Humphries, M. P. | bioRxiv Science

bioRxiv2021

HistoClean: Open-source Software for Histological Image Pre-processing and Augmentation to Improve Development of Robust Convolutional Neural Networks

The growth of digital pathology over the past decade has opened new research pathways and insights in cancer prediction and prognosis. In particular, there has been a surge in deep learning and computer vision techniques to analyse digital images. Common practice in this area is to use image pre-processing and augmentation to prevent bias and overfitting, creating a more robust deep learning model. Herein we introduce HistoClean; user-friendly, graphical user interface that brings together multiple image processing modules into one easy to use toolkit. In this study, we utilise HistoClean to pre-process images for a simple convolutional neural network used to detect stromal maturity, improving the accuracy of the model at a tile, region of interest, and patient level. HistoClean is free and open-source and can be downloaded from the Github repository here: https://github.com/HistoCleanQUB/HistoClean.

cancer biology↗

bioRxiv2020

The clinical and molecular significance associated with STING signaling in estrogen receptor-positive early breast cancer

STING signaling in cancer is a crucial component of response to immunotherapy and other anti-cancer treatments. Conversely, STING signaling can promote tumor invasion and metastasis. Currently, there is no robust method of measuring STING activation in cancer. Here, we describe an immunohistochemistry-based assay with digital pathology assessment of STING in tumor cells. Using this novel approach, we identify perinuclear-localized expression of STING (pnSTING) in estrogen receptor-positive (ER+) breast cancer as an independent predictor of good prognosis, associated with immune cell infiltration and upregulation of immune checkpoints. Tumors with low pnSTING are immunosuppressed with increased infiltration of "M2" -polarised macrophages. In ER-disease, pnSTING does not have a significant prognostic role, and STING appears to be uncoupled from interferon responses. Importantly, a gene signature defining low pnSTING expression in ER+ disease is predictive of poor prognosis in independent datasets. Low pnSTING is associated with chromosomal instability, MYC amplification and mTOR signaling, suggesting novel therapeutic approaches for this subgroup.

cancer biology↗