iCAIRD & Kheiron mammography AI MIA evaluation project
iCAIRD are working with SME partner Kheiron Medical Technologies on an exemplar project evaluating performance of their mammography screening platform MIA for potential use in clinical breast screening services. MIA, or Mammography Intelligent Assessment, is an AI tool which has been developed by Kheiron to support radiologists in breast screening. It does this by acting as a 2nd reader for reporting, scanning mammograms to detect any signs of breast cancer and flagging up anything which looks suspicious/abnormal to the radiologist. This enables radiologists to make clinical decisions accurately and rapidly.
Kheiron machine learning specialists are working together with NHS Grampian clinicians & health data scientists from the University of Aberdeen to train and evaluate MIA for potential deployment in clinical practices for breast cancer screening across NHS Scotland. Currently the Scottish Breast Screening Programme (SBSP) carries out around 600,000 mammogram examinations in any 3-year cycle. The intention is that using MIA in breast screening will help radiologists manage this workload, leading to faster screening & treatment turnaround and improved patient outcomes.
Click on the image opp to learn more about MIA
Researching Patient & Clinician attitudes to potential AI use in breast screening
Part of iCAIRDs’ current research in this area includes conducting a study of people’s views on using artificial intelligence systems for breast cancer screening. This research has 2 key areas: one is focussing on screening clinic attendees whilst the other is looking at healthcare service practitioners and providers. Click on the links below to learn more about each of these.
Attitudes to AI Use in Breast Screening
As part of the iCAIRD WP5 programme, we have created a short questionnaire to find out what women who attend for routine breast screening in NHS Grampian think about the use of AI in breast screening. They were presented our questionnaire which can be completed in a few minutes.
Currently two experts examine each mammogram. If the two experts disagree on whether a woman is invited back for additional examinations, a third expert decides. Through these additional exams, it is determined whether a woman has breast cancer.
The questionnaire presented several potential options for the use of AI in breast screening:
1. AI replaces one of the two experts
2. AI replaces all human experts
3. AI examines all mammograms. Difficult images are sent to a human expert for review
4. All human experts have access to AI to help them reach a decision
Participants were asked to indicate whether they approve or disapprove of each option. 187 women have completed the questionnaire. These are the results:
Women on average approved of the scenarios posed in questions 1, 3 and 4. They approved most of scenario 1 (one human expert and one AI examine your mammogram), and scenario 4 (all human experts have access to an AI to help make their decision). To a lesser extent, they approved of scenario 3: only difficult images are seen by a human expert.
Women were most divided on scenario 2: only AI reads your mammograms.
We also observed a relationship between self-reported understanding of AI and women’s response to questions 1 to 5: women who believed they had a good understanding of AI were more likely to approve of all scenarios presented.
by Dr Clarisse de Vries, Clinical Research Fellow, University of Aberdeen
PREPare for the real-world evaluation of an Artificial Intelligent system for breast cancer scReening: identify and assess stakeholders’ needs (PREP-AIR)
Breast cancer is increasingly prevalent with 1 in 8 women in Scotland being diagnosed with this condition at some point in their lives. To facilitate its early detection the Scottish Government has created a Breast Screening Programme (BSP), which periodically invites women in Scotland aged between 50 and 70 years old for a mammography, a technique that uses X-rays to examine a breast allowing early detection of breast cancer.
Currently, this process is overseen by two separate highly trained radiologists, who analyse the breast images and give a diagnosis, whilst ensuring that the cancer standards established by the government are adhered to. With demand and availability of diagnostic images rapidly exceeding the capacity of available radiologists, developing technological innovations to aid this process is a priority for Scotland.
Artificial intelligent (AI) systems show considerable promise for improving accuracy and speed of diagnosis. For this system to be deployed to aid the early detection of breast cancer in Scotland, its utility needs to be evaluated in medical practice. To determine how to best evaluate it, we propose to explore ‘wants’ and ‘needs’ of members of groups/organisations that have interest in, the potential to influence or be affected by potential deployment of this AI system as a part of the BSP in Scotland.
This research is using qualitative online sources analysis, telephone interviews, focus groups and workshops to identify and assess needs of stakeholders, for example, women of breast cancer screening age, patient representatives from cancer charities, NHS non-clinical and NHS clinical staff.
The overall aim of this study is to inform strategies for future assessment of AI systems to detect breast cancer in real world settings in Scotland. The study is currently ongoing with results to be published on its completion.
by Dr Rumana Newlands, Research Fellow, University of Aberdeen
Clinician Perspective
Dr Gerald Lip, clinical lead on iCAIRDs’ mammography project shares his thoughts: “As someone working specifically in breast cancer screening, I’m excited by the possible opportunities that AI is presenting. Although still in evaluation stages we are already seeing where benefits such as reduction in waiting times and improving accuracy in diagnosis can potentially come from adopting this technology.”
“AI will not replace medical professionals, but by working with technology, we will be able to ensure patients are receiving the most accurate diagnosis and care possible.”
Dr Gerald LipClinical Director, North East of Scotland Breast Screening Programme and Consultant Radiologist, NHS Grampian said
The Patient Perspective
Dr Liz O’Riordan shares her experience as a consultant breast surgeon and a breast cancer patient.
“31 Wishes”
Our partner Kheiron Medicals’ 2021 Breast Cancer Awareness activity this year focuses on their “31 Wishes” feature, sharing a story each day during October of 31 people who have been affected by cancer, as they share their experience and their wish for breast cancer awareness. Click on the image to read more:-
Further information /learning & support resources for breast cancer:
NHS Inform – Breast Screening In Scotland
Cancer Research UK – About Breast Cancer
If you would like to donate or join in fundraising for Wear It Pink 2021 further details are available on the Breast Cancer Now charity site at www.wearitpink.org
