Active data enrichment by learning what to annotate in digital pathology
Abstract: Our work aims to link pathology with radiology with the goal to improve the early detection of lung cancer. Rather than utilising a set of predefined radiomics features, we propose to learn a new set of features from histology. Generating a comprehensive lung histology report is the first vital step towards this goal. Deep learning has revolutionised the computational assessment of digital pathology images. Today, we have mature algorithms for assessing morphological features at the cellular and tissue levels. In addition, there are promising efforts that link morphological features with biologically relevant information. While promising, these efforts mostly focus on narrow well-defined questions. Developing a comprehensive report that is required in our setting requires an annotation strategy that captures all clinically relevant patterns specified in the WHO guidelines. Here, we propose and compare approaches aimed to balance the dataset and mitigate the biases in learning by automatically prioritising regions with clinical patterns underrepresented in the dataset. Our study demonstrates the opportunities active data enrichment can provide and results in a new lung-cancer dataset annotated to a degree that is not readily available in public domain.
Comparing imaging techniques to diagnose lung cancer – looking for advisors
Are you interested in helping research to improve accuracy in earlier diagnosis of lung cancer from different imaging types?
We’re looking for people who’ve had imaging for suspected lung cancer whether or not they’ve gone on to be diagnosed, as well as those who’ve been diagnosed and treated.
What is it for?
A group led by the University of Oxford is planning a research study exploring whether a CT scan can diagnose lung cancer earlier than the chest x-ray currently performed when lung cancer is suspected. As part of this research, the team is inviting people to join a patient and public involvement (PPI) advisory group.
What is the purpose of the PPI advisory group?
The group’s purpose is to ensure the experiences and views of those referred for chest imaging for suspected lung cancer, and those diagnosed with lung cancer or their family members/carers are taken into account when planning and delivering this study.
What would PPI advisory group members be expected to do?
PPI Group members will be asked to attend up to 3 workshops, review documents being designed for study participants, join a study steering committee and help share the final study results. You can choose to take part in all or some of these activities.
Timeline/tenure
The first workshop is expected to be held in June/July 2023 during the preparation of the final funding application to the National Institute for Health and Care Research (NIHR).
If funded, the study will start early 2024 and there will be an opportunity to continue participation in the PPI advisory group at that point.
Remuneration
You will receive payment for any time spent in these activities and travel expenses for any face to face meetings, although most of the work will be completed remotely.
If you are interested and would like more information please contact dart@oncology.ox.ac.uk
Newcastle upon Tyne Hospitals NHS Foundation Trust
Royal Brompton & Harefield Clinical Group, Part of Guy’s and St Thomas’ NHS Foundation Trust
University Hospitals of North Midlands NHS Trust
We have received:
92,454 clinical records
902 CTs
81 patient’s pathology slides which have been scanned
SCOOT, DART’s companion project, has collected 94 blood samples.
Presentation at OxCODE Symposium
Dr Weiqi Liao presented a Lightning Talk at the OxCODE (Oxford Centre for Early Cancer Detection) symposium 2022 on Tuesday 13th September 2022 at Worcester College. His talk was entitled “Predicting the future risk of lung cancer: Development and validation of QCancer2 (10-year risk) lung model and evaluating the performance of nine prediction models”
Congratulations to those sites which have been able to begin data transfer to DART. We have received over 38,000 records already and look forward to more sites being able to contribute shortly.
DART scientists will use the data gathered by the Targeted Lung Health Checks during patient screening and care to develop:
Enhanced information about patient outcomes
New digital pathology AI and AI derived radiomics for diagnosis and stratification of patients
Algorithms to better evaluate risks from comorbidities such as chronic obstructive pulmonary disease (COPD)
New insights about the right at-risk individuals to be selected for screening, using linked data from primary care
Collectively this will define a new set of standards for lung cancer screening, to increase the number of lung cancers diagnosed earlier, and therefore treated more successfully and with fewer invasive clinical procedures.
Data collected includes:
Clinical data
CT scans (LDCT and PET-CT) and
Digitised images of stained tissue sections (digital pathology)
Celebrate Doncaster LHC’s success
The Doncaster programme has been operational since mid-March 2021 and in the first year 11,857 lung health check calls and more than 5,000 initial scans have taken place; 50 lung cancers have been found, 73% of which were early-stage lung cancers. Seven other cancers have also been confirmed and 41 patients (75%) have been given life-saving cancer treatment so far.
Read more about this pioneering lung screening trial that is saving lives in Doncaster here
Optellum attains CE marking in Europe
DART partner Optellum has attained CE marking in Europe.
This latest certification will allow for use in the European Union (EU) and the United Kingdom (UK), and opens the door to a European expansion for the growing company. It is the latest milestone for Optellum, which received FDA 510(k) clearance in early 2021 as the first AI-assisted diagnosis application for lung cancer. You can read the full press release online here.
In addition, researchers from the University of Oxford and Oxford University Hospitals NHS Foundation Trust (OUH) have published the results of an academic study in European Radiology which takes a closer look at the Optellum Lung Cancer Prediction (LCP) model. You can read the full story on Optellum’s website here.
Approvals and adoption
NIHR
DART is very pleased that the project has been adopted onto the NIHR portfolio and is therefore deemed eligible for NIHR Clinical Research Network support. Further information about CRN support can be found on the NIHR website.
The 15 Local Clinical Research Networks (LCRN) cover the length and breadth of England and are available to coordinate and support the delivery of research across the NHS in England.
Health Research Authority HRA
DART has had confirmation that HRA and Health and Care Research Wales (HCRW) Approval has been given for the study, on the basis described in the application form, protocol, supporting documentation and clarifications requested and received.
DART has also had notification of Confidentiality Advisory Group (CAG) conditional support, as per the excerpt from their letter below.
“The role of the Confidentiality Advisory Group (CAG) is to review applications submitted under these Regulations and to provide advice to the Health Research Authority on whether application activity should be supported, and if so, any relevant conditions. This application was considered at the CAG meeting held on 10 February 2022.
Health Research Authority decision
The Health Research Authority, having considered the advice from the Confidentiality Advisory Group as set out below, has determined the following:
The application, to allow the disclosure of confidential patient information from participating trusts to the Oxford University Hospitals NHS Foundation Trust, is conditionally supported, subject to compliance with the standard and specific conditions of support.
Please note that the legal basis to allow access to the specified confidential patient information without consent is now in effect.”
The conditions of support allow for the project to proceed as planned and will be addressed by mid-May 2022. DART will comply with the HRA annual review.
CAG reference: 22/CAG/0010
IRAS project ID: 301420
REC reference: 21/WM/0278
GE Healthcare and Optellum join forces to advance Lung Cancer Diagnosis with Artificial Intelligence
DART partners GE Healthcare and Optellum announced that they have signed a letter of intent to collaborate to advance precision diagnosis and treatment of lung cancer. GE Healthcare is a global leader in medical imaging solutions. Optellum is the leader in AI decision support for the early diagnosis and optimal treatment of lung cancer.
A clinician’s AI-assisted diagnosis of malignancy may enable patients whose nodules are not malignant to avoid unnecessary and aggressive procedures such as biopsy and surgical resection, while expediting the diagnostic process, and enabling the right treatment to start earlier. This has the potential to provide patients with personalized diagnosis and treatment plans, enabling lung cancer patients to be treated at the earliest possible stage when survival rates are the highest. This is also the ambition of DART.
NCIMI: Seven Strategies for Success – for adoption and deployment of Artificial Intelligence in Medical Diagnostics
NCIMI, working with The Behavioural Architects have created a guide to support their ambition of driving greater adoption of AI within the healthcare system, through the application of behavioural science.
