A robust nonlinear tissue-component discrimination method for computational pathology

Abstract

Advances in digital pathology, specifically imaging instrumentation and data management, have allowed for the development of computational pathology tools with the potential for better, faster, and cheaper diagnosis, prognosis, and prediction of disease. Images of tissue sections frequently vary in color appearance across research laboratories and medical facilities due to differences in tissue fixation, staining protocols, and imaging instrumentation, leading to difficulty in the development of robust computational tools. To address this challenge, we propose a novel non-linear tissue-component discrimination (NLTD) method to automatically register the color space of histopathology images and visualize individual tissue components, independent of color differences between images. Our results show that the NLTD method could effectively discriminate different tissue components from different types of tissues prepared at different institutions. Further, we demonstrate that NLTD can improve the accuracy of nuclear detection and segmentation algorithms, compared to using conventional color deconvolution methods, and can quantitatively analyze immunohistochemistry images. Together, the NLTD method is objective, robust, and effective, and can be easily implemented in the emerging field of computational pathology.

Collagen IV and laminin‐1 expression in embryonic mouse lens using principal components analysis technique

“…This reduces the number of samples to be used and the types of labelling (Gry et al ., ; Sobhani et al ., ; Laas et al ., ). In other cases, the immunohistochemical images are used to illustrate the morphological patterns characteristic of the expression of the molecular markers (Moro et al ., ). Enhancement of the immunohistochemistry imaging contrast by performing the PCA analysis on the RGB image taken with a colour CCD camera, together with automatic classification algorithms or Machine Learning (Loukas et al ., ; Khorshed, ; Sarnecki et al ., ; Komura & Ishikawa, ; Van Eycke et al ., ). The degree of marking of an immunohistochemical image is usually done through a subjective numerical ‘scoring’ that assigns a ‘score’ to each image. The PCA method is usually used on the data of these numerical values (Ocak et al ., ).…”

“…Enhancement of the immunohistochemistry imaging contrast by performing the PCA analysis on the RGB image taken with a colour CCD camera, together with automatic classification algorithms or Machine Learning (Loukas et al ., ; Khorshed, ; Sarnecki et al ., ; Komura & Ishikawa, ; Van Eycke et al ., ).…”

“…(2) Enhancement of the immunohistochemistry imaging contrast by performing the PCA analysis on the RGB image taken with a colour CCD camera, together with automatic classification algorithms or Machine Learning (Loukas et al, 2003;Khorshed, 2013;Sarnecki et al, 2016;Komura & Ishikawa, 2018;Van Eycke et al, 2017). (3) The degree of marking of an immunohistochemical image is usually done through a subjective numerical 'scoring' that assigns a 'score' to each image.…”

Spatial Statistics for Segmenting Histological Structures in H&E Stained Tissue Images

“…As future work, we can automate the color preprocessing step to be completely unsupervised using the recently proposed non-linear tissue-component separation method [35]. In addition, we can improve the performance of our segmentation algorithms, by combining multi-scale methods with higher-order statistics in H&E-hue space.…”

Computational discovery of tissue morphology biomarkers in very long-term survivors with pancreatic ductal adenocarcinoma

“…Briefly, each/a raw WSI is converted into two nuclei-rich and extracellular matrix-rich areas in the WSI using the previously published nonlinear tissue-component discrimination (NLTD) method 29 . To minimize the memory loading of the computer, WSIs are computationally partitioned into an array of smaller tissue tiles (500 m x 500 m).…”