Clinical Integration of Artificial Intelligence for Breast Imaging.

This article describes an approach to planning and implementing artificial intelligence products in a breast screening service. It highlights the importance of an in-depth understanding of the end-to-end workflow and effective project planning by a multidisciplinary team. It discusses the need for monitoring to ensure that performance is stable and meets expectations, as well as focusing on the potential for inadvertantly generating inequality. New cross-discipline roles and expertise will be needed to enhance service delivery.

Evidence-informed recommendations on managing breast screening atypia: perspectives from an expert panel consensus meeting reviewing results from the Sloane atypia project.

Evidence-based clinical guidelines are essential to maximize patient benefit and to reduce clinical uncertainty and inconsistency in clinical practice. Gaps in the evidence base can be addressed by data acquired in routine practice. At present, there is no international consensus on management of women diagnosed with atypical lesions in breast screening programmes. Here, we describe how routine NHS breast screening data collected by the Sloane atypia project was used to inform a management pathway that maximizes early detection of cancer and minimizes over-investigation of lesions with uncertain malignant potential. A half-day consensus meeting with 11 clinical experts, 1 representative from Independent Cancer Patients' Voice, 6 representatives from NHS England (NHSE) including from Commissioning, and 2 researchers was held to facilitate discussions of findings from an analysis of the Sloane atypia project. Key considerations of the expert group in terms of the management of women with screen detected atypia were: (1) frequency and purpose of follow-up; (2) communication to patients; (3) generalizability of study results; and (4) workforce challenges. The group concurred that the new evidence does not support annual surveillance mammography for women with atypia, irrespective of type of lesion, or woman's age. Continued data collection is paramount to monitor and audit the change in recommendations.

Development and quality appraisal of a new English breast screening linked data set as part of the age, test threshold, and frequency of mammography screening (ATHENA-M) study.

OBJECTIVES: To build a data set capturing the whole breast cancer screening journey from individual breast cancer screening records to outcomes and assess data quality. METHODS: Routine screening records (invitation, attendance, test results) from all 79 English NHS breast screening centres between January 1, 1988 and March 31, 2018 were linked to cancer registry (cancer characteristics and treatment) and national mortality data. Data quality was assessed using comparability, validity, timeliness, and completeness. RESULTS: Screening records were extracted from 76/79 English breast screening centres, 3/79 were not possible due to software issues. Data linkage was successful from 1997 after introduction of a universal identifier for women (NHS number). Prior to 1997 outcome data are incomplete due to linkage issues, reducing validity. Between January 1, 1997 and March 31, 2018, a total of 11 262 730 women were offered screening of whom 9 371 973 attended at least one appointment, with 139 million person-years of follow-up (a median of 12.4 person years for each woman included) with 73 810 breast cancer deaths and 1 111 139 any-cause deaths. Comparability to reference data sets and internal validity were demonstrated. Data completeness was high for core screening variables (>99%) and main cancer outcomes (>95%). CONCLUSIONS: The ATHENA-M project has created a large high-quality and representative data set of individual women's screening trajectories and outcomes in England from 1997 to 2018, data before 1997 are lower quality. ADVANCES IN KNOWLEDGE: This is the most complete data set of English breast screening records and outcomes constructed to date, which can be used to evaluate and optimize screening.

Are mammography image acquisition factors, compression pressure and paddle tilt, associated with breast cancer detection in screening?

OBJECTIVES: To assess the associations between objectively measured mammographic compression pressure and paddle tilt and breast cancer (BC) detected at the same ("contemporaneous") screen, subsequent screens, or in-between screens (interval cancers). METHODS: Automated pressure and paddle tilt estimates were derived for 80,495 mammographic examinations in a UK population-based screening programme. Adjusted logistic regression models were fitted to estimate the associations of compression parameters with BC detected at contemporaneous screen (777 cases).Nested case-control designs were used to estimate associations of pressure and tilt with: (a) interval cancer (148 cases/625 age-matched controls) and (b) subsequent screen-detected cancer (344/1436), via conditional logistic regression. RESULTS: Compression pressure was negatively associated with odds of BC at contemporaneous screen (odds ratio (OR) for top versus bottom third of the pressure distribution: 0.74; 95% CI 0.60, 0.92; P-for-linear-trend (Pt) = 0.007). There was weak evidence that moderate pressure at screening was associated with lower odds of interval cancer (OR for middle versus bottom third: 0.63; 95% CI 0.38, 1.05; p = 0.079), but no association was found between pressure and the odds of BC at subsequent screen. There was no evidence that paddle tilt was associated with the odds of contemporaneous, subsequent screen or interval cancer detection. CONCLUSIONS: Findings are consistent with compression pressure, but not paddle tilt, affecting the performance of mammographic screening by interfering with its ability to detect cancers. ADVANCES IN KNOWLEDGE: Inadequate or excessive compression pressure at screening may contribute to a reduced ability to detect cancers, resulting in a greater number of interval cancer cases.

Should we share breast density information during breast cancer screening in the United Kingdom? an integrative review.

OBJECTIVE: Dense breasts are an established risk factor for breast cancer and also reduce the sensitivity of mammograms. There is increasing public concern around breast density in the UK, with calls for this information to be shared at breast cancer screening. METHODS: We searched the PubMed database, Cochrane Library and grey literature, using broad search terms in October 2022. Two reviewers extracted data and assessed the risk of bias of each included study. The results were narratively synthesised by five research questions: desire for information, communication formats, psychological impact, knowledge impact and behaviour change. RESULTS: We identified 19 studies: three Randomised Controlled Trials (RCTs), three cohort studies, nine cross-sectional studies, one qualitative interview study, one mixed methods study and two 2021 systematic reviews. Nine studies were based in the United States of America (USA), five in Australia, two in the UK and one in Croatia. One systematic review included 14 USA studies, and the other 27 USA studies, 1 Australian and 1 Canadian. The overall GRADE evidence quality rating for each research question was very low to low.Generally, participants wanted to receive breast density information. Conversations with healthcare professionals were more valued and effective than letters. Breast density awareness after notification varied greatly between studies.Breast density information either did not impact frequency of mammography screening or increased the intentions of participants to return for routine screening as well as intention to access, and uptake of, supplementary screening. People from ethnic minority groups or of lower socioeconomic status (SES) had greater confusion following notification, and, along with those without healthcare insurance, were less likely to access supplementary screening. CONCLUSION: Breast density specific research in the UK, including different communities, is needed before the UK considers sharing breast density information at screening. There are also practical considerations around implementation and recording, which need to be addressed. ADVANCES IN KNOWLEDGE: Currently, sharing breast density information at breast cancer screening in the UK may not be beneficial to participants and could widen inequalities. UK specific research is needed, and measurement, communication and future testing implications need to be carefully considered.

Defining the road map to a UK national lung cancer screening programme.

Lung cancer screening with low-dose CT was recommended by the UK National Screening Committee (UKNSC) in September, 2022, on the basis of data from trials showing a reduction in lung cancer mortality. These trials provide sufficient evidence to show clinical efficacy, but further work is needed to prove deliverability in preparation for a national roll-out of the first major targeted screening programme. The UK has been world leading in addressing logistical issues with lung cancer screening through clinical trials, implementation pilots, and the National Health Service (NHS) England Targeted Lung Health Check Programme. In this Policy Review, we describe the consensus reached by a multiprofessional group of experts in lung cancer screening on the key requirements and priorities for effective implementation of a programme. We summarise the output from a round-table meeting of clinicians, behavioural scientists, stakeholder organisations, and representatives from NHS England, the UKNSC, and the four UK nations. This Policy Review will be an important tool in the ongoing expansion and evolution of an already successful programme, and provides a summary of UK expert opinion for consideration by those organising and delivering lung cancer screenings in other countries.

Quantitative background parenchymal enhancement and fibro-glandular density at breast MRI: Association with BRCA status.

OBJECTIVES: To investigate whether MRI-based measurements of fibro-glandular tissue volume, breast density (MRBD), and background parenchymal enhancement (BPE) could be used to stratify two cohorts of healthy women: BRCA carriers and women at population risk of breast cancer. METHODS: Pre-menopausal women aged 40-50 years old were scanned at 3 T, employing a standard breast protocol including a DCE-MRI (35 and 30 participants in high- and low-risk groups, respectively). The dynamic range of the DCE protocol was characterised and both breasts were masked and segmented with minimal user input to produce measurements of fibro-glandular tissue volume, MRBD, and voxelwise BPE. Statistical tests were performed to determine inter- and intra-user repeatability, evaluate the symmetry between metrics derived from left and right breasts, and investigate MRBD and BPE differences between the high- and low-risk cohorts. RESULTS: Intra- and inter-user reproducibility in estimates of fibro-glandular tissue volume, MRBD, and median BPE estimations were good, with coefficients of variation < 15%. Coefficients of variation between left and right breasts were also low (< 25%). There were no significant correlations between fibro-glandular tissue volume, MRBD, and BPE for either risk group. However, the high-risk group had higher BPE kurtosis, although linear regression analysis did not reveal significant associations between BPE kurtosis and breast cancer risk. CONCLUSIONS: This study found no significant differences or correlations in fibro-glandular tissue volume, MRBD, or BPE metrics between the two groups of women with different levels of breast cancer risk. However, the results support further investigation into the heterogeneity of parenchymal enhancement. KEY POINTS: • A semi-automated method enabled quantitative measurements of fibro-glandular tissue volume, breast density, and background parenchymal enhancement with minimal user intervention. • Background parenchymal enhancement was quantified over the entire parenchyma, segmented in pre-contrast images, thus avoiding region selection. • No significant differences and correlations in fibro-glandular tissue volume, breast density, and breast background parenchymal enhancement were found between two cohorts of women at high and low levels of breast cancer risk.

To what extent are objectively measured mammographic imaging techniques associated with compression outcomes.

OBJECTIVE: To describe the association between objectively measurable imaging techniques and the resulting compression thickness and dose. METHODS: The study included 80,495 routine screens from the South-West London Breast Screening Service between March 2013 and July 2017. Average compression force, paddle tilt and dose were calculated. The Volpara® DensityTM algorithm was used to estimate pressure, breast volume and density.Linear regression models, using generalized estimating equations (GEEs) to account for clustering by practitioner, assessed the strength of the associations between the imaging compression outcomes, (thickness, dose) and imaging techniques (force, pressure and paddle tilt), adjusting for the subject's characteristics (age, ethnicity, breast volume and percent mammographic density). RESULTS: Fully adjusted linear regression models showed that compression thickness decreased by ~1 mm (~2% of mean thickness) for every 1daN increase in force and decreased by ~0.8 mm with an increase of 1 kPa of pressure (at median pressure). Increasing pressure above 15 kPa resulted in minimal reduction in thickness. Dose increased with increased force but decreased by ~1% of mean dose with every increase in 1 kPa of pressure. For 1o increase in paddle tilt, the compression thickness increased by ~1.5 mm (~2.5%) and dose increased by ~2.5%, (Pt <0.001 in all cases). CONCLUSION: Differences in imaging technique are associated with imaging outcome measures (thickness and dose). A better understanding of the association between objective image acquisition parameters and tumour conspicuity could lead to clearer guidelines for practitioners. ADVANCES IN KNOWLEDGE: Increased paddle tilt is associated with increased compression thickness and increased dose after adjustment for breast volume and force applied.

A national Delphi study: experiences of general practitioner (GP) trainees in the Republic of Ireland undertaking online small group learning as part of the GP Training Programme during COVID-19.

INTRODUCTION: During the COVID-19 pandemic, GP training day release was redirected from face-to-face to an online setting. With this study, our aim was to assess trainee experiences of online small group learning and to make recommendations with regards to future GP training. METHODS: A qualitative study using the Delphi survey technique, approved by the Irish College of General Practitioners (ICGP) Ethics Committee. A series of three sequential online questionnaires were sent to our trainee cohort in all 14 training schemes in Ireland. The first questionnaire explored GP trainee experiences and key themes were generated. Subsequent questionnaires were developed using these themes, with second and third round questionnaires establishing consensus on these experiences. RESULTS: In total, 64 GP trainees responded. Each training scheme was represented. Response rates for round 1 and 2 were 76% and 56% respectively, with round 3 currently underway. Trainees felt that online teaching was convenient, reduced commuting costs, and provided peer support. They also reported loss in unstructured discussion, practical teaching sessions and relationship building. Seven key themes were generated: future format of GP training; accessibility and flexibility; teaching experience; provision of GP training; support and collegiality; educational experience; and technical problems. There is a consensus that some online teaching should be retained for the future. DISCUSSION: Online teaching provided a continuation in training that was more convenient and accessible but affected social interactions and relationship building amongst trainees. Future online sessions could be utilised in a hybrid model of teaching going forward.

Lessons learned from independent external validation of an AI tool to detect breast cancer using a representative UK data set.

OBJECTIVE: To pilot a process for the independent external validation of an artificial intelligence (AI) tool to detect breast cancer using data from the NHS breast screening programme (NHSBSP). METHODS: A representative data set of mammography images from 26,000 women attending 2 NHS screening centres, and an enriched data set of 2054 positive cases were used from the OPTIMAM image database. The use case of the AI tool was the replacement of the first or second human reader. The performance of the AI tool was compared to that of human readers in the NHSBSP. RESULTS: Recommendations for future external validations of AI tools to detect breast cancer are provided. The tool recalled different breast cancers to the human readers. This study showed the importance of testing AI tools on all types of cases (including non-standard) and the clarity of any warning messages. The acceptable difference in sensitivity and specificity between the AI tool and human readers should be determined. Any information vital for the clinical application should be a required output for the AI tool. It is recommended that the interaction of radiologists with the AI tool, and the effect of the AI tool on arbitration be investigated prior to clinical use. CONCLUSION: This pilot demonstrated several lessons for future independent external validation of AI tools for breast cancer detection. ADVANCES IN KNOWLEDGE: Knowledge has been gained towards best practice procedures for performing independent external validations of AI tools for the detection of breast cancer using data from the NHS Breast Screening Programme.

Understanding and supporting peer relationships in adolescents with acquired brain injury: A stakeholder engagement study.

Peer relationship difficulties in adolescents with acquired brain injury (ABI) are under-recognized and targets for intervention are unclear. From a social constructionist position, this study aimed to engage with stakeholders to develop a collaborative understanding of peer relationship difficulties in adolescents with ABI and seek consultation on what might be required to improve them. Focus groups and semi-structured interviews were conducted with four stakeholder groups: adolescents with ABI (n = 4); parents of adolescents with ABI (n = 7); adults who sustained an ABI in adolescence (n = 2); and specialist practitioners (n = 3). Qualitative data were analysed using thematic analysis. The analysis yielded 11 themes, grouped into two domains. The first, understanding peer relationship difficulties, included themes from "exclusion and a need to belong", to "loss of past self". The second, supporting peer relationships, comprised themes of "building understanding" and "meaningful social connection", amongst others. A logic model of stakeholder experiences of peer relationship difficulties was constructed. Difficulties with peers can increase vulnerability to feelings of loneliness, shame, and hopelessness for adolescents post-ABI. Stakeholders described that a meaningful intervention would be multi-layered, targeting change within the adolescent's environment and within the adolescent themselves. The presented logic model provides a framework for future intervention development.

Recommendations for the development and use of imaging test sets to investigate the test performance of artificial intelligence in health screening.

Rigorous evaluation of artificial intelligence (AI) systems for image classification is essential before deployment into health-care settings, such as screening programmes, so that adoption is effective and safe. A key step in the evaluation process is the external validation of diagnostic performance using a test set of images. We conducted a rapid literature review on methods to develop test sets, published from 2012 to 2020, in English. Using thematic analysis, we mapped themes and coded the principles using the Population, Intervention, and Comparator or Reference standard, Outcome, and Study design framework. A group of screening and AI experts assessed the evidence-based principles for completeness and provided further considerations. From the final 15 principles recommended here, five affect population, one intervention, two comparator, one reference standard, and one both reference standard and comparator. Finally, four are appliable to outcome and one to study design. Principles from the literature were useful to address biases from AI; however, they did not account for screening specific biases, which we now incorporate. The principles set out here should be used to support the development and use of test sets for studies that assess the accuracy of AI within screening programmes, to ensure they are fit for purpose and minimise bias.

UK National Screening Committee's approach to reviewing evidence on artificial intelligence in breast cancer screening.

Artificial intelligence (AI) could have the potential to accurately classify mammograms according to the presence or absence of radiological signs of breast cancer, replacing or supplementing human readers (radiologists). The UK National Screening Committee's assessments of the use of AI systems to examine screening mammograms continues to focus on maximising benefits and minimising harms to women screened, when deciding whether to recommend the implementation of AI into the Breast Screening Programme in the UK. Maintaining or improving programme specificity is important to minimise anxiety from false positive results. When considering cancer detection, AI test sensitivity alone is not sufficiently informative, and additional information on the spectrum of disease detected and interval cancers is crucial to better understand the benefits and harms of screening. Although large retrospective studies might provide useful evidence by directly comparing test accuracy and spectrum of disease detected between different AI systems and by population subgroup, most retrospective studies are biased due to differential verification (ie, the use of different reference standards to verify the target condition among study participants). Enriched, multiple-reader, multiple-case, test set laboratory studies are also biased due to the laboratory effect (ie, radiologists' performance in retrospective, laboratory, observer studies is substantially different to their performance in a clinical environment). Therefore, assessment of the effect of incorporating any AI system into the breast screening pathway in prospective studies is required as it will provide key evidence for the effect of the interaction of medical staff with AI, and the impact on women's outcomes.

Virtual clinical trial to compare cancer detection using combinations of 2D mammography, digital breast tomosynthesis and synthetic 2D imaging.

OBJECTIVES: This study was designed to compare the detection of subtle lesions (calcification clusters or masses) when using the combination of digital breast tomosynthesis (DBT) and synthetic mammography (SM) with digital mammography (DM) alone or combined with DBT. METHODS: A set of 166 cases without cancer was acquired on a DBT mammography system. Realistic subtle calcification clusters and masses in the DM images and DBT planes were digitally inserted into 104 of the acquired cases. Three study arms were created: DM alone, DM with DBT and SM with DBT. Five mammographic readers located the centre of any lesion within the images that should be recalled for further investigation and graded their suspiciousness. A JAFROC figure of merit (FoM) and lesion detection fraction (LDF) were calculated for each study arm. The visibility of the lesions in the DBT images was compared with SM and DM images. RESULTS: For calcification clusters, there were no significant differences (p > 0.075) in FoM or LDF. For masses, the FoM and LDF were significantly improved in the arms using DBT compared to DM alone (p < 0.001). On average, both calcification clusters and masses were more visible on DBT than on DM and SM images. CONCLUSIONS: This study demonstrated that masses were detected better with DBT than with DM alone and there was no significant difference (p = 0.075) in LDF between DM&DBT and SM&DBT for calcifications clusters. Our results support previous studies that it may be acceptable to not acquire digital mammography alongside tomosynthesis for subtle calcification clusters and ill-defined masses. KEY POINTS: • The detection of masses was significantly better using DBT than with digital mammography alone. • The detection of calcification clusters was not significantly different between digital mammography and synthetic 2D images combined with tomosynthesis. • Our results support previous studies that it may be acceptable to not acquire digital mammography alongside tomosynthesis for subtle calcification clusters and ill-defined masses for the imaging technology used.

Understanding breast cancer as a global health concern.

Breast cancer is now the most commonly diagnosed cancer in the world. The most recent global cancer burden figures estimate that there were 2.26 million incident breast cancer cases in 2020 and the disease is the leading cause of cancer mortality in women worldwide. The incidence is strongly correlated with human development, with a large rise in cases anticipated in regions of the world that are currently undergoing economic transformation. Survival, however, is far less favourable in less developed regions. There are a multitude of factors behind disparities in the global survival rates, including delays in diagnosis and lack of access to effective treatment. The World Health Organization's new Global Breast Cancer Initiative was launched this year to address this urgent global health challenge. It aims to improve survival across the world through three pillars: health promotion, timely diagnosis, and comprehensive treatment and supportive care. In this article, we discuss the key challenges of breast cancer care and control in a global context.

Quantitative breast density analysis to predict interval and node-positive cancers in pursuit of improved screening protocols: a case-control study.

BACKGROUND: This study investigates whether quantitative breast density (BD) serves as an imaging biomarker for more intensive breast cancer screening by predicting interval, and node-positive cancers. METHODS: This case-control study of 1204 women aged 47-73 includes 599 cancer cases (302 screen-detected, 297 interval; 239 node-positive, 360 node-negative) and 605 controls. Automated BD software calculated fibroglandular volume (FGV), volumetric breast density (VBD) and density grade (DG). A radiologist assessed BD using a visual analogue scale (VAS) from 0 to 100. Logistic regression and area under the receiver operating characteristic curves (AUC) determined whether BD could predict mode of detection (screen-detected or interval); node-negative cancers; node-positive cancers, and all cancers vs. controls. RESULTS: FGV, VBD, VAS, and DG all discriminated interval cancers (all p < 0.01) from controls. Only FGV-quartile discriminated screen-detected cancers (p < 0.01). Based on AUC, FGV discriminated all cancer types better than VBD or VAS. FGV showed a significantly greater discrimination of interval cancers, AUC = 0.65, than of screen-detected cancers, AUC = 0.61 (p < 0.01) as did VBD (0.63 and 0.53, respectively, p < 0.001). CONCLUSION: FGV, VBD, VAS and DG discriminate interval cancers from controls, reflecting some masking risk. Only FGV discriminates screen-detected cancers perhaps adding a unique component of breast cancer risk.

OPTIMAM Mammography Image Database: A Large-Scale Resource of Mammography Images and Clinical Data.

Supplemental material is available for this article.

Digital breast tomosynthesis: sensitivity for cancer in younger symptomatic women.

OBJECTIVE: Full-field digital mammography (FFDM) has limited sensitivity for cancer in younger women with denser breasts. Digital breast tomosynthesis (DBT) can reduce the risk of cancer being obscured by overlying tissue. The primary study aim was to compare the sensitivity of FFDM, DBT and FFDM-plus-DBT in women under 60 years old with clinical suspicion of breast cancer. METHODS: This multicentre study recruited 446 patients from UK breast clinics. Participants underwent both standard FFDM and DBT. A blinded retrospective multireader study involving 12 readers and 300 mammograms (152 malignant and 148 benign cases) was conducted. RESULTS: Sensitivity for cancer was 86.6% with FFDM [95% CI (85.2-88.0%)], 89.1% with DBT [95% CI (88.2-90%)], and 91.7% with FFDM+DBT [95% CI (90.7-92.6%)]. In the densest breasts, the maximum sensitivity increment with FFDM +DBT over FFDM alone was 10.3%, varying by density measurement method. Overall specificity was 81.4% with FFDM [95% CI (80.5-82.3%)], 84.6% with DBT [95% CI (83.9-85.3%)], and 79.6% with FFDM +DBT [95% CI (79.0-80.2%)]. No differences were detected in accuracy of tumour measurement in unifocal cases. CONCLUSIONS: Where available, DBT merits first-line use in the under 60 age group in symptomatic breast clinics, particularly in women known to have very dense breasts. ADVANCES IN KNOWLEDGE: This study is one of very few to address the accuracy of DBT in symptomatic rather than screening patients. It quantifies the diagnostic gains of DBT in direct comparison with standard digital mammography, supporting informed decisions on appropriate use of DBT in this population.