Biology subjects

Bankhead, P.

Publications and source records attributed to Bankhead, P..

11 recordsLinked to original sources

bioRxiv2026

Explainable machine learning relates histological to genomic pathology

Background & AimsHaematoxylin and eosin (H&E) staining remains the diagnostic gold standard for solid cancers, including hepatocellular carcinoma, and is increasingly complemented by genomic profiling for precision medicine. Inferring genomic alterations directly from H&E images could streamline testing, but heterogeneity and biases in human training data limit interpretation of genotype-phenotype associations. Here, we aimed to relate histologic to genomic pathology to provide biological explainability for mutation prediction models and assess the impact of germline variation on model performance. MethodsWe analysed 597 murine liver tumours with matched whole-genome sequencing and histopathology (163,835 image tiles; 22.9 million nuclei). Our controlled in vivo design accounted for germline variation, biological sex, and causal mutagen (N-diethylnitrosamine), removing confounding factors present in human cohorts. We trained and evaluated deep learning and supervised machine learning models to predict germline variation and cancer driver alterations from H&E. ResultsModelling accurately predicted germline and somatic alterations from histology, at both locus-specific and genome-wide scales. Quantitative image analysis revealed an unexpected association between Egfr driver mutations and hepatic steatosis, linking genotype to an interpretable morphological phenotype. While model performance declined when applied to tumours from unrepresented genetic backgrounds, this limitation was biologically informative, revealing strain-dependent differences in tumour evolution, notably the prevalence of whole-genome duplication. ConclusionsMachine learning integration of histological and genomic pathology enables accurate, interpretable inference of genetic alterations from H&E, potentially reducing reliance on costly ancillary molecular assays. Our predictions are supported by human-interpretable biological features, addressing concerns around "black-box" technologies. However, caution is required when applying such methods to samples with a genetic background that, even if closely related, is beyond the genetic horizon of training data.

pathology↗

bioRxiv2024

A novel channel invariant architecture for the segmentation of cells and nuclei in multiplexed images using InstanSeg

The quantitative analysis of bioimaging data increasingly depends on the accurate segmentation of cells and nuclei, a significant challenge for the analysis of high-plex imaging data. Current deep learning-based approaches to segment cells in multiplexed images require reducing the input to a small and fixed number of input channels, discarding imaging information in the process. We present Channel Net, a novel deep learning architecture for generating three-channel representations of multiplexed images irrespective of the number or ordering of imaged biomarkers. When combined with InstanSeg, ChannelNet sets a new benchmark for the segmentation of cells and nuclei on public multiplexed imaging datasets. We provide an open implementation of our method and integrate it in open source software. Our code and models are available on https://github.com/instanseg/instanseg.

bioinformatics↗

bioRxiv2024

GelGenie: an AI-powered framework for gel electrophoresis image analysis

Gel electrophoresis is a ubiquitous laboratory method for the separation and semi-quantitative analysis of biomolecules. However, gel image analysis principles have barely advanced for decades, in stark contrast to other fields where AI has revolutionised data processing. Here, we show that an AI-based system can automatically identify gel bands in seconds for a wide range of experimental conditions, far surpassing the capabilities of current software. We used a dataset containing 500+ images of manually-labelled gels to train various U-Nets to accurately identify bands through segmentation, i.e. classifying pixels as band or background. When applied to gel electrophoresis data from other laboratories, our system generated results that quantitatively matched those of the original authors. We have publicly released our models through GelGenie, an open-source application that allows users to extract bands from gel images on their own devices, with no expert knowledge or experience required.

molecular biology↗

bioRxiv2024

CLEAR-IT: Contrastive Learning Enabled Accurate Registration of Immune and Tumor cells from multiplexed images with limited labels in a platform-independent manner

Accurate phenotyping of cells in the tumor microenvironment is crucial for understanding cancer biology and developing effective therapies. However, current methods require precise cell segmentations and struggle to generalize across different imaging modalities, limiting their utility in digital pathology. Here, we show that Contrastive Learning Enabled Accurate Registration of Immune and Tumor Cells (CLEAR-IT) overcomes these limitations, providing a robust and versatile tool for cell phenotyping. CLEAR-IT accurately phenotypes cells comparable to state-of-the-art methods, generalizes across multiplex imaging modalities, maintains high performance even with limited number of labels, and enables the extraction of prognostic markers. Additionally, CLEAR-IT can be combined with existing methods to boost their performance, whereas its lack of need for precise cell segmentations significantly reduces training efforts. This method enhances the robustness and efficiency of digital pathology workflows, making it a valuable tool for cancer research and diagnostics.

bioinformatics↗

bioRxiv2023

AimSeg: a machine-learning-aided tool for axon, inner tongue and myelin segmentation

Electron microscopy (EM) images of axons and their ensheathing myelin from both the central and peripheral nervous system are used for assessing myelin formation, degeneration (demyelination) and regeneration (remyelination). The g-ratio is the gold standard measure of assessing myelin thickness and quality, and traditionally is determined from measurements done manually from EM images - a time-consuming endeavour with limited reproducibility. These measurements have also historically neglected the innermost uncompacted myelin sheath, known as the inner myelin tongue. Nonetheless, the inner tongue has been shown to be important for myelin growth and some studies have reported that certain conditions can elicit its enlargement. Ignoring this fact may bias the standard g-ratio analysis, whereas quantifying the uncompacted myelin has the potential to provide novel insights in the myelin field. In this regard, we have developed AimSeg, a bioimage analysis tool for axon, inner tongue and myelin segmentation. Aided by machine learning classifiers trained on tissue undergoing remyelination, AimSeg can be used either as an automated workflow or as a user-assisted segmentation tool. Validation results show good performance segmenting all three fibre components, with the assisted segmentation showing the potential for further improvement with minimal user intervention. This results in a considerable reduction in time for analysis compared with manual annotation. AimSeg could also be used to build larger, high quality ground truth datasets to train novel deep learning models. Implemented in Fiji, AimSeg can use machine learning classifiers trained in ilastik. This, combined with a user-friendly interface and the ability to quantify uncompacted myelin, makes AimSeg a unique tool to assess myelin growth. Author SummaryMyelin is formed by specialised cells that wrap themselves around axons and has a major role in the function, protection, and maintenance of nerves. These functions are disturbed by demyelinating diseases, such as multiple sclerosis. In this work we present AimSeg, a new tool based on artificial intelligence algorithms (machine learning) to assess myelin growth on electron microscopy images. Whereas standard metrics and previous computational methods focus on quantifying compact myelin, AimSeg also quantifies the inner myelin tongue (uncompacted myelin). This structure has been largely overlooked despite the fact that it has an important role in the process of myelin growth (both during development and in the adult brain) and recent studies have reported morphological changes associated with some diseases. We report the performance of AimSeg, both as a fully automated approach and in an assisted segmentation workflow that enables the user to curate the results on-the-fly while reducing human intervention to the minimum. Therefore, AimSeg stands as a novel bioimage analysis tool that meets the challenges of assessing myelin growth by supporting both standard metrics for myelin evaluation and the quantification of the uncompacted myelin in different conditions.

bioinformatics↗

bioRxiv2022

BioImage Model Zoo: A Community-Driven Resource for Accessible Deep Learning in BioImage Analysis

Deep learning-based approaches are revolutionizing imaging-driven scientific research. However, the accessibility and reproducibility of deep learning-based workflows for imaging scientists remain far from sufficient. Several tools have recently risen to the challenge of democratizing deep learning by providing user-friendly interfaces to analyze new data with pre-trained or fine-tuned models. Still, few of the existing pre-trained models are interoperable between these tools, critically restricting a models overall utility and the possibility of validating and reproducing scientific analyses. Here, we present the BioImage Model Zoo (https://bioimage.io): a community-driven, fully open resource where standardized pre-trained models can be shared, explored, tested, and downloaded for further adaptation or direct deployment in multiple end user-facing tools (e.g., ilastik, deepImageJ, QuPath, StarDist, ImJoy, ZeroCostDL4Mic, CSBDeep). To enable everyone to contribute and consume the Zoo resources, we provide a model standard to enable cross-compatibility, a rich list of example models and practical use-cases, developer tools, documentation, and the accompanying infrastructure for model upload, download and testing. Our contribution aims to lay the groundwork to make deep learning methods for microscopy imaging findable, accessible, interoperable, and reusable (FAIR) across software tools and platforms.

bioinformatics↗

bioRxiv2021

Development of a semi-automated method for tumor budding assessment in colorectal cancer and comparison with manual methods

Tumor budding is an established prognostic feature in multiple cancers but routine assessment has not yet been incorporated into clinical pathology practice. Recent efforts to standardize and automate assessment have shifted away from haematoxylin and eosin (H&E)-stained images towards cytokeratin (CK) immunohistochemistry. In this study, we compare established manual H&E and cytokeratin budding assessment methods with a new, semi-automated approach built within the QuPath open-source software. We applied our method to tissue cores from the advancing tumor edge in a cohort of stage II/III colon cancers (n=186). The total number of buds detected by each method, over the 186 TMA cores, were as follows; manual H&E (n=503), manual CK (n=2290) and semi-automated (n=5138). More than four times the number of buds were detected using CK compared to H&E. A total of 1734 individual buds were identified both using manual assessment and semi-automated detection on CK images, representing 75.7% of the total buds identified manually (n=2290) and 33.7% of the total buds detected using our proposed semi-automated method (n=5138). Higher bud scores by the semi-automated method were due to any discrete area of CK immunopositivity within an accepted area range being identified as a bud, regardless of shape or crispness of definition, and to inclusion of tumor cell clusters within glandular lumina ("luminal pseudobuds"). Although absolute numbers differed, semi-automated and manual bud counts were strongly correlated across cores ({rho}=0.81, p<0.0001). Despite the random, rather than "hotspot", nature of tumor core sampling, all methods of budding assessment demonstrated poorer survival associated with higher budding scores. In conclusion, we present a new QuPath-based approach to tumor budding assessment, which compares favorably to current established methods and offers a freely-available, rapid and transparent tool that is also applicable to whole slide images.

pathology↗

bioRxiv2019

Pan-cancer image-based detection of clinically actionable genetic alterations

Precision treatment of cancer relies on genetic alterations which are diagnosed by molecular biology assays.1 These tests can be a bottleneck in oncology workflows because of high turnaround time, tissue usage and costs.2 Here, we show that deep learning can predict point mutations, molecular tumor subtypes and immune-related gene expression signatures3,4 directly from routine histological images of tumor tissue. We developed and systematically optimized a one-stop-shop workflow and applied it to more than 4000 patients with breast5, colon and rectal6, head and neck7, lung8,9, pancreatic10, prostate11 cancer, melanoma12 and gastric13 cancer. Together, our findings show that a single deep learning algorithm can predict clinically actionable alterations from routine histology data. Our method can be implemented on mobile hardware14, potentially enabling point-of-care diagnostics for personalized cancer treatment in individual patients.

genetics↗

bioRxiv2019

Deep learning detects virus presence in cancer histology

Oncogenic viruses like human papilloma virus (HPV) or Epstein Barr virus (EBV) are a major cause of human cancer. Viral oncogenesis has a direct impact on treatment decisions because virus-associated tumors can demand a lower intensity of chemotherapy and radiation or can be more susceptible to immune check-point inhibition. However, molecular tests for HPV and EBV are not ubiquitously available.\n\nWe hypothesized that the histopathological features of virus-driven and non-virus driven cancers are sufficiently different to be detectable by artificial intelligence (AI) through deep learning-based analysis of images from routine hematoxylin and eosin (HE) stained slides. We show that deep transfer learning can predict presence of HPV in head and neck cancer with a patient-level 3-fold cross validated area-under-the-curve (AUC) of 0.89 [0.82; 0.94]. The same workflow was used for Epstein-Barr virus (EBV) driven gastric cancer achieving a cross-validated AUC of 0.80 [0.70; 0.92] and a similar performance in external validation sets. Reverse-engineering our deep neural networks, we show that the key morphological features can be made understandable to humans.\n\nThis workflow could enable a fast and low-cost method to identify virus-induced cancer in clinical trials or clinical routine. At the same time, our approach for feature visualization allows pathologists to look into the black box of deep learning, enabling them to check the plausibility of computer-based image classification.

cancer biology↗

bioRxiv2017

Bcl-xL as a poor prognostic biomarker and predictor of response to adjuvant chemotherapy specifically in BRAF-mutant stage II and III colon cancer

BackgroundBRAF mutation occurs in 8-15% of colon cancers (CC), and is associated with poor prognosis in metastatic disease. Compared to wild-type BRAF (BRAFWT) disease, stage II/III CC patients with BRAF mutant (BRAFMT) tumors have shorter overall survival after relapse; however, time-to-relapse is not significantly different. The aim of this investigation was to identify, and validate, novel predictors of relapse of stage II/III BRAFMT CC.\n\nPatients and methodsWe used gene expression data from a cohort of 460 patients (GSE39582) to perform a supervised classification analysis based on risk-of-relapse within BRAFMT stage II/III CC, to identify transcriptomic biomarkers associated with prognosis within this genotype. These findings were validated using immunohistochemistry in an independent population-based cohort of Stage II/III CC (n=691), applying Cox proportional hazards analysis to determine associations with survival.\n\nResultsHigh gene expression levels of Bcl-xL, a key regulator of apoptosis, were associated with increased risk of relapse, specifically in BRAFMT tumors (HR=8.3, 95% CI 1.7-41.7), but not KRASMT/BRAFWT or KRASWT/BRAFWT tumors. High Bcl-xL protein expression in BRAFMT, untreated, stage II/III CC was confirmed to be associated with an increased risk of death in an independent cohort (HR=12.13, 95% CI 2.49-59.13). Additionally, BRAFMT tumors with high levels of Bcl-xL protein expression appeared to benefit from adjuvant chemotherapy (P for interaction =0.006), indicating the potential predictive value of Bcl-xL expression in this setting.\n\nConclusionsThese findings provide evidence that Bcl-xL gene and/or protein expression identifies a poor prognostic subgroup of BRAFMT stage II/III CC patients, who may benefit from adjuvant chemotherapy.\n\nKey MessageO_LIUsing a combination of computational biology discovery and immunohistochemistry validation in independent patient cohorts, we show that high expression of the apoptosis regulator Bcl-xL is associated with disease relapse specifically within BRAF mutant stage II/III colon cancer.\nC_LIO_LIThis data could enable tailored disease management to reduce relapse rates in the most aggressive subtype.\nC_LI

cancer biology↗

bioRxiv2017

QuPath: Open source software for digital pathology image analysis

QuPath is new bioimage analysis software designed to meet the growing need for a user-friendly, extensible, open-source solution for digital pathology and whole slide image analysis. In addition to offering a comprehensive panel of tumor identification and high-throughput biomarker evaluation tools, QuPath provides researchers with powerful batch-processing and scripting functionality, and an extensible platform with which to develop and share new algorithms to analyze complex tissue images. Furthermore, QuPaths flexible design make it suitable for a wide range of additional image analysis applications across biomedical research.

pathology↗