative Characterization of Liver Iron Overload

Abstract

Hemochromatosis is a pathological condition characterized by the excessive absorption and storage of iron, primarily in the liver, which can lead to life-threatening complications. Histological tissue sections stained with Perls’ Prussian Blue (PB) are used to visualize iron accumulation under transmitted light. In clinical practice, pathologists evaluate the density and intensity of these Prussian Blue deposits through direct visual inspection, a semiquantitative process that can be influenced by individual interpretation and experience. Whole-slide scanners now digitize entire tissue sections into high-resolution RGB whole-slide images (WSIs), allowing image analysis algorithms to support pathologists with quantitative diagnostics. In this context, we developed a WSI processing method based on a multimodal phasor approach to quantify the extent, size and distribution of iron deposits, providing quantitative data that closely correlate with pathologists’ diagnoses based on the Scheuer scoring system. The Discrete Fourier Transform was used to map the RGB spectral data into the phasor plane, where the contribution of the PB stain was isolated using an unsupervised clustering method. Additionally, the size of Prussian Blue aggregates was determined through phasor analysis of the spatial autocorrelation function. Our results highlight the benefits of phasor-based analysis in segmentation, comprehensive analysis, and the efficient management of large image data sets. This method has potential applications in analyzing immunohistochemistry and immunofluorescence spectral images, as well as in simultaneously studying nanoparticle size and diffusion properties within tissues.

Ähnliche Arbeiten