AI based pre-screening of large bowel cancer via weakly supervised learning of colorectal biopsy histology images

Abstract

Histopathological examination is a pivotal step in the diagnosis and treatment planning of many major diseases. To facilitate the diagnostic decision-making and reduce the workload of pathologists, we present an AI-based pre-screening tool capable of identifying normal and neoplastic colon biopsies. To learn the differential histological patterns from whole slides images (WSIs) stained with hematoxylin and eosin (H&E), our proposed weakly supervised deep learning method requires only slide-level labels and no detailed cell or region-level annotations. The proposed method was developed and validated on an internal cohort of biopsy slides (n=4 292) from two hospitals labeled with corresponding diagnostic categories assigned by pathologists after reviewing case reports. Performance of the proposed colon cancer pre-screening tool was evaluated in a cross-validation setting using the internal cohort (n=4 292) and also by an external validation on The Cancer Genome Atlas (TCGA) cohort (n=731). With overall cross-validated classification accuracy (AUROC = 0.9895) and external validation accuracy (AUROC = 0.9746), the proposed tool promises high accuracy to assist with the pre-screening of colorectal biopsies in clinical practice. Analysis of saliency maps confirms the representation of disease heterogeneity in model predictions and their association with relevant pathological features. The proposed AI tool correctly reported some slides as neoplastic while clinical reports suggested they were normal. Additionally, we analyzed genetic mutations and gene enrichment analysis of AI-generated neoplastic scores to gain further insight into the model predictions and explore the association between neoplastic histology and genetic heterogeneity through representative genes and signaling pathways.

EBHI-Seg: A novel enteroscope biopsy histopathological hematoxylin and eosin image dataset for image segmentation tasks

“…This dataset contained 3,125 images with an image size of 224 × 224 pixels and is small in type and number. In Bilal et al ( 41 ), two datasets named “UHCW” and “TCGA” are proposed. The first dataset is a colorectal cancer biopsy sequence developed at the University Hospital of Coventry and Warwickshire (UHCW) for internal validation of the rectal biopsy trial.…”

Human-AI Collaboration: The Effect of AI Delegation on Human Task Performance and Task Satisfaction

“…AI delegation could also be used in high-stakes decisionmaking domains (e.g., medicine) to reduce the workload of medical experts. For instance, in the context of cancer screening, the AI model can be used to identify simple cases so that the medical experts can focus on the complex cases delegated to them [10]. Several recent works propose approaches that enable the AI model to delegate a subset of instances to a human while taking both its own and the human's capabilities into consideration [31,39,50,54,56,72].…”

Classification of Colorectal Polyps from Histopathological Images using Ensemble of ConvNeXt Variants

“…As the number of biopsies increases, it gets harder and harder to detect early-stage disease, and more diagnostic information may be required to provide the best standard of care. Moreover, over the past decade according to [2], pathologic colon biopsy slide volumes are doubled. This increase in the workload increases the cost.…”

“…Moreover, DL based approaches are actively employed for pattern classification or analysis of the medical images. There are extensive studies conducted with numerous methods focusing on the individual diagnosis of colorectal cancer from histopathology images, such as classification of colorectal adenocarcinoma [ [2], [5], [6], [7], [8], [9], [10], [11], [12], [13], [14], [15]], colon polyp classification [ [16], [17], [4], [18], [19], [20], [21]] and colon gland classification [ [5], [22], [23]].…”

“…Moreover, in their work they compare the model performance on gray scale, RGB and Macenko stain normalized histology images. In [10] [2] propose a framework to classify colonic WSIs as neoplastic or normal with a weakly supervised deep learning. They evaluate the performance of the system by using custom collected WSIs and The Cancer Genome Atlas (TCGA) database, and overall area under the receiver operating characteristic curve (AUROC) of 0.9746 is achieved.…”