Glandular Morphometrics for Objective Grading of Colorectal Adenocarcinoma Histology Images

Abstract

Determining the grade of colon cancer from tissue slides is a routine part of the pathological analysis. In the case of colorectal adenocarcinoma (CRA), grading is partly determined by morphology and degree of formation of glandular structures. Achieving consistency between pathologists is difficult due to the subjective nature of grading assessment. An objective grading using computer algorithms will be more consistent, and will be able to analyse images in more detail. In this paper, we measure the shape of glands with a novel metric that we call the Best Alignment Metric (BAM). We show a strong correlation between a novel measure of glandular shape and grade of the tumour. We used shape specific parameters to perform a two-class classification of images into normal or cancerous tissue and a three-class classification into normal, low grade cancer, and high grade cancer. The task of detecting gland boundaries, which is a prerequisite of shape-based analysis, was carried out using a deep convolutional neural network designed for segmentation of glandular structures. A support vector machine (SVM) classifier was trained using shape features derived from BAM. Through cross-validation, we achieved an accuracy of 97% for the two-class and 91% for three-class classification.

CGC-Net: Cell Graph Convolutional Network for Grading of Colorectal Cancer Histology Images

“…Image Accuracy BAM-1 [2] 87.79 ± 2.32 BAM-2 [2] 90.66 ± 2.45 Context -G [42] 89.96 ± 3.54 ResNet50 [26] 92.08 ± 2.08 MobileNet [27] 92.78 ± 2.74 InceptionV3 [46] 91.37 ± 3.55 Xception [8] 92.09 ± 0.98 CA-CNN [41] 95.70 ± 3.04 Ours 97.00 ± 1.10 Nuclei sampling strategy: As introduced in Section 3.2, to reduce the size of the cell graph and preserve the cell architecture information in the graph, we propose a representative nuclei sampling strategy. Experiments of utilizing different sample strategy can be seen in Table 1, where Random, Farthest and Fuse denote the random sampling, farthest sampling and our proposed sampling method.…”

“…Colorectal Cancer(CRC) dataset: The proposed method is evaluated on the CRC dataset [2], which consists of 139 images taken from WSIs with an average size of 4548×7520 at 20× magnification. The images are divided into normal, low grade and high grade based on the degree of gland differentiation.…”

Artificial intelligence in digital pathology: a roadmap to routine use in clinical practice

“…Her‐2 and Ki67 tools are already available, with many other markers in development), disease quantification, morphometrics, tumour detection and cancer grading, and rare event screening (e.g. highlighting samples where tumour or micrometastases are detected and need pathologist review, and those which are negative and may not need review) .…”

Context-Aware Convolutional Neural Network for Grading of Colorectal Cancer Histology Images

“…Awan et al [11] presented a method for two-tier CRC grading based on the extent of deviation of the gland from its normal shape (circular/elliptical). They proposed a novel Best Alignment Metric (BAM) for this purpose.…”

“…The proposed framework is evaluated on two different tissue types for two different tasks in order to demonstrate its capabilities. Our colorectal cancer dataset [11] is comprised of visual fields (refer as images for simplicity) extracted from colorectal histology images based on a two-tier [13], [14] grading system. The CRC dataset consists of 139 images with an average size of 4, 548 × 7, 520 pixels obtained at 20× magnification.…”