A paper on “Automated Annotator: Capturing Expert Knowledge for Free” by the DART digital pathology and radiomics team was presented at EMBC, the 43rd Annual International Conference of the IEEE Engineering in Medicine and Biology Society Intervention, November 1-5, 2021. EMBC is the flagship conference of the IEEE EMB Society and the peer-reviewed conference proceedings attract submission of innovative work through the international gathering.
Paper Abstract: Deep learning enabled medical image analysis is heavily reliant on expert annotations which is costly. We present a simple yet effective automated annotation pipeline that uses autoencoder based heatmaps to exploit high level information that can be extracted from a histology viewer in an unobtrusive fashion. By predicting heatmaps on unseen images the model effectively acts like a robot annotator. The method is demonstrated in the context of coeliac disease histology images in this initial work, but the approach is task agnostic and may be used for other medical image annotation applications. The results are evaluated by a pathologist and also empirically using a deep network for coeliac disease classification. Initial results using this simple but effective approach are encouraging and merit further investigation, especially considering the possibility of scaling this up to a large number of users.
Co-author, Tapabrata (Rohan) Chakraborty said “We present a pilot demonstration to show how AI maybe be used to learn the annotations of pathologists from a relatively few examples without interrupting their workflow, and then use that knowledge to annotate new images in an automated fashion, thus reducing time and cost. This paper is based on an earlier work on coeliac disease histology data by current DART members and past students, and it is being exploited and extended as a generalised approach in the current DART project on lung histology data.”
DART presentation at Machine Learning in Medical Imaging workshop
A paper on “Contrastive Representations for Continual Learning of Fine-grained Histology Images” by the DART digital pathology and radiomics team has been presented at the Machine Learning in Medical Imaging (MLMI) pre-conference workshop at MICCAI 2021, the 24th International Conference on Medical Image Computing and Computer Assisted Intervention, September 27th to October 1st 2021. The annual MICCAI conference attracts world leading biomedical scientists, engineers, and clinicians from a wide range of disciplines associated with medical imaging and computer assisted intervention. See the Conference Proceedings here.
Paper Abstract: We show how a simple autoencoder based deep network with a contrastive loss can effectively learn representations in a continual/incremental manner with limited labelling. This is of particular interest to the biomedical imaging research community, for whom the visual task is often a binary decision (healthy vs. disease) with limited quantity data and costly labelling. For such applications, the proposed method provides a light-weight option of 1) representing patterns with relatively few training samples using a novel collaborative contrastive loss function 2) update the autoencoder based deep network in an unsupervised fashion for continual learning for new incoming data. We overcome the drawbacks of existing methods through planned technical design, and demonstrate the efficacy of the proposed method on three histology image classification tasks (lung, colon, breast cancer) with SOTA results.
Lead author, Tapabrata (Rohan) Chakraborty says “It is great to present the first paper of the DART project at the MLMI workshop at MICCAI, a leading venue on engineering applications in biomedical imaging. This introductory methodological paper should motivate more translational articles from the DART project in the near future as we delve into clinical applications.”
AI software could help clinicians diagnose lung cancer earlier and reduce waiting lists for patients
A new artificial intelligence software that will help doctors to make quicker and more accurate decisions when diagnosing potentially cancerous lung lesions, has received major government funding.
This is a step towards bringing the benefits of earlier diagnosis of lung cancer, the leading source of cancer death, to all patients across the National Health Service.
The lung cancer predication AI software, Virtual Nodule Clinic (VNC), which has been developed by Optellum (Oxford, UK), will examine lung nodules to determine whether they are benign or malignant. The pioneering AI solution has been shown to outperform existing methods to predict malignancy in nodules in a multi-centre study conducted by Nottingham, Oxford and Leeds clinicians with results published in BMJ Thorax last year.
Lung cancer is the leading cause of cancer deaths in the UK, accounting for 21% of all cancer deaths in any one year. When diagnosed at an early stage, almost 57% of people in the UK with lung cancer will survive their disease for five years or more, compared with only 3% when the disease is the latest stage. Currently, around three-quarters of lung cancer cases are diagnosed at a late stage in the UK, although the survival rate for small tumours treated at Stage IA is up to 90%.
Waiting lists are a problem for the NHS even at the best of times and due to the pandemic, researchers have estimated between 1,235 and 1,372 additional deaths in lung cancer due to diagnostic delays.
DOLCE is a landmark research project led by Professor David Baldwin, Honorary Professor of Medicine at the University of Nottingham, and Consultant Physician at Nottingham University Hospitals NHS Trust. It will show how many CT scans, expensive PET scans and biopsies are saved by the Optellum software and how much faster the diagnosis of cancer is confirmed. If the utility and safety are confirmed, the solution could be implemented nationally with fewer harms to patients, reduced anxiety for patients waiting for tests and substantial savings in precious radiology resources.
The project is part of the NHS AI Lab’s £140million AI in Health and Care Award. The AI in Health and Care Award aims to accelerate the testing and evaluation of AI in the NHS so patients can benefit from faster and more personalised diagnosis and greater efficiency in screening services.
The Health and Social Care Secretary Matt Hancock, announced this latest award last week, which will see 38 projects awarded a share of £36million to test state of the art AI technology.
Professor Baldwin, who is also Chair of the Clinical Expert Group for Lung Cancer, NHS England and co-author of the current clinical guidelines for lung cancer in the UK, said: “We are delighted to receive this award. This technology is truly transformative and we have previously shown that this software can help us to safely discharge more people with pulmonary nodules earlier, reducing anxiety amongst patients waiting for repeat scans and also the need for potentially harmful tests. This is also very important to the NHS because it will reduce the pressure on radiology resources.
“In this latest study – DOLCE, we aim to confirm the findings shown by three previous published studies so clinicians will be able to confidently improve patient care beyond current practice. It is also about earlier identification of the relatively small number of cancers, which will be tested and again may change practice and bring forward diagnosis of lung cancer to improve survival and mortality.”
The team at Optellum and Professor Baldwin will now work with 10 leading NHS hospitals to deploy the technology for clinical evaluation, taking the solution one-step closer to being widely deployed to benefit patients across the entire country as a new standard of care.
Dr Vaclav Potesil, co-founder and CEO of Optellum, said: “We are delighted to be working in partnership with ten hospitals and leading experts across the NHS to continue to develop our lung cancer prediction software. It is already in clinical use and benefitting patients at leading hospitals in the United States. The NHSX award will help us accelerate bringing the AI-driven early diagnosis and treatment of lung cancer to all NHS patients as soon as possible.”
Dr Indra Joshi, Director of AI, NHSX, said: “With this latest round of AI Award winners, we now have an incredible breadth of expertise across a wide range of clinical and operational areas. Through this award, X will be at the forefront of applying artificial intelligence in new ways to transform health and care.”
Other project partners are leading lung cancer experts across the NHS; Professor Fergus Gleeson of Oxford University Hospitals NHS Foundation Trust; Professor Matthew Callister and Professor Andrew Scarsbrook of Leeds Teaching Hospitals NHS Trust; Dr. Richard Lee (co-ordinating Royal Marsden NHS Foundation Trust and satellite sites); and Professor Sam Janes coordinating University College London Hospital NHS Foundation Trust and satellite sites.
Optellum (Oxford, UK) is a commercial-stage lung health company providing Artificial Intelligence decision support software that assists physicians in early diagnosis and optimal treatment for their patients. The company was founded so that every lung disease patient is diagnosed and treated at the earliest possible stage when chances of cure are the highest. www.optellum.com
NHSX is a joint unit of teams from the Department of Health and Social Care and NHS England and Improvement, driving forward the digital transformation of health and social care. www.nhsx.nhs.uk
The Accelerated Access Collaborative (AAC) is a unique partnership between patient groups, government bodies, industry and NHS bodies, working together to streamline the adoption of new innovations in healthcare. www.england.nhs.uk/aac
The National Institute for Health and Research (NIHR) provides the people, facilities, and technology that enable research to thrive. www.nihr.ac.uk
