Fairness in Cardiac Magnetic Resonance Imaging: Assessing Sex and Racial Bias in Deep Learning-Based Segmentation

Addressing fairness issues in deep learning-based medical image analysis: a systematic review

Deep learning algorithms have demonstrated remarkable efficacy in various medical image analysis (MedIA) applications. However, recent research highlights a performance disparity in these algorithms when applied to specific subgroups, such as exhibiting poorer predictive performance in elderly females. Addressing this fairness issue has become a collaborative effort involving AI scientists and clinicians seeking to understand its origins and develop solutions for mitigation within MedIA. In this survey, we thoroughly examine the current advancements in addressing fairness issues in MedIA, focusing on methodological approaches. We introduce the basics of group fairness and subsequently categorize studies on fair MedIA into fairness evaluation and unfairness mitigation. Detailed methods employed in these studies are presented too. Our survey concludes with a discussion of existing challenges and opportunities in establishing a fair MedIA and healthcare system. By offering this comprehensive review, we aim to foster a shared understanding of fairness among AI researchers and clinicians, enhance the development of unfairness mitigation methods, and contribute to the creation of an equitable MedIA society.

Advancing Fairness in Cardiac Care: Strategies for Mitigating Bias in Artificial Intelligence Models Within Cardiology

In the dynamic field of medical artificial intelligence (AI), cardiology stands out as a key area for its technological advancements and clinical application. In this review we explore the complex issue of data bias, specifically addressing those encountered during the development and implementation of AI tools in cardiology. We dissect the origins and effects of these biases, which challenge their reliability and widespread applicability in health care. Using a case study, we highlight the complexities involved in addressing these biases from a clinical viewpoint. The goal of this review is to equip researchers and clinicians with the practical knowledge needed to identify, understand, and mitigate these biases, advocating for the creation of AI solutions that are not just technologically sound, but also fair and effective for all patients.

Applying Artificial Intelligence for Phenotyping of Inherited Arrhythmia Syndromes

Inherited arrhythmia disorders account for a significant proportion of sudden cardiac death, particularly among young individuals. Recent advances in our understanding of these syndromes have improved patient diagnosis and care, yet certain clinical gaps remain, particularly within case ascertainment, access to genetic testing, and risk stratification. Artificial intelligence (AI), specifically machine learning and its subset deep learning, present promising solutions to these challenges. The capacity of AI to process vast amounts of patient data and identify disease patterns differentiates them from traditional methods, which are time- and resource-intensive. To date, AI models have shown immense potential in condition detection (including asymptomatic/concealed disease) and genotype and phenotype identification, exceeding expert cardiologists in these tasks. Additionally, they have exhibited applicability for general population screening, improving case ascertainment in a set of conditions that are often asymptomatic such as left ventricular dysfunction. Third, models have shown the ability to improve testing protocols; through model identification of disease and genotype, specific clinical testing (eg, drug challenges or further diagnostic imaging) can be avoided, reducing health care expenses, speeding diagnosis, and possibly allowing for more incremental or targeted genetic testing approaches. These significant benefits warrant continued investigation of AI, particularly regarding the development and implementation of clinically applicable screening tools. In this review we summarize key developments in AI, including studies in long QT syndrome, Brugada syndrome, hypertrophic cardiomyopathy, and arrhythmogenic cardiomyopathies, and provide direction for effective future AI implementation in clinical practice.

Application of Artificial Intelligence in Cardiology: A Bibliometric Analysis

Recent advancements in artificial intelligence (AI) applications in medicine have been significant over the past 30 years. To monitor current research developments, it is crucial to examine the latest trends in AI adoption across various medical fields. This bibliometric analysis focuses on AI applications in cardiology. Unlike existing literature reviews, this study specifically examines journal articles published in the last decade, sourced from both Scopus and Web of Science databases, to illustrate the recent trends in AI within cardiology. The bibliometric analysis involves a statistical and quantitative evaluation of the literature on AI application in cardiovascular medicine over a defined period. A comprehensive global literature review is conducted to identify key research areas, authors, and their interrelationships through published works. The leading institutions and most influential authors in research on the role of AI in cardiology were located in the United States, the United Kingdom, and China. This study also provides researchers with an overview of the evolution of research in AI and cardiology. The main contribution of this study is to highlight the prominent authors, countries, journals, institutions, keywords, and trends in the development of AI in cardiology.

Assessment of the Prognostic Value of MRI Left Ventricular Global Function Index (LVGFI) in Patients With End-Stage Renal Disease Under Maintenance Dialysis

BACKGROUND

Left ventricular global function index (LVGFI) integrates LV volumetric and functional parameters. In patients with end-stage renal disease (ESRD), cardiac injury manifests as LV hypertrophy and dysfunction. However, the prognostic value of LVGFI in this population remains unclear.

PURPOSE

To investigate the association of LVGFI with major adverse cardiac events (MACE) in patients with ESRD.

STUDY TYPE

Prospective.

POPULATION

One hundred fifty-eight ESRD patients (mean age: 54.1 ± 14.4 years; 105 male) on maintenance dialysis.

FILED STRENGTH/SEQUENCE

3.0 T, balanced steady-state free precession (bSSFP) cine and modified Look-Locker inversion recovery (MOLLI) sequences.

ASSESSMENT

LV volumetric and functional parameters were determined from bSSFP images. LVGFI was calculated as the ratio of stroke volume to global volume and native T1 was determined from MOLLI T1 maps. MACE was recorded on follow up. Models were developed to predict MACE from conventional risk factors combined with LVGFI, GLS, native T1, and LV mass index (LVMI), respectively. Subgroup analyses were further performed in participants with LVEF above median.

STATISTICAL TESTS

Cox proportional hazard regression and log-rank test were used to investigate the association between LVGFI and MACE. The predictive models were evaluated and compared using Harrell's C-statistics and DeLong tests. A P value <0.05 was considered statistically significant.

RESULTS

Thirty-four MACE occurred during the median follow-up period of 26 months. The hazard of MACE increased by 114% for each 10% decrease in LVGFI in univariable analysis. The predictive model consisting of LVGFI (C-statistic: 0.724) had significantly better predictive performance than the others (all P < 0.001). These results were consistent in patients (N = 79) with LVEF > median (63.54%).

DATA CONCLUSION

LVGFI is a novel marker for MACE risk stratification in patients with ESRD and was better able to predict MACE than native T1 mapping and GLS.

EVIDENCE LEVEL

2 TECHNICAL EFFICACY: Stage 3.

A Framework for Interpretability in Machine Learning for Medical Imaging

Interpretability for machine learning models in medical imaging (MLMI) is an important direction of research. However, there is a general sense of murkiness in what interpretability means. Why does the need for interpretability in MLMI arise? What goals does one actually seek to address when interpretability is needed? To answer these questions, we identify a need to formalize the goals and elements of interpretability in MLMI. By reasoning about real-world tasks and goals common in both medical image analysis and its intersection with machine learning, we identify five core elements of interpretability: localization, visual recognizability, physical attribution, model transparency, and actionability. From this, we arrive at a framework for interpretability in MLMI, which serves as a step-by-step guide to approaching interpretability in this context. Overall, this paper formalizes interpretability needs in the context of medical imaging, and our applied perspective clarifies concrete MLMI-specific goals and considerations in order to guide method design and improve real-world usage. Our goal is to provide practical and didactic information for model designers and practitioners, inspire developers of models in the medical imaging field to reason more deeply about what interpretability is achieving, and suggest future directions of interpretability research.

Demographic bias in misdiagnosis by computational pathology models

Despite increasing numbers of regulatory approvals, deep learning-based computational pathology systems often overlook the impact of demographic factors on performance, potentially leading to biases. This concern is all the more important as computational pathology has leveraged large public datasets that underrepresent certain demographic groups. Using publicly available data from The Cancer Genome Atlas and the EBRAINS brain tumor atlas, as well as internal patient data, we show that whole-slide image classification models display marked performance disparities across different demographic groups when used to subtype breast and lung carcinomas and to predict IDH1 mutations in gliomas. For example, when using common modeling approaches, we observed performance gaps (in area under the receiver operating characteristic curve) between white and Black patients of 3.0% for breast cancer subtyping, 10.9% for lung cancer subtyping and 16.0% for IDH1 mutation prediction in gliomas. We found that richer feature representations obtained from self-supervised vision foundation models reduce performance variations between groups. These representations provide improvements upon weaker models even when those weaker models are combined with state-of-the-art bias mitigation strategies and modeling choices. Nevertheless, self-supervised vision foundation models do not fully eliminate these discrepancies, highlighting the continuing need for bias mitigation efforts in computational pathology. Finally, we demonstrate that our results extend to other demographic factors beyond patient race. Given these findings, we encourage regulatory and policy agencies to integrate demographic-stratified evaluation into their assessment guidelines.

Generative models improve fairness of medical classifiers under distribution shifts

Domain generalization is a ubiquitous challenge for machine learning in healthcare. Model performance in real-world conditions might be lower than expected because of discrepancies between the data encountered during deployment and development. Underrepresentation of some groups or conditions during model development is a common cause of this phenomenon. This challenge is often not readily addressed by targeted data acquisition and 'labeling' by expert clinicians, which can be prohibitively expensive or practically impossible because of the rarity of conditions or the available clinical expertise. We hypothesize that advances in generative artificial intelligence can help mitigate this unmet need in a steerable fashion, enriching our training dataset with synthetic examples that address shortfalls of underrepresented conditions or subgroups. We show that diffusion models can automatically learn realistic augmentations from data in a label-efficient manner. We demonstrate that learned augmentations make models more robust and statistically fair in-distribution and out of distribution. To evaluate the generality of our approach, we studied three distinct medical imaging contexts of varying difficulty: (1) histopathology, (2) chest X-ray and (3) dermatology images. Complementing real samples with synthetic ones improved the robustness of models in all three medical tasks and increased fairness by improving the accuracy of clinical diagnosis within underrepresented groups, especially out of distribution.

Implications of Bias in Artificial Intelligence: Considerations for Cardiovascular Imaging

PURPOSE OF REVIEW

Bias in artificial intelligence (AI) models can result in unintended consequences. In cardiovascular imaging, biased AI models used in clinical practice can negatively affect patient outcomes. Biased AI models result from decisions made when training and evaluating a model. This paper is a comprehensive guide for AI development teams to understand assumptions in datasets and chosen metrics for outcome/ground truth, and how this translates to real-world performance for cardiovascular disease (CVD).

RECENT FINDINGS

CVDs are the number one cause of mortality worldwide; however, the prevalence, burden, and outcomes of CVD vary across gender and race. Several biomarkers are also shown to vary among different populations and ethnic/racial groups. Inequalities in clinical trial inclusion, clinical presentation, diagnosis, and treatment are preserved in health data that is ultimately used to train AI algorithms, leading to potential biases in model performance. Despite the notion that AI models themselves are biased, AI can also help to mitigate bias (e.g., bias auditing tools). In this review paper, we describe in detail implicit and explicit biases in the care of cardiovascular disease that may be present in existing datasets but are not obvious to model developers. We review disparities in CVD outcomes across different genders and race groups, differences in treatment of historically marginalized groups, and disparities in clinical trials for various cardiovascular diseases and outcomes. Thereafter, we summarize some CVD AI literature that shows bias in CVD AI as well as approaches that AI is being used to mitigate CVD bias.

Enhancing the fairness of AI prediction models by Quasi-Pareto improvement among heterogeneous thyroid nodule population

Artificial Intelligence (AI) models for medical diagnosis often face challenges of generalizability and fairness. We highlighted the algorithmic unfairness in a large thyroid ultrasound dataset with significant diagnostic performance disparities across subgroups linked causally to sample size imbalances. To address this, we introduced the Quasi-Pareto Improvement (QPI) approach and a deep learning implementation (QP-Net) combining multi-task learning and domain adaptation to improve model performance among disadvantaged subgroups without compromising overall population performance. On the thyroid ultrasound dataset, our method significantly mitigated the area under curve (AUC) disparity for three less-prevalent subgroups by 0.213, 0.112, and 0.173 while maintaining the AUC for dominant subgroups; we also further confirmed the generalizability of our approach on two public datasets: the ISIC2019 skin disease dataset and the CheXpert chest radiograph dataset. Here we show the QPI approach to be widely applicable in promoting AI for equitable healthcare outcomes.

Understanding skin color bias in deep learning-based skin lesion segmentation

BACKGROUND

The field of dermatological image analysis using deep neural networks includes the semantic segmentation of skin lesions, pivotal for lesion analysis, pathology inference, and diagnoses. While biases in neural network-based dermatoscopic image classification against darker skin tones due to dataset imbalance and contrast disparities are acknowledged, a comprehensive exploration of skin color bias in lesion segmentation models is lacking. It is imperative to address and understand the biases in these models.

METHODS

Our study comprehensively evaluates skin tone bias within prevalent neural networks for skin lesion segmentation. Since no information about skin color exists in widely used datasets, to quantify the bias we use three distinct skin color estimation methods: Fitzpatrick skin type estimation, Individual Typology Angle estimation as well as manual grouping of images by skin color. We assess bias across common models by training a variety of U-Net-based models on three widely-used datasets with 1758 different dermoscopic and clinical images. We also evaluate commonly suggested methods to mitigate bias.

RESULTS

Our findings expose a significant and large correlation between segmentation performance and skin color, revealing consistent challenges in segmenting lesions for darker skin tones across diverse datasets. Using various methods of skin color quantification, we have found significant bias in skin lesion segmentation against darker-skinned individuals when evaluated both in and out-of-sample. We also find that commonly used methods for bias mitigation do not result in any significant reduction in bias.

CONCLUSIONS

Our findings suggest a pervasive bias in most published lesion segmentation methods, given our use of commonly employed neural network architectures and publicly available datasets. In light of our findings, we propose recommendations for unbiased dataset collection, labeling, and model development. This presents the first comprehensive evaluation of fairness in skin lesion segmentation.

Machine Learning and Bias in Medical Imaging: Opportunities and Challenges

Bias in health care has been well documented and results in disparate and worsened outcomes for at-risk groups. Medical imaging plays a critical role in facilitating patient diagnoses but involves multiple sources of bias including factors related to access to imaging modalities, acquisition of images, and assessment (ie, interpretation) of imaging data. Machine learning (ML) applied to diagnostic imaging has demonstrated the potential to improve the quality of imaging-based diagnosis and the precision of measuring imaging-based traits. Algorithms can leverage subtle information not visible to the human eye to detect underdiagnosed conditions or derive new disease phenotypes by linking imaging features with clinical outcomes, all while mitigating cognitive bias in interpretation. Importantly, however, the application of ML to diagnostic imaging has the potential to either reduce or propagate bias. Understanding the potential gain as well as the potential risks requires an understanding of how and what ML models learn. Common risks of propagating bias can arise from unbalanced training, suboptimal architecture design or selection, and uneven application of models. Notwithstanding these risks, ML may yet be applied to improve gain from imaging across all 3A's (access, acquisition, and assessment) for all patients. In this review, we present a framework for understanding the balance of opportunities and challenges for minimizing bias in medical imaging, how ML may improve current approaches to imaging, and what specific design considerations should be made as part of efforts to maximize the quality of health care for all.

Factors associated with attitudes toward research MRI in older Asian Americans

INTRODUCTION

South Asian (SA) and East Asian (EA) older adults represent the fastest-growing racial/ethnic groups of Americans at risk for dementia. While recruiting older SA adults into a brain health study, we encountered unexpected hesitancy toward structural brain magnetic resonance imaging (MRI) analysis and stigmatizing attitudes related to internal locus of control (LoC) for future dementia risks. We hypothesized that support for MRI-related research was influenced by these attitudes as well as personal MRI experience, perceived MRI safety, and concerns for personal risk for future dementia/stroke.

METHODS

We developed a brief cross-sectional survey to assess older adults' MRI experiences and perceptions, desire to learn of six incidental findings of increasing impact on health, and attitudes related to dementia (including LoC) and research participation. We recruited a convenience sample of 256 respondents (74% reporting as 50+) from the New Jersey/New York City area to complete the survey (offered in English, Chinese, Korean, and Spanish) and modeled the proportional odds (PO) for favorable attitudes toward research activities.

RESULTS

Seventy-seven SA and 84 EA respondents were analyzed alongside 95 White, Black, or Hispanic adults. White (PO = 2.54, p = 0.013) and EA (PO = 2.14, p = 0.019) respondents were both more likely than SA respondents to endorse healthy volunteers' participation in research, and the difference between White and SA respondents was mediated by the latter's greater internal LoC for dementia risks. EA respondents had more worries for future dementia/stroke than SA respondents (p = 0.006) but still shared SA respondents' lower wish (measured by proportion of total) to learn of incidental MRI findings.

DISCUSSION

SA-and EA compared to SA-older adults had low desire to learn of incidental MRI findings but had different attitudes toward future dementia/stroke risks. A culturally appropriate protocol to disclose incidental MRI findings may improve SA and EA participation in brain health research.

Highlights

Older Asian Americans have limited interest in incidental findings on research MRISouth Asians are most likely to attribute dementia to people's own behaviorsSouth Asians' attitudes mediate lower support for healthy volunteers in researchSouth and East Asians differ in dementia worries and research-related attitudes.

An Investigation into Race Bias in Random Forest Models Based on Breast DCE-MRI Derived Radiomics Features

Recent research has shown that artificial intelligence (AI) models can exhibit bias in performance when trained using data that are imbalanced by protected attribute(s). Most work to date has focused on deep learning models, but classical AI techniques that make use of hand-crafted features may also be susceptible to such bias. In this paper we investigate the potential for race bias in random forest (RF) models trained using radiomics features. Our application is prediction of tumour molecular subtype from dynamic contrast enhanced magnetic resonance imaging (DCE-MRI) of breast cancer patients. Our results show that radiomics features derived from DCE-MRI data do contain race-identifiable information, and that RF models can be trained to predict White and Black race from these data with 60-70% accuracy, depending on the subset of features used. Furthermore, RF models trained to predict tumour molecular subtype using race-imbalanced data seem to produce biased behaviour, exhibiting better performance on test data from the race on which they were trained.

Toward advanced diagnosis and management of inherited arrhythmia syndromes: Harnessing the capabilities of artificial intelligence and machine learning

The use of advanced computational technologies, such as artificial intelligence (AI), is now exerting a significant influence on various aspects of life, including health care and science. AI has garnered remarkable public notice with the release of deep learning models that can model anything from artwork to academic papers with minimal human intervention. Machine learning, a method that uses algorithms to extract information from raw data and represent it in a model, and deep learning, a method that uses multiple layers to progressively extract higher-level features from the raw input with minimal human intervention, are increasingly leveraged to tackle problems in the health sector, including utilization for clinical decision support in cardiovascular medicine. Inherited arrhythmia syndromes are a clinical domain where multiple unanswered questions remain despite unprecedented progress over the past 2 decades with the introduction of large panel genetic testing and the first steps in precision medicine. In particular, AI tools can help address gaps in clinical diagnosis by identifying individuals with concealed or transient phenotypes; enhance risk stratification by elevating recognition of underlying risk burden beyond widely recognized risk factors; improve prediction of response to therapy, and further prognostication. In this contemporary review, we provide a summary of the AI models developed to solve challenges in inherited arrhythmia syndromes and also outline gaps that can be filled with the development of intelligent AI models.

Mitigating Racial And Ethnic Bias And Advancing Health Equity In Clinical Algorithms: A Scoping Review

In August 2022 the Department of Health and Human Services (HHS) issued a notice of proposed rulemaking prohibiting covered entities, which include health care providers and health plans, from discriminating against individuals when using clinical algorithms in decision making. However, HHS did not provide specific guidelines on how covered entities should prevent discrimination. We conducted a scoping review of literature published during the period 2011-22 to identify health care applications, frameworks, reviews and perspectives, and assessment tools that identify and mitigate bias in clinical algorithms, with a specific focus on racial and ethnic bias. Our scoping review encompassed 109 articles comprising 45 empirical health care applications that included tools tested in health care settings, 16 frameworks, and 48 reviews and perspectives. We identified a wide range of technical, operational, and systemwide bias mitigation strategies for clinical algorithms, but there was no consensus in the literature on a single best practice that covered entities could employ to meet the HHS requirements. Future research should identify optimal bias mitigation methods for various scenarios, depending on factors such as patient population, clinical setting, algorithm design, and types of bias to be addressed.

Understanding Biases and Disparities in Radiology AI Datasets: A Review

Artificial intelligence (AI) continues to show great potential in disease detection and diagnosis on medical imaging with increasingly high accuracy. An important component of AI model creation is dataset development for training, validation, and testing. Diverse and high-quality datasets are critical to ensure robust and unbiased AI models that maintain validity, especially in traditionally underserved populations globally. Yet publicly available datasets demonstrate problems with quality and inclusivity. In this literature review, the authors evaluate publicly available medical imaging datasets for demographic, geographic, genetic, and disease representation or lack thereof and call for an increase emphasis on dataset development to maximize the impact of AI models.

Too much information? Asian Americans' preferences for incidental brain MRI findings

INTRODUCTION

South Asian (SA) and East Asian (EA) older adults represent the fastest growing group of Americans at risk for dementia, but their participation in aging and dementia research has been limited. While recruiting healthy SA older adults into a brain health study, we encountered unexpected hesitancy towards structural brain MRI analysis along with some stigmatizing attitudes related to internal locus of control (LoC) for future dementia risks. We hypothesized that support for MRI-related research was influenced by these attitudes as well as one's own MRI experience, perceived MRI safety, and concerns for one's own risks for future dementia/stroke.

METHODS

We developed a brief cross-sectional survey to assess older adults' MRI experiences and perceptions, desire to learn of six incidental findings of increasing health implications, and attitudes related to dementia as well as research participation. We recruited a convenience sample of 256 respondents (74% reporting as 50+) from the New Jersey/New York City area to complete the survey, and modeled the proportional odds (P.O.) for pro-research attitudes.

RESULTS

77 SA and 84 EA respondents were analyzed with 95 non-Asian adults. White (P.O.=2.54, p=0.013) and EA (P.O.=2.14, p=0.019) respondents were both more likely than SA respondents to endorse healthy volunteers' participation in research, and the difference between White and SA respondents was mediated by the latter's greater internal LoC for dementia risks. EA respondents had more worries for future dementia/stroke than SA respondents (p=0.006), but still shared SA respondents' low desire to learn of incidental MRI findings.

DISCUSSION

SA and EA older adults had different attitudes towards future dementia/stroke risks, but shared a low desire to learn of incidental MRI findings. A culturally-appropriate protocol to disclose incidental MRI findings may improve SA and EA participation in brain health research.

Digital health-high tech or high touch?

Digital transformation in medicine refers to the implementation of information technology-driven developments in the healthcare system and their impact on the way we teach, share, and practice medicine. We would like to provide an overview of current developments and opportunities but also of the risks of digital transformation in medicine. Therefore, we examine the possibilities wearables and digital biomarkers provide for early detection and monitoring of diseases and discuss the potential of artificial intelligence applications in medicine. Furthermore, we outline new opportunities offered by telemedicine applications and digital therapeutics, discuss the aspects of social media in healthcare, and provide an outlook on "Health 4.0."

Clinician's guide to trustworthy and responsible artificial intelligence in cardiovascular imaging

A growing number of artificial intelligence (AI)-based systems are being proposed and developed in cardiology, driven by the increasing need to deal with the vast amount of clinical and imaging data with the ultimate aim of advancing patient care, diagnosis and prognostication. However, there is a critical gap between the development and clinical deployment of AI tools. A key consideration for implementing AI tools into real-life clinical practice is their "trustworthiness" by end-users. Namely, we must ensure that AI systems can be trusted and adopted by all parties involved, including clinicians and patients. Here we provide a summary of the concepts involved in developing a "trustworthy AI system." We describe the main risks of AI applications and potential mitigation techniques for the wider application of these promising techniques in the context of cardiovascular imaging. Finally, we show why trustworthy AI concepts are important governing forces of AI development.

Quality of reporting in AI cardiac MRI segmentation studies - A systematic review and recommendations for future studies

Background

There has been a rapid increase in the number of Artificial Intelligence (AI) studies of cardiac MRI (CMR) segmentation aiming to automate image analysis. However, advancement and clinical translation in this field depend on researchers presenting their work in a transparent and reproducible manner. This systematic review aimed to evaluate the quality of reporting in AI studies involving CMR segmentation.

Methods

MEDLINE and EMBASE were searched for AI CMR segmentation studies in April 2022. Any fully automated AI method for segmentation of cardiac chambers, myocardium or scar on CMR was considered for inclusion. For each study, compliance with the Checklist for Artificial Intelligence in Medical Imaging (CLAIM) was assessed. The CLAIM criteria were grouped into study, dataset, model and performance description domains.

Results

209 studies published between 2012 and 2022 were included in the analysis. Studies were mainly published in technical journals (58%), with the majority (57%) published since 2019. Studies were from 37 different countries, with most from China (26%), the United States (18%) and the United Kingdom (11%). Short axis CMR images were most frequently used (70%), with the left ventricle the most commonly segmented cardiac structure (49%). Median compliance of studies with CLAIM was 67% (IQR 59-73%). Median compliance was highest for the model description domain (100%, IQR 80-100%) and lower for the study (71%, IQR 63-86%), dataset (63%, IQR 50-67%) and performance (60%, IQR 50-70%) description domains.

Conclusion

This systematic review highlights important gaps in the literature of CMR studies using AI. We identified key items missing-most strikingly poor description of patients included in the training and validation of AI models and inadequate model failure analysis-that limit the transparency, reproducibility and hence validity of published AI studies. This review may support closer adherence to established frameworks for reporting standards and presents recommendations for improving the quality of reporting in this field.

Systematic Review Registration

[www.crd.york.ac.uk/prospero/], identifier [CRD42022279214].