On the Interpretability of Artificial Intelligence in Radiology: Challenges and Opportunities

Equity within AI systems: What can health leaders expect?

Artificial Intelligence (AI) for health has a great potential; it has already proven to be successful in enhancing patient outcomes, facilitating professional work and benefiting administration. However, AI presents challenges related to health equity defined as an opportunity for people to reach their fullest health potential. This article discusses the opportunities and challenges that AI presents in health and examines ways in which inequities related to AI can be mitigated.

AI in Pathology: What could possibly go wrong?

The field of medicine is undergoing rapid digital transformation. Pathologists are now striving to digitize their data, workflows, and interpretations, assisted by the enabling development of whole-slide imaging. Going digital means that the analog process of human diagnosis can be augmented or even replaced by rapidly evolving AI approaches, which are just now entering into clinical practice. But with such progress comes challenges that reflect a variety of stressors, including the impact of unrepresentative training data with accompanying implicit bias, data privacy concerns, and fragility of algorithm performance. Beyond such core digital aspects, considerations arise related to difficulties presented by changing disease presentations, diagnostic approaches, and therapeutic options. While some tools such as data federation can help with broadening data diversity while preserving expertise and local control, they may not be the full answer to some of these issues. The impact of AI in pathology on the field's human practitioners is still very much unknown: installation of unconscious bias and deference to AI guidance need to be understood and addressed. If AI is widely adopted, it may remove many inefficiencies in daily practice and compensate for staff shortages. It may also cause practitioner deskilling, dethrilling, and burnout. We discuss the technological, clinical, legal, and sociological factors that will influence the adoption of AI in pathology, and its eventual impact for good or ill.

Attri-VAE: Attribute-based interpretable representations of medical images with variational autoencoders

Deep learning (DL) methods where interpretability is intrinsically considered as part of the model are required to better understand the relationship of clinical and imaging-based attributes with DL outcomes, thus facilitating their use in the reasoning behind the medical decisions. Latent space representations built with variational autoencoders (VAE) do not ensure individual control of data attributes. Attribute-based methods enforcing attribute disentanglement have been proposed in the literature for classical computer vision tasks in benchmark data. In this paper, we propose a VAE approach, the Attri-VAE, that includes an attribute regularization term to associate clinical and medical imaging attributes with different regularized dimensions in the generated latent space, enabling a better-disentangled interpretation of the attributes. Furthermore, the generated attention maps explained the attribute encoding in the regularized latent space dimensions. Using the Attri-VAE approach we analyzed healthy and myocardial infarction patients with clinical, cardiac morphology, and radiomics attributes. The proposed model provided an excellent trade-off between reconstruction fidelity, disentanglement, and interpretability, outperforming state-of-the-art VAE approaches according to several quantitative metrics. The resulting latent space allowed the generation of realistic synthetic data in the trajectory between two distinct input samples or along a specific attribute dimension to better interpret changes between different cardiac conditions.

Graph Node Based Interpretability Guided Sample Selection for Active Learning

While supervised learning techniques have demonstrated state-of-the-art performance in many medical image analysis tasks, the role of sample selection is important. Selecting the most informative samples contributes to the system attaining optimum performance with minimum labeled samples, which translates to fewer expert interventions and cost. Active Learning (AL) methods for informative sample selection are effective in boosting performance of computer aided diagnosis systems when limited labels are available. Conventional approaches to AL have mostly focused on the single label setting where a sample has only one disease label from the set of possible labels. These approaches do not perform optimally in the multi-label setting where a sample can have multiple disease labels (e.g. in chest X-ray images). In this paper we propose a novel sample selection approach based on graph analysis to identify informative samples in a multi-label setting. For every analyzed sample, each class label is denoted as a separate node of a graph. Building on findings from interpretability of deep learning models, edge interactions in this graph characterize similarity between corresponding interpretability saliency map model encodings. We explore different types of graph aggregation to identify informative samples for active learning. We apply our method to public chest X-ray and medical image datasets, and report improved results over state-of-the-art AL techniques in terms of model performance, learning rates, and robustness.

Artificial intelligence in liver cancers: Decoding the impact of machine learning models in clinical diagnosis of primary liver cancers and liver cancer metastases

Liver cancers are the fourth leading cause of cancer-related mortality worldwide. In the past decade, breakthroughs in the field of artificial intelligence (AI) have inspired development of algorithms in the cancer setting. A growing body of recent studies have evaluated machine learning (ML) and deep learning (DL) algorithms for pre-screening, diagnosis and management of liver cancer patients through diagnostic image analysis, biomarker discovery and predicting personalized clinical outcomes. Despite the promise of these early AI tools, there is a significant need to explain the 'black box' of AI and work towards deployment to enable ultimate clinical translatability. Certain emerging fields such as RNA nanomedicine for targeted liver cancer therapy may also benefit from application of AI, specifically in nano-formulation research and development given that they are still largely reliant on lengthy trial-and-error experiments. In this paper, we put forward the current landscape of AI in liver cancers along with the challenges of AI in liver cancer diagnosis and management. Finally, we have discussed the future perspectives of AI application in liver cancer and how a multidisciplinary approach using AI in nanomedicine could accelerate the transition of personalized liver cancer medicine from bench side to the clinic.

AI: Can It Make a Difference to the Predictive Value of Ultrasound Breast Biopsy?

(1) Background: This study aims to compare the ground truth (pathology results) against the BI-RADS classification of images acquired while performing breast ultrasound diagnostic examinations that led to a biopsy and against the result of processing the same images through the AI algorithm KOIOS DS ^TM (KOIOS). (2) Methods: All results of biopsies performed with ultrasound guidance during 2019 were recovered from the pathology department. Readers selected the image which better represented the BI-RADS classification, confirmed correlation to the biopsied image, and submitted it to the KOIOS AI software. The results of the BI-RADS classification of the diagnostic study performed at our institution were set against the KOIOS classification and both were compared to the pathology reports. (3) Results: 403 cases were included in this study. Pathology rendered 197 malignant and 206 benign reports. Four biopsies on BI-RADS 0 and two images are included. Of fifty BI-RADS 3 cases biopsied, only seven rendered cancers. All but one had a positive or suspicious cytology; all were classified as suspicious by KOIOS. Using KOIOS, 17 B3 biopsies could have been avoided. Of 347 BI-RADS 4, 5, and 6 cases, 190 were malignant (54.7%). Because only KOIOS suspicious and probably malignant categories should be biopsied, 312 biopsies would have resulted in 187 malignant lesions (60%), but 10 cancers would have been missed. (4) Conclusions: KOIOS had a higher ratio of positive biopsies in this selected case study vis-à-vis the BI-RADS 4, 5 and 6 categories. A large number of biopsies in the BI-RADS 3 category could have been avoided.

A review of the application of three-dimensional convolutional neural networks for the diagnosis of Alzheimer's disease using neuroimaging

Alzheimer's disease (AD) is a degenerative disorder that leads to progressive, irreversible cognitive decline. To obtain an accurate and timely diagnosis and detect AD at an early stage, numerous approaches based on convolutional neural networks (CNNs) using neuroimaging data have been proposed. Because 3D CNNs can extract more spatial discrimination information than 2D CNNs, they have emerged as a promising research direction in the diagnosis of AD. The aim of this article is to present the current state of the art in the diagnosis of AD using 3D CNN models and neuroimaging modalities, focusing on the 3D CNN architectures and classification methods used, and to highlight potential future research topics. To give the reader a better overview of the content mentioned in this review, we briefly introduce the commonly used imaging datasets and the fundamentals of CNN architectures. Then we carefully analyzed the existing studies on AD diagnosis, which are divided into two levels according to their inputs: 3D subject-level CNNs and 3D patch-level CNNs, highlighting their contributions and significance in the field. In addition, this review discusses the key findings and challenges from the studies and highlights the lessons learned as a roadmap for future research. Finally, we summarize the paper by presenting some major findings, identifying open research challenges, and pointing out future research directions.

The current status and future of FDA-approved artificial intelligence tools in chest radiology in the United States

Artificial intelligence (AI) is becoming more widespread within radiology. Capabilities that AI algorithms currently provide include detection, segmentation, classification, and quantification of pathological findings. Artificial intelligence software have created challenges for the traditional United States Food and Drug Administration (FDA) approval process for medical devices given their abilities to evolve over time with incremental data input. Currently, there are 190 FDA-approved radiology AI-based software devices, 42 of which pertain specifically to thoracic radiology. The majority of these algorithms are approved for the detection and/or analysis of pulmonary nodules, for monitoring placement of endotracheal tubes and indwelling catheters, for detection of emergent findings, and for assessment of pulmonary parenchyma; however, as technology evolves, there are many other potential applications that can be explored. For example, evaluation of non-idiopathic pulmonary fibrosis interstitial lung diseases, synthesis of imaging, clinical and/or laboratory data to yield comprehensive diagnoses, and survival or prognosis prediction of certain pathologies. With increasing physician and developer engagement, transparency and frequent communication between developers and regulatory agencies, such as the FDA, AI medical devices will be able to provide a critical supplement to patient management and ultimately enhance physicians' ability to improve patient care.

AAPM task group report 273: Recommendations on best practices for AI and machine learning for computer-aided diagnosis in medical imaging

Rapid advances in artificial intelligence (AI) and machine learning, and specifically in deep learning (DL) techniques, have enabled broad application of these methods in health care. The promise of the DL approach has spurred further interest in computer-aided diagnosis (CAD) development and applications using both "traditional" machine learning methods and newer DL-based methods. We use the term CAD-AI to refer to this expanded clinical decision support environment that uses traditional and DL-based AI methods. Numerous studies have been published to date on the development of machine learning tools for computer-aided, or AI-assisted, clinical tasks. However, most of these machine learning models are not ready for clinical deployment. It is of paramount importance to ensure that a clinical decision support tool undergoes proper training and rigorous validation of its generalizability and robustness before adoption for patient care in the clinic. To address these important issues, the American Association of Physicists in Medicine (AAPM) Computer-Aided Image Analysis Subcommittee (CADSC) is charged, in part, to develop recommendations on practices and standards for the development and performance assessment of computer-aided decision support systems. The committee has previously published two opinion papers on the evaluation of CAD systems and issues associated with user training and quality assurance of these systems in the clinic. With machine learning techniques continuing to evolve and CAD applications expanding to new stages of the patient care process, the current task group report considers the broader issues common to the development of most, if not all, CAD-AI applications and their translation from the bench to the clinic. The goal is to bring attention to the proper training and validation of machine learning algorithms that may improve their generalizability and reliability and accelerate the adoption of CAD-AI systems for clinical decision support.

User-centred design of a clinical decision support system for palliative care: Insights from healthcare professionals

Objective

Although clinical decision support systems (CDSS) have many benefits for clinical practice, they also have several barriers to their acceptance by professionals. Our objective in this study was to design and validate The Aleph palliative care (PC) CDSS through a user-centred method, considering the predictions of the artificial intelligence (AI) core, usability and user experience (UX).

Methods

We performed two rounds of individual evaluation sessions with potential users. Each session included a model evaluation, a task test and a usability and UX assessment.

Results

The machine learning (ML) predictive models outperformed the participants in the three predictive tasks. System Usability Scale (SUS) reported 62.7  ±  14.1 and 65  ±  26.2 on a 100-point rating scale for both rounds, respectively, while User Experience Questionnaire - Short Version (UEQ-S) scores were 1.42 and 1.5 on the -3 to 3 scale.

Conclusions

The think-aloud method and including the UX dimension helped us to identify most of the workflow implementation issues. The system has good UX hedonic qualities; participants were interested in the tool and responded positively to it. Performance regarding usability was modest but acceptable.

Interpretable machine learning for automated left ventricular scar quantification in hypertrophic cardiomyopathy patients

Scar quantification on cardiovascular magnetic resonance (CMR) late gadolinium enhancement (LGE) images is important in risk stratifying patients with hypertrophic cardiomyopathy (HCM) due to the importance of scar burden in predicting clinical outcomes. We aimed to develop a machine learning (ML) model that contours left ventricular (LV) endo- and epicardial borders and quantifies CMR LGE images from HCM patients.We retrospectively studied 2557 unprocessed images from 307 HCM patients followed at the University Health Network (Canada) and Tufts Medical Center (USA). LGE images were manually segmented by two experts using two different software packages. Using 6SD LGE intensity cutoff as the gold standard, a 2-dimensional convolutional neural network (CNN) was trained on 80% and tested on the remaining 20% of the data. Model performance was evaluated using the Dice Similarity Coefficient (DSC), Bland-Altman, and Pearson's correlation. The 6SD model DSC scores were good to excellent at 0.91 ± 0.04, 0.83 ± 0.03, and 0.64 ± 0.09 for the LV endocardium, epicardium, and scar segmentation, respectively. The bias and limits of agreement for the percentage of LGE to LV mass were low (-0.53 ± 2.71%), and correlation high (r = 0.92). This fully automated interpretable ML algorithm allows rapid and accurate scar quantification from CMR LGE images. This program does not require manual image pre-processing, and was trained with multiple experts and software, increasing its generalizability.

Brain age gap in neuromyelitis optica spectrum disorders and multiple sclerosis

OBJECTIVE

To evaluate the clinical significance of deep learning-derived brain age prediction in neuromyelitis optica spectrum disorder (NMOSD) relative to relapsing-remitting multiple sclerosis (RRMS).

METHODS

This cohort study used data retrospectively collected from 6 tertiary neurological centres in China between 2009 and 2018. In total, 199 patients with NMOSD and 200 patients with RRMS were studied alongside 269 healthy controls. Clinical follow-up was available in 85 patients with NMOSD and 124 patients with RRMS (mean duration NMOSD=5.8±1.9 (1.9-9.9) years, RRMS=5.2±1.7 (1.5-9.2) years). Deep learning was used to learn 'brain age' from MRI scans in the healthy controls and estimate the brain age gap (BAG) in patients.

RESULTS

A significantly higher BAG was found in the NMOSD (5.4±8.2 years) and RRMS (13.0±14.7 years) groups compared with healthy controls. A higher baseline disability score and advanced brain volume loss were associated with increased BAG in both patient groups. A longer disease duration was associated with increased BAG in RRMS. BAG significantly predicted Expanded Disability Status Scale worsening in patients with NMOSD and RRMS.

CONCLUSIONS

There is a clear BAG in NMOSD, although smaller than in RRMS. The BAG is a clinically relevant MRI marker in NMOSD and RRMS.

Identifying the regional substrates predictive of Alzheimer's disease progression through a convolutional neural network model and occlusion

Progressive brain atrophy is a key neuropathological hallmark of Alzheimer's disease (AD) dementia. However, atrophy patterns along the progression of AD dementia are diffuse and variable and are often missed by univariate methods. Consequently, identifying the major regional atrophy patterns underlying AD dementia progression is challenging. In the current study, we propose a method that evaluates the degree to which specific regional atrophy patterns are predictive of AD dementia progression, while holding all other atrophy changes constant using a total sample of 334 subjects. We first trained a dense convolutional neural network model to differentiate individuals with mild cognitive impairment (MCI) who progress to AD dementia versus those with a stable MCI diagnosis. Then, we retested the model multiple times, each time occluding different regions of interest (ROIs) from the model's testing set's input. We also validated this approach by occluding ROIs based on Braak's staging scheme. We found that the hippocampus, fusiform, and inferior temporal gyri were the strongest predictors of AD dementia progression, in agreement with established staging models. We also found that occlusion of limbic ROIs defined according to Braak stage III had the largest impact on the performance of the model. Our predictive model reveals the major regional patterns of atrophy predictive of AD dementia progression. These results highlight the potential for early diagnosis and stratification of individuals with prodromal AD dementia based on patterns of cortical atrophy, prior to interventional clinical trials.

[Validation and implementation of artificial intelligence in radiology : Quo vadis in 2022?]

BACKGROUND

The hype around artificial intelligence (AI) in radiology continues and the number of approved AI tools is growing steadily. Despite the great potential, integration into clinical routine in radiology remains limited. In addition, the large number of individual applications poses a challenge for clinical routine, as individual applications have to be selected for different questions and organ systems, which increases the complexity and time required.

OBJECTIVES

This review will discuss the current status of validation and implementation of AI tools in clinical routine, and identify possible approaches for an improved assessment of the generalizability of results of AI tools.

MATERIALS AND METHODS

A literature search in various literature and product databases as well as publications, position papers, and reports from various stakeholders was conducted for this review.

RESULTS

Scientific evidence and independent validation studies are available for only a few commercial AI tools and the generalizability of the results often remains questionable.

CONCLUSIONS

One challenge is the multitude of offerings for individual, specific application areas by a large number of manufacturers, making integration into the existing site-specific IT infrastructure more difficult. Furthermore, remuneration for the use of AI tools in clinical routine by health insurance companies in Germany is lacking. But in order for reimbursement to be granted, the clinical utility of new applications must first be proven. Such proof, however, is lacking for most applications.

Artificial intelligence-based model for COVID-19 prognosis incorporating chest radiographs and clinical data; a retrospective model development and validation study

OBJECTIVES

The purpose of this study was to develop an artificial intelligence-based model to prognosticate COVID-19 patients at admission by combining clinical data and chest radiographs.

METHODS

This retrospective study used the Stony Brook University COVID-19 dataset of 1384 inpatients. After exclusions, 1356 patients were randomly divided into training (1083) and test datasets (273). We implemented three artificial intelligence models, which classified mortality, ICU admission, or ventilation risk. Each model had three submodels with different inputs: clinical data, chest radiographs, and both. We showed the importance of the variables using SHapley Additive exPlanations (SHAP) values.

RESULTS

The mortality prediction model was best overall with area under the curve, sensitivity, specificity, and accuracy of 0.79 (0.72-0.86), 0.74 (0.68-0.79), 0.77 (0.61-0.88), and 0.74 (0.69-0.79) for the clinical data-based model; 0.77 (0.69-0.85), 0.67 (0.61-0.73), 0.81 (0.67-0.92), 0.70 (0.64-0.75) for the image-based model, and 0.86 (0.81-0.91), 0.76 (0.70-0.81), 0.77 (0.61-0.88), 0.76 (0.70-0.81) for the mixed model. The mixed model had the best performance (p value < 0.05). The radiographs ranked fourth for prognostication overall, and first of the inpatient tests assessed.

CONCLUSIONS

These results suggest that prognosis models become more accurate if AI-derived chest radiograph features and clinical data are used together.

ADVANCES IN KNOWLEDGE

This AI model evaluates chest radiographs together with clinical data in order to classify patients as having high or low mortality risk. This work shows that chest radiographs taken at admission have significant COVID-19 prognostic information compared to clinical data other than age and sex.

Artificial intelligence for precision medicine in autoimmune liver disease

Autoimmune liver diseases (AiLDs) are rare autoimmune conditions of the liver and the biliary tree with unknown etiology and limited treatment options. AiLDs are inherently characterized by a high degree of complexity, which poses great challenges in understanding their etiopathogenesis, developing novel biomarkers and risk-stratification tools, and, eventually, generating new drugs. Artificial intelligence (AI) is considered one of the best candidates to support researchers and clinicians in making sense of biological complexity. In this review, we offer a primer on AI and machine learning for clinicians, and discuss recent available literature on its applications in medicine and more specifically how it can help to tackle major unmet needs in AiLDs.

A manifesto on explainability for artificial intelligence in medicine

The rapid increase of interest in, and use of, artificial intelligence (AI) in computer applications has raised a parallel concern about its ability (or lack thereof) to provide understandable, or explainable, output to users. This concern is especially legitimate in biomedical contexts, where patient safety is of paramount importance. This position paper brings together seven researchers working in the field with different roles and perspectives, to explore in depth the concept of explainable AI, or XAI, offering a functional definition and conceptual framework or model that can be used when considering XAI. This is followed by a series of desiderata for attaining explainability in AI, each of which touches upon a key domain in biomedicine.

Artificial Intelligence in Clinical Practice: Implementation Considerations and Barriers

The rapid growth of artificial intelligence (AI) in radiology has led to Food and Drug Administration clearance of more than 20 AI algorithms for breast imaging. The steps involved in the clinical implementation of an AI product include identifying all stakeholders, selecting the appropriate product to purchase, evaluating it with a local data set, integrating it into the workflow, and monitoring its performance over time. Despite the potential benefits of improved quality and increased efficiency with AI, several barriers, such as high costs and liability concerns, may limit its widespread implementation. This article lists currently available AI products for breast imaging, describes the key elements of clinical implementation, and discusses barriers to clinical implementation.

Influence of contrast and texture based image modifications on the performance and attention shift of U-Net models for brain tissue segmentation

Contrast and texture modifications applied during training or test-time have recently shown promising results to enhance the generalization performance of deep learning segmentation methods in medical image analysis. However, a deeper understanding of this phenomenon has not been investigated. In this study, we investigated this phenomenon using a controlled experimental setting, using datasets from the Human Connectome Project and a large set of simulated MR protocols, in order to mitigate data confounders and investigate possible explanations as to why model performance changes when applying different levels of contrast and texture-based modifications. Our experiments confirm previous findings regarding the improved performance of models subjected to contrast and texture modifications employed during training and/or testing time, but further show the interplay when these operations are combined, as well as the regimes of model improvement/worsening across scanning parameters. Furthermore, our findings demonstrate a spatial attention shift phenomenon of trained models, occurring for different levels of model performance, and varying in relation to the type of applied image modification.

Computer vision in surgery: from potential to clinical value

Hundreds of millions of operations are performed worldwide each year, and the rising uptake in minimally invasive surgery has enabled fiber optic cameras and robots to become both important tools to conduct surgery and sensors from which to capture information about surgery. Computer vision (CV), the application of algorithms to analyze and interpret visual data, has become a critical technology through which to study the intraoperative phase of care with the goals of augmenting surgeons' decision-making processes, supporting safer surgery, and expanding access to surgical care. While much work has been performed on potential use cases, there are currently no CV tools widely used for diagnostic or therapeutic applications in surgery. Using laparoscopic cholecystectomy as an example, we reviewed current CV techniques that have been applied to minimally invasive surgery and their clinical applications. Finally, we discuss the challenges and obstacles that remain to be overcome for broader implementation and adoption of CV in surgery.

Machine Learning and Deep Learning in Cardiothoracic Imaging: A Scoping Review

Machine-learning (ML) and deep-learning (DL) algorithms are part of a group of modeling algorithms that grasp the hidden patterns in data based on a training process, enabling them to extract complex information from the input data. In the past decade, these algorithms have been increasingly used for image processing, specifically in the medical domain. Cardiothoracic imaging is one of the early adopters of ML/DL research, and the COVID-19 pandemic resulted in more research focus on the feasibility and applications of ML/DL in cardiothoracic imaging. In this scoping review, we systematically searched available peer-reviewed medical literature on cardiothoracic imaging and quantitatively extracted key data elements in order to get a big picture of how ML/DL have been used in the rapidly evolving cardiothoracic imaging field. During this report, we provide insights on different applications of ML/DL and some nuances pertaining to this specific field of research. Finally, we provide general suggestions on how researchers can make their research more than just a proof-of-concept and move toward clinical adoption.

Improving disease classification performance and explainability of deep learning models in radiology with heatmap generators

As deep learning is widely used in the radiology field, the explainability of Artificial Intelligence (AI) models is becoming increasingly essential to gain clinicians' trust when using the models for diagnosis. In this research, three experiment sets were conducted with a U-Net architecture to improve the disease classification performance while enhancing the heatmaps corresponding to the model's focus through incorporating heatmap generators during training. All experiments used the dataset that contained chest radiographs, associated labels from one of the three conditions ["normal", "congestive heart failure (CHF)", and "pneumonia"], and numerical information regarding a radiologist's eye-gaze coordinates on the images. The paper that introduced this dataset developed a U-Net model, which was treated as the baseline model for this research, to show how the eye-gaze data can be used in multi-modal training for explainability improvement and disease classification. To compare the classification performances among this research's three experiment sets and the baseline model, the 95% confidence intervals (CI) of the area under the receiver operating characteristic curve (AUC) were measured. The best method achieved an AUC of 0.913 with a 95% CI of [0.860, 0.966]. "Pneumonia" and "CHF" classes, which the baseline model struggled the most to classify, had the greatest improvements, resulting in AUCs of 0.859 with a 95% CI of [0.732, 0.957] and 0.962 with a 95% CI of [0.933, 0.989], respectively. The decoder of the U-Net for the best-performing proposed method generated heatmaps that highlight the determining image parts in model classifications. These predicted heatmaps, which can be used for the explainability of the model, also improved to align well with the radiologist's eye-gaze data. Hence, this work showed that incorporating heatmap generators and eye-gaze information into training can simultaneously improve disease classification and provide explainable visuals that align well with how the radiologist viewed the chest radiographs when making diagnosis.

Benchmarking saliency methods for chest X-ray interpretation

Saliency methods, which produce heat maps that highlight the areas of the medical image that influence model prediction, are often presented to clinicians as an aid in diagnostic decision-making. However, rigorous investigation of the accuracy and reliability of these strategies is necessary before they are integrated into the clinical setting. In this work, we quantitatively evaluate seven saliency methods, including Grad-CAM, across multiple neural network architectures using two evaluation metrics. We establish the first human benchmark for chest X-ray segmentation in a multilabel classification set-up, and examine under what clinical conditions saliency maps might be more prone to failure in localizing important pathologies compared with a human expert benchmark. We find that (1) while Grad-CAM generally localized pathologies better than the other evaluated saliency methods, all seven performed significantly worse compared with the human benchmark, (2) the gap in localization performance between Grad-CAM and the human benchmark was largest for pathologies that were smaller in size and had shapes that were more complex, and (3) model confidence was positively correlated with Grad-CAM localization performance. Our work demonstrates that several important limitations of saliency methods must be addressed before we can rely on them for deep learning explainability in medical imaging.

Artificial intelligence for multimodal data integration in oncology

In oncology, the patient state is characterized by a whole spectrum of modalities, ranging from radiology, histology, and genomics to electronic health records. Current artificial intelligence (AI) models operate mainly in the realm of a single modality, neglecting the broader clinical context, which inevitably diminishes their potential. Integration of different data modalities provides opportunities to increase robustness and accuracy of diagnostic and prognostic models, bringing AI closer to clinical practice. AI models are also capable of discovering novel patterns within and across modalities suitable for explaining differences in patient outcomes or treatment resistance. The insights gleaned from such models can guide exploration studies and contribute to the discovery of novel biomarkers and therapeutic targets. To support these advances, here we present a synopsis of AI methods and strategies for multimodal data fusion and association discovery. We outline approaches for AI interpretability and directions for AI-driven exploration through multimodal data interconnections. We examine challenges in clinical adoption and discuss emerging solutions.

Interplay between Artificial Intelligence and Biomechanics Modeling in the Cardiovascular Disease Prediction

Cardiovascular disease (CVD) is the most common cause of morbidity and mortality worldwide, and early accurate diagnosis is the key point for improving and optimizing the prognosis of CVD. Recent progress in artificial intelligence (AI), especially machine learning (ML) technology, makes it possible to predict CVD. In this review, we first briefly introduced the overview development of artificial intelligence. Then we summarized some ML applications in cardiovascular diseases, including ML-based models to directly predict CVD based on risk factors or medical imaging findings and the ML-based hemodynamics with vascular geometries, equations, and methods for indirect assessment of CVD. We also discussed case studies where ML could be used as the surrogate for computational fluid dynamics in data-driven models and physics-driven models. ML models could be a surrogate for computational fluid dynamics, accelerate the process of disease prediction, and reduce manual intervention. Lastly, we briefly summarized the research difficulties and prospected the future development of AI technology in cardiovascular diseases.

Chest X-ray analysis empowered with deep learning: A systematic review

Chest radiographs are widely used in the medical domain and at present, chest X-radiation particularly plays an important role in the diagnosis of medical conditions such as pneumonia and COVID-19 disease. The recent developments of deep learning techniques led to a promising performance in medical image classification and prediction tasks. With the availability of chest X-ray datasets and emerging trends in data engineering techniques, there is a growth in recent related publications. Recently, there have been only a few survey papers that addressed chest X-ray classification using deep learning techniques. However, they lack the analysis of the trends of recent studies. This systematic review paper explores and provides a comprehensive analysis of the related studies that have used deep learning techniques to analyze chest X-ray images. We present the state-of-the-art deep learning based pneumonia and COVID-19 detection solutions, trends in recent studies, publicly available datasets, guidance to follow a deep learning process, challenges and potential future research directions in this domain. The discoveries and the conclusions of the reviewed work have been organized in a way that researchers and developers working in the same domain can use this work to support them in taking decisions on their research.

Natural Language Processing in Radiology: Update on Clinical Applications

Radiological reports are a valuable source of information used to guide clinical care and support research. Organizing and managing this content, however, frequently requires several manual curations due to the more common unstructured nature of the reports. However, manual review of these reports for clinical knowledge extraction is costly and time-consuming. Natural language processing (NLP) is a set of methods developed to extract structured meaning from a body of text and can be used to optimize the workflow of health care professionals. Specifically, NLP methods can help radiologists as decision support systems and improve the management of patients' medical data. In this study, we highlight the opportunities offered by NLP in the field of radiology. A comprehensive review of the most commonly used NLP methods to extract information from radiological reports and the development of tools to improve radiological workflow using this information is presented. Finally, we review the important limitations of these tools and discuss the relevant observations and trends in the application of NLP to radiology that could benefit the field in the future.

Artificial intelligence in multi-parametric magnetic resonance imaging: A review

Multi-parametric magnetic resonance imaging (mpMRI) is an indispensable tool in the clinical workflow for the diagnosis and treatment planning of various diseases. Machine learning-based artificial intelligence (AI) methods, especially those adopting the deep learning technique, have been extensively employed to perform mpMRI image classification, segmentation, registration, detection, reconstruction, and super-resolution. The current availability of increasing computational power and fast-improving AI algorithms have empowered numerous computer-based systems for applying mpMRI to disease diagnosis, imaging-guided radiotherapy, patient risk and overall survival time prediction, and the development of advanced quantitative imaging technology for magnetic resonance fingerprinting. However, the wide application of these developed systems in the clinic is still limited by a number of factors, including robustness, reliability, and interpretability. This survey aims to provide an overview for new researchers in the field as well as radiologists with the hope that they can understand the general concepts, main application scenarios, and remaining challenges of AI in mpMRI. This article is protected by copyright. All rights reserved.

Breast cancer patient characterisation and visualisation using deep learning and fisher information networks

Breast cancer is the most commonly diagnosed female malignancy globally, with better survival rates if diagnosed early. Mammography is the gold standard in screening programmes for breast cancer, but despite technological advances, high error rates are still reported. Machine learning techniques, and in particular deep learning (DL), have been successfully used for breast cancer detection and classification. However, the added complexity that makes DL models so successful reduces their ability to explain which features are relevant to the model, or whether the model is biased. The main aim of this study is to propose a novel visualisation to help characterise breast cancer patients using Fisher Information Networks on features extracted from mammograms using a DL model. In the proposed visualisation, patients are mapped out according to their similarities and can be used to study new patients as a ‘patient-like-me’ approach. When applied to the CBIS-DDSM dataset, it was shown that it is a competitive methodology that can (i) facilitate the analysis and decision-making process in breast cancer diagnosis with the assistance of the FIN visualisations and ‘patient-like-me’ analysis, and (ii) help improve diagnostic accuracy and reduce overdiagnosis by identifying the most likely diagnosis based on clinical similarities with neighbouring patients.

Automated diagnosis and prognosis of COVID-19 pneumonia from initial ER chest X-rays using deep learning

Background

Airspace disease as seen on chest X-rays is an important point in triage for patients initially presenting to the emergency department with suspected COVID-19 infection. The purpose of this study is to evaluate a previously trained interpretable deep learning algorithm for the diagnosis and prognosis of COVID-19 pneumonia from chest X-rays obtained in the ED.

Methods

This retrospective study included 2456 (50% RT-PCR positive for COVID-19) adult patients who received both a chest X-ray and SARS-CoV-2 RT-PCR test from January 2020 to March of 2021 in the emergency department at a single U.S. institution. A total of 2000 patients were included as an additional training cohort and 456 patients in the randomized internal holdout testing cohort for a previously trained Siemens AI-Radiology Companion deep learning convolutional neural network algorithm. Three cardiothoracic fellowship-trained radiologists systematically evaluated each chest X-ray and generated an airspace disease area-based severity score which was compared against the same score produced by artificial intelligence. The interobserver agreement, diagnostic accuracy, and predictive capability for inpatient outcomes were assessed. Principal statistical tests used in this study include both univariate and multivariate logistic regression.

Results

Overall ICC was 0.820 (95% CI 0.790–0.840). The diagnostic AUC for SARS-CoV-2 RT-PCR positivity was 0.890 (95% CI 0.861–0.920) for the neural network and 0.936 (95% CI 0.918–0.960) for radiologists. Airspace opacities score by AI alone predicted ICU admission (AUC = 0.870) and mortality (0.829) in all patients. Addition of age and BMI into a multivariate log model improved mortality prediction (AUC = 0.906).

Conclusion

The deep learning algorithm provides an accurate and interpretable assessment of the disease burden in COVID-19 pneumonia on chest radiographs. The reported severity scores correlate with expert assessment and accurately predicts important clinical outcomes. The algorithm contributes additional prognostic information not currently incorporated into patient management.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12879-022-07617-7.

Interpretable Dimension Reduction for MRI Channel Suppression

Channel suppression can reduce the redundant information in multiple channel receiver coils and accelerate reconstruction speed to meet real-time imaging requirements. The principal component analysis has been used for channel suppression, but it is difficult to be interpreted because all channels contribute to principal components. Furthermore, the importance of interpretability in machine learning has recently attracted increasing attention in radiology. To improve the interpretability of PCA-based channel suppression, a sparse PCA method is proposed to reduce the most coils' loadings to be zero. Channel suppression is formulated as solving a nonlinear eigenvalue problem using the inverse power method instead of the direct matrix decomposition. Experimental results of in vivo data show that the sparse PCA-based channel suppression not only improves the interpretability with sparse channels, but also improves reconstruction quality compared to the standard PCA-based reconstruction with the similar reconstruction time.

Toward understanding deep learning classification of anatomic sites: lessons from the development of a CBCT projection classifier

Purpose: Deep learning (DL) applications strongly depend on the training dataset and convolutional neural network architecture; however, it is unclear how to objectively select such parameters. We investigate the classification performance of different DL models and training schemes for the anatomic classification of cone-beam computed tomography (CBCT) projections. Approach: CBCT scans from 1055 patients were collected and manually classified into five anatomic classes and used to develop DL models to predict the anatomic class from single x-ray projections. VGG-16, Xception, and Inception v3 architectures were trained with 75% of the data, and the remaining 25% was used for testing and evaluation. To study the dependence of the classification performance on dataset size, training data was downsampled to various dataset sizes. Gradient-weighted class activation maps (grad-CAM) were generated using the model with highest classification performance, to identify regions with strong influence on CNN decisions. Results: The highest precision and recall values were achieved with VGG-16. One of the best performing combinations was the VGG-16 trained with 90 deg projections (mean class precision = 0.87). The training dataset size could be reduced to ∼ 50 % of its initial size, without compromising the classification performance. For correctly classified cases, Grad-CAM were more heavily weighted for anatomically relevant regions. Conclusions: It was possible to determine those dependencies with a higher influence on the classification performance of DL models for the studied task. Grad-CAM enabled the identification of possible sources of class confusion.

Imaging in Osteoarthritis

Osteoarthritis (OA) is the most frequent form of arthritis with major implications on both individual and public health care levels. The field of joint imaging, and particularly magnetic resonance imaging (MRI), has evolved rapidly due to the application of technical advances to the field of clinical research. This narrative review will provide an introduction to the different aspects of OA imaging aimed at an audience of scientists, clinicians, students, industry employees, and others who are interested in OA but who do not necessarily focus on OA. The current role of radiography and recent advances in measuring joint space width will be discussed. The status of cartilage morphology assessment and evaluation of cartilage biochemical composition will be presented. Advances in quantitative three-dimensional morphologic cartilage assessment and semi-quantitative whole-organ assessment of OA will be reviewed. Although MRI has evolved as the most important imaging method used in OA research, other modalities such as ultrasound, computed tomography, and metabolic imaging play a complementary role and will also be discussed.

A survey on the interpretability of deep learning in medical diagnosis

Deep learning has demonstrated remarkable performance in the medical domain, with accuracy that rivals or even exceeds that of human experts. However, it has a significant problem that these models are "black-box" structures, which means they are opaque, non-intuitive, and difficult for people to understand. This creates a barrier to the application of deep learning models in clinical practice due to lack of interpretability, trust, and transparency. To overcome this problem, several studies on interpretability have been proposed. Therefore, in this paper, we comprehensively review the interpretability of deep learning in medical diagnosis based on the current literature, including some common interpretability methods used in the medical domain, various applications with interpretability for disease diagnosis, prevalent evaluation metrics, and several disease datasets. In addition, the challenges of interpretability and future research directions are also discussed here. To the best of our knowledge, this is the first time that various applications of interpretability methods for disease diagnosis have been summarized.

Beyond automatic medical image segmentation-the spectrum between fully manual and fully automatic delineation

Semi-automatic and fully automatic contouring tools have emerged as an alternative to fully manual segmentation to reduce time spent contouring and to increase contour quality and consistency. Particularly, fully automatic segmentation has seen exceptional improvements through the use of deep learning in recent years. These fully automatic methods may not require user interactions, but the resulting contours are often not suitable to be used in clinical practice without a review by the clinician. Furthermore, they need large amounts of labelled data to be available for training. This review presents alternatives to manual or fully automatic segmentation methods along the spectrum of variable user interactivity and data availability. The challenge lies to determine how much user interaction is necessary and how this user interaction can be used most effectively. While deep learning is already widely used for fully automatic tools, interactive methods are just at the starting point to be transformed by it. Interaction between clinician and machine, via artificial intelligence, can go both ways and this review will present the avenues that are being pursued to improve medical image segmentation.

Towards a safe and efficient clinical implementation of machine learning in radiation oncology by exploring model interpretability, explainability and data-model dependency

The interest in machine learning (ML) has grown tremendously in recent years, partly due to the performance leap that occurred with new techniques of deep learning, convolutional neural networks for images, increased computational power, and wider availability of large datasets. Most fields of medicine follow that popular trend and, notably, radiation oncology is one of those that are at the forefront, with already a long tradition in using digital images and fully computerized workflows. ML models are driven by data, and in contrast with many statistical or physical models, they can be very large and complex, with countless generic parameters. This inevitably raises two questions, namely, the tight dependence between the models and the datasets that feed them, and the interpretability of the models, which scales with its complexity. Any problems in the data used to train the model will be later reflected in their performance. This, together with the low interpretability of ML models, makes their implementation into the clinical workflow particularly difficult. Building tools for risk assessment and quality assurance of ML models must involve then two main points: interpretability and data-model dependency. After a joint introduction of both radiation oncology and ML, this paper reviews the main risks and current solutions when applying the latter to workflows in the former. Risks associated with data and models, as well as their interaction, are detailed. Next, the core concepts of interpretability, explainability, and data-model dependency are formally defined and illustrated with examples. Afterwards, a broad discussion goes through key applications of ML in workflows of radiation oncology as well as vendors' perspectives for the clinical implementation of ML.

The stability of oncologic MRI radiomic features and the potential role of deep learning: a review

The use of MRI radiomic models for the diagnosis, prognosis and treatment response prediction of tumors has been increasingly reported in literature. However, its widespread adoption in clinics is hampered by issues related to features stability. In the MRI radiomic workflow, the main factors that affect radiomic features computation can be found in the image acquisition and reconstruction phase, in the image pre-processing steps, and in the segmentation of the region of interest on which radiomic indices are extracted. Deep Neural Networks (DNNs), having shown their potentiality in the medical image processing and analysis field, can be seen as an attractive strategy to partially overcome the issues related to radiomic stability and mitigate their impact. In fact, DNN approaches can be prospectively integrated in the MRI radiomic workflow to improve image quality, obtain accurate and reproducible segmentations and generate standardized images. In this review, DNN methods that can be included in the image processing steps of the radiomic workflow are described and discussed, in the light of a detailed analysis of the literature in the context of MRI radiomic reliability.

Residual RAKI: A hybrid linear and non-linear approach for scan-specific k-space deep learning

Parallel imaging is the most clinically used acceleration technique for magnetic resonance imaging (MRI) in part due to its easy inclusion into routine acquisitions. In k-space based parallel imaging reconstruction, sub-sampled k-space data are interpolated using linear convolutions. At high acceleration rates these methods have inherent noise amplification and reduced image quality. On the other hand, non-linear deep learning methods provide improved image quality at high acceleration, but the availability of training databases for different scans, as well as their interpretability hinder their adaptation. In this work, we present an extension of Robust Artificial-neural-networks for k-space Interpolation (RAKI), called residual-RAKI (rRAKI), which achieves scan-specific machine learning reconstruction using a hybrid linear and non-linear methodology. In rRAKI, non-linear CNNs are trained jointly with a linear convolution implemented via a skip connection. In effect, the linear part provides a baseline reconstruction, while the non-linear CNN that runs in parallel provides further reduction of artifacts and noise arising from the linear part. The explicit split between the linear and non-linear aspects of the reconstruction also help improve interpretability compared to purely non-linear methods. Experiments were conducted on the publicly available fastMRI datasets, as well as high-resolution anatomical imaging, comparing GRAPPA and its variants, compressed sensing, RAKI, Scan Specific Artifact Reduction in K-space (SPARK) and the proposed rRAKI. Additionally, highly-accelerated simultaneous multi-slice (SMS) functional MRI reconstructions were also performed, where the proposed rRAKI was compred to Read-out SENSE-GRAPPA and RAKI. Our results show that the proposed rRAKI method substantially improves the image quality compared to conventional parallel imaging, and offers sharper images compared to SPARK and ℓ₁-SPIRiT. Furthermore, rRAKI shows improved preservation of time-varying dynamics compared to both parallel imaging and RAKI in highly-accelerated SMS fMRI.

U-Net-Based Medical Image Segmentation

Deep learning has been extensively applied to segmentation in medical imaging. U-Net proposed in 2015 shows the advantages of accurate segmentation of small targets and its scalable network architecture. With the increasing requirements for the performance of segmentation in medical imaging in recent years, U-Net has been cited academically more than 2500 times. Many scholars have been constantly developing the U-Net architecture. This paper summarizes the medical image segmentation technologies based on the U-Net structure variants concerning their structure, innovation, efficiency, etc.; reviews and categorizes the related methodology; and introduces the loss functions, evaluation parameters, and modules commonly applied to segmentation in medical imaging, which will provide a good reference for the future research.