A5041645982- AI in cancer detection
- Radiomics and Machine Learning in Medical Imaging
- Cancer Immunotherapy and Biomarkers
- Digital Imaging for Blood Diseases
- Immunotherapy and Immune Responses
- Cell Image Analysis Techniques
- Colorectal Cancer Screening and Detection
- Monoclonal and Polyclonal Antibodies Research
- Lung Cancer Treatments and Mutations
- Single-cell and spatial transcriptomics
- Ferroptosis and cancer prognosis
- Oral Health Pathology and Treatment
- Lung Cancer Diagnosis and Treatment
- Cancer Genomics and Diagnostics
- Medical Image Segmentation Techniques
- Image Processing Techniques and Applications
- Esophageal Cancer Research and Treatment
- Cancer Cells and Metastasis
- Medical Imaging and Analysis
- Oral and Maxillofacial Pathology
- Gastric Cancer Management and Outcomes
- Advanced Biosensing Techniques and Applications
- Brain Tumor Detection and Classification
- Advanced Neural Network Applications
- Pancreatic and Hepatic Oncology Research
University of Warwick
2016-2025
Institute of Cancer Research
2018-2023
Royal Stoke University Hospital
2023
University Hospitals Coventry and Warwickshire NHS Trust
2022
Bridge University
2022
Cancer Research UK
2018-2019
University of Kashmir
2019
Detection and classification of cell nuclei in histopathology images cancerous tissue stained with the standard hematoxylin eosin stain is a challenging task due to cellular heterogeneity. Deep learning approaches have been shown produce encouraging results on various studies. In this paper, we propose Spatially Constrained Convolutional Neural Network (SC-CNN) perform nucleus detection. SC-CNN regresses likelihood pixel being center nucleus, where high probability values are spatially...
BackgroundDetermining the status of molecular pathways and key mutations in colorectal cancer is crucial for optimal therapeutic decision making. We therefore aimed to develop a novel deep learning pipeline predict from whole-slide images haematoxylin eosin-stained slides as an alternative current tests.MethodsIn this retrospective study, we used 502 diagnostic primary tumours 499 patients The Cancer Genome Atlas colon rectal (TCGA-CRC-DX) cohort developed weakly supervised framework...
Accurate and timely detection of plant diseases can help mitigate the worldwide losses experienced by horticulture agriculture industries each year. Thermal imaging provides a fast non-destructive way scanning plants for diseased regions has been used various researchers to study effect disease on thermal profile plant. However, image affected known be environmental conditions which include leaf angles depth canopy areas accessible camera. In this paper, we combine visible light data with...
The diagnosis and surgical resection using Magnetic Resonance (MR) images in brain tumors is a challenging task to minimize the neurological defects after surgery owing non-linear nature of size, shape, textural variation. Radiologists, clinical experts, surgeons examine MRI scans available methods, which are tedious, error-prone, time-consuming, still exhibit positional accuracy up 2–3 mm, very high case cells. In this context, we propose an automated Ultra-Light Brain Tumor Detection...
The development of deep segmentation models for computational pathology (CPath) can help foster the investigation interpretable morphological biomarkers. Yet, there is a major bottleneck in success such approaches be-cause supervised learning require an abundance accurately labelled data. This issue exacerbated field CPath because generation detailed annotations usually demands input pathologist to be able distinguish between different tissue constructs and nuclei. Manually labelling nuclei...
Abstract Before squamous cell lung cancer develops, precancerous lesions can be found in the airways. From longitudinal monitoring, we know that only half of such become cancer, whereas a third spontaneously regress. Although recent studies have described presence an active immune response high-grade lesions, mechanisms underpinning clinical regression remain unknown. Here, show host surveillance is strongly implicated lesion regression. Using bronchoscopic biopsies from human subjects, find...
The recent surge in performance for image analysis of digitised pathology slides can largely be attributed to the advances deep learning. Deep models used initially localise various structures tissue and hence facilitate extraction interpretable features biomarker discovery. However, these are typically trained a single task therefore scale poorly as we wish adapt model an increasing number different tasks. Also, supervised learning very data hungry rely on large amounts training perform...
Computational pathology has seen rapid growth in recent years, driven by advanced deep-learning algorithms. Due to the sheer size and complexity of multi-gigapixel whole-slide images, best our knowledge, there is no open-source software library providing a generic end-to-end API for image analysis using practices. Most researchers have designed custom pipelines from bottom up, restricting development algorithms specialist users. To help overcome this bottleneck, we present TIAToolbox, Python...
Recent advances in whole-slide imaging (WSI) technology have led to the development of a myriad computer vision and artificial intelligence-based diagnostic, prognostic, predictive algorithms. Computational Pathology (CPath) offers an integrated solution utilise information embedded pathology WSIs beyond what can be obtained through visual assessment. For automated analysis validation machine learning (ML) models, annotations at slide, tissue, cellular levels are required. The annotation...
Diagnostic, prognostic and therapeutic decision-making of cancer in pathology clinics can now be carried out based on analysis multi-gigapixel tissue images, also known as whole-slide images (WSIs). Recently, deep convolutional neural networks (CNNs) have been proposed to derive unsupervised WSI representations; these are attractive they rely less expert annotation which is cumbersome. However, a major trade-off that higher predictive power generally comes at the cost interpretability,...
Abstract Beyond tertiary lymphoid structures, a significant number of immune-rich areas without germinal center-like structures are observed in non–small cell lung cancer. Here, we integrated transcriptomic data and digital pathology images to study the prognostic implications, spatial locations, constitution immune rich (immune hotspots) cohort 935 patients with cancer from The Cancer Genome Atlas. A high intratumoral hotspot score, which measures proportion hotspots interfacing tumor...
Nuclear detection, segmentation and morphometric profiling are essential in helping us further understand the relationship between histology patient outcome. To drive innovation this area, we setup a community-wide challenge using largest available dataset of its kind to assess nuclear cellular composition. Our challenge, named CoNIC, stimulated development reproducible algorithms for recognition with real-time result inspection on public leaderboards. We conducted an extensive...
Counting of mitotic figures is a fundamental step in grading and prognostication several cancers. However, manual mitosis counting tedious time-consuming. In addition, variation the appearance causes high degree discordance among pathologists. With advances deep learning models, automatic detection algorithms have been proposed but they are sensitive to domain shift often seen histology images. We propose robust efficient two-stage framework, which comprises candidate segmentation (Detecting...
Abstract Oral epithelial dysplasia (OED) is a premalignant histopathological diagnosis given to lesions of the oral cavity. Its grading suffers from significant inter-/intra-observer variability, and does not reliably predict malignancy progression, potentially leading suboptimal treatment decisions. To address this, we developed an artificial intelligence (AI) algorithm, that assigns Malignant Transformation (OMT) risk score based on Haematoxylin Eosin (H&E) stained whole slide images...
Stain colour estimation is a prominent factor of the analysis pipeline in most histology image processing algorithms. Providing reliable and efficient stain deconvolution approach fundamental for robust algorithm. In this paper, we propose novel method images. This statistically analyses multi-resolutional representation to separate independent observations out correlated ones. We then estimate mixing matrix using filtered uncorrelated data. conducted an extensive set experiments compare...
Detecting various types of cells in and around the tumor matrix holds a special significance characterizing micro-environment for cancer prognostication research. Automating tasks detecting, segmenting, classifying nuclei can free up pathologists' time higher value reduce errors due to fatigue subjectivity. To encourage computer vision research community develop test algorithms these tasks, we prepared large diverse dataset nucleus boundary annotations class labels. The has over 46,000 from...
Triple-negative breast cancer (TNBC) is known to have a relatively poor outcome with variable prognoses, raising the need for more informative risk stratification. We investigated set of digital, artificial intelligence (AI)-based spatial tumour microenvironment (sTME) features and explored their prognostic value in TNBC. After performing tissue classification on digitised haematoxylin eosin (H&E) slides TNBC cases, we employed deep learning-based algorithm segment regions into tumour,...
Objective To develop an interpretable artificial intelligence algorithm to rule out normal large bowel endoscopic biopsies, saving pathologist resources and helping with early diagnosis. Design A graph neural network was developed incorporating domain knowledge classify 6591 whole-slides images (WSIs) of biopsies from 3291 patients (approximately 54% female, 46% male) as or abnormal (non-neoplastic neoplastic) using clinically driven features. One UK National Health Service (NHS) site used...
Tumour infiltrating lymphocytes (TILs) are a prognostic parameter in triple-negative and human epidermal growth factor receptor 2 (HER2)-positive breast cancer (BC). However, their role luminal (oestrogen positive HER2 negative (ER + /HER2-)) BC remains unclear. In this study, we used artificial intelligence (AI) to assess the significance of TILs large well-characterised cohort BC.
We propose a novel multiple-input multiple-output convolution neural network (MIMO-Net) for cell segmentation in fluorescence microscopy images. The proposed trains the parameters using multiple resolutions of input image, connects intermediate layers better localization and context generates output multi-resolution deconvolution filters. MIMO-Net allows us to deal with variable intensity boundaries highly size mouse pancreatic tissue by adding extra convolutional which bypass max-pooling...
Detection of tumor nuclei in cancer histology images requires sophisticated techniques due to the irregular shape, size and chromatin texture nuclei. Some very recently proposed methods employ deep convolutional neural networks (CNNs) detect cells H&E stained images. However, all such use some form raw pixel intensities as input rely on CNN learn features. In this work, we extend a spatially constrained (SC-CNN) by proposing features that capture characteristics show although produces good...