Automatic cell detection and segmentation from H and E stained pathology slides using colorspace decorrelation stretching

Abstract

Purpose: Automatic cell segmentation plays an important role in reliable diagnosis and prognosis of patients. Most of the state-of-the-art cell detection and segmentation techniques focus on complicated methods to subtract foreground cells from the background. In this study, we introduce a preprocessing method which leads to a better detection and segmentation results compared to a well-known state-of-the-art work. Method: We transform the original red-green-blue (RGB) space into a new space defined by the top eigenvectors of the RGB space. Stretching is done by manipulating the contrast of each pixel value to equalize the color variances. New pixel values are then inverse transformed to the original RGB space. This altered RGB image is then used to segment cells. Result: The validation of our method with a well-known state-of-the-art technique revealed a statistically significant improvement on an identical validation set. We achieved a mean F1-score of 0.901. Conclusion: Preprocessing steps to decorrelate colorspaces may improve cell segmentation performances.

Characterization of active and infiltrative tumorous subregions from normal tissue in brain gliomas using multiparametric MRI

“…Quantitative assessment of histology images was performed automatically by a cell segmentation method which applies decorrelation and stretching of the colorspace in the preprocessing step for improving the performance of cell segmentation (Fig. ) . Cellular count (CC) within the specified area was calculated from the segmented images and used as a representative parameter of cellular density.…”

“…2). 27 Cellular count (CC) within the specified area was calculated from the segmented images and used as a representative parameter of cellular density. The relationship between the calculated CC on all regions and interpretation of the pathologist, as the gold standard, was examined.…”

Automatic cellularity assessment from post‐treated breast surgical specimens

“…While some studies focus on producing high quality nuclei segmentation for subsequent automated processing of pathology images , others claim perfect nuclei segmentation does not necessarily lead to a better classification performance . Most of the currently present nuclei segmentation techniques in the literature utilize one or a combination of watershed segmentation , graph‐cuts , matching‐based , Laplacian of Gaussian (LoG) , active contours , or pixel classification methods combined with pre‐ or post‐processing phases .…”

The Emergence of Pathomics

“…Mobadersany et al [12•] implemented a method that combines image analysis by CNNs with genomic markers into a unified machine learning model to predict the survival of patients with glioma, where the deep learning architecture consists of convolutional layers that are trained to predict image patterns associated with survival, fully connected layers that further transform image features from the convolutional layers, and a Cox proportional hazard layer that models survival data. Peikari and Martel [28] proposed a color transformation step that maps the red-green-blue (RGB) color space by computing eigenvectors of the RGB space to perform cell segmentation by utilizing the color-mapped image. A deep learning method is employed by Sirinukunwattana et al [29] to detect and classify nuclei in H&E stained color cancer tissue images by implementing a spatially constrained CNN for nucleus detection followed by a predictor that is coupled with a CNN for classification.…”

“…Ensembles of support vector machines (SVMs) were used by Manivannan et al [30] to detect and classify cellular patterns. Peikari et al [28,31] designed an analysis pipeline where a clustering operation is executed on input data to detect the structure of the data space, where a semi-supervised learning method is then executed to carry out classification using clustering information. Chen et al [32] developed a deep learning framework for segmentation that implemented a multi-task learning approach by the use of multi-level CNNs.…”