Artificial Intelligence In Medical Imaging Research Articles

The Evolution of Artificial Intelligence in Medical Imaging: From Computer Science to Machine and Deep Learning.

Artificial intelligence (AI), the wide spectrum of technologies aiming to give machines or computers the ability to perform human-like cognitive functions, began in the 1940s with the first abstract models of intelligent machines. Soon after, in the 1950s and 1960s, machine learning algorithms such as neural networks and decision trees ignited significant enthusiasm. More recent advancements include the refinement of learning algorithms, the development of convolutional neural networks to efficiently analyze images, and methods to synthesize new images. This renewed enthusiasm was also due to the increase in computational power with graphical processing units and the availability of large digital databases to be mined by neural networks. AI soon began to be applied in medicine, first through expert systems designed to support the clinician's decision and later with neural networks for the detection, classification, or segmentation of malignant lesions in medical images. A recent prospective clinical trial demonstrated the non-inferiority of AI alone compared with a double reading by two radiologists on screening mammography. Natural language processing, recurrent neural networks, transformers, and generative models have both improved the capabilities of making an automated reading of medical images and moved AI to new domains, including the text analysis of electronic health records, image self-labeling, and self-reporting. The availability of open-source and free libraries, as well as powerful computing resources, has greatly facilitated the adoption of deep learning by researchers and clinicians. Key concerns surrounding AI in healthcare include the need for clinical trials to demonstrate efficacy, the perception of AI tools as 'black boxes' that require greater interpretability and explainability, and ethical issues related to ensuring fairness and trustworthiness in AI systems. Thanks to its versatility and impressive results, AI is one of the most promising resources for frontier research and applications in medicine, in particular for oncological applications.

Semiautomated Three-Dimensional Landmark Placement on Knee Models Is a Reliable Method to Describe Bone Shape and Alignment

PurposeThe purpose of this study was to assess the inter- and intra-rater reliability of 21 anatomical landmarks initially placed with an AI-algorithm and then manually verified with human input. Methods30 knee CT scans of participants from the Multicenter Osteoarthritis Study (MOST) ages 45-55 years were included. Approximately one-half experienced progression of patellofemoral osteoarthritis, defined as an increased cartilage score in the patellofemoral compartment on MRI over 2 years. The algorithm automatically placed 19 anatomic landmarks on the femur, tibia, and patella. An additional 2 landmarks were added manually. Two landmark reviewers separately reviewed all 30 scans and verified all landmarks. After 2 weeks, one reviewer repeated the process for the same dataset. The mean Euclidean distance of manual landmark displacement, mean absolute disagreement between and within raters, and intraclass correlation coefficients for inter- and intra-rater reliability were calculated. ResultsAll landmarks had excellent inter-rater reliability. The tibial and femoral shaft centers had ICCs of 1, indicating their positions did not differ. Seventeen landmarks had ICCs between 0.90-0.99 and the tibial tuberosity had an ICC of 0.87. Intra-rater reliability scores were 1 for 16 landmarks and between 0.90-0.99 for the remaining five. ConclusionThere was excellent agreement on the locations of all 21 landmarks evaluated in this study. Clinical RelevanceThe potential role of artificial intelligence in medical imaging and orthopedic research is a growing area of interest. The excellent reliability demonstrated across multiple landmarks in our study reveals the potential for semi-automated 3D methods to enhance precision of anatomical measurements of the knee over 2D methods.

Gender and Ethnicity Bias of Text-to-Image Generative Artificial Intelligence in Medical Imaging, Part 2: Analysis of DALL-E 3.

Disparity among gender and ethnicity remains an issue across medicine and health science. Only 26%-35% of trainee radiologists are female, despite more than 50% of medical students' being female. Similar gender disparities are evident across the medical imaging professions. Generative artificial intelligence text-to-image production could reinforce or amplify gender biases. Methods: In March 2024, DALL-E 3 was utilized via GPT-4 to generate a series of individual and group images of medical imaging professionals: radiologist, nuclear medicine physician, radiographer, nuclear medicine technologist, medical physicist, radiopharmacist, and medical imaging nurse. Multiple iterations of images were generated using a variety of prompts. Collectively, 120 images were produced for evaluation of 524 characters. All images were independently analyzed by 3 expert reviewers from medical imaging professions for apparent gender and skin tone. Results: Collectively (individual and group images), 57.4% (n = 301) of medical imaging professionals were depicted as male, 42.4% (n = 222) as female, and 91.2% (n = 478) as having a light skin tone. The male gender representation was 65% for radiologists, 62% for nuclear medicine physicians, 52% for radiographers, 56% for nuclear medicine technologists, 62% for medical physicists, 53% for radiopharmacists, and 26% for medical imaging nurses. For all professions, this overrepresents men compared with women. There was no representation of persons with a disability. Conclusion: This evaluation reveals a significant overrepresentation of the male gender associated with generative artificial intelligence text-to-image production using DALL-E 3 across the medical imaging professions. Generated images have a disproportionately high representation of white men, which is not representative of the diversity of the medical imaging professions.

Gender and Ethnicity Bias of Text-to-Image Generative Artificial Intelligence in Medical Imaging, Part 1: Preliminary Evaluation.

Generative artificial intelligence (AI) text-to-image production could reinforce or amplify gender and ethnicity biases. Several text-to-image generative AI tools are used for producing images that represent the medical imaging professions. White male stereotyping and masculine cultures can dissuade women and ethnically divergent people from being drawn into a profession. Methods: In March 2024, DALL-E 3, Firefly 2, Stable Diffusion 2.1, and Midjourney 5.2 were utilized to generate a series of individual and group images of medical imaging professionals: radiologist, nuclear medicine physician, radiographer, and nuclear medicine technologist. Multiple iterations of images were generated using a variety of prompts. Collectively, 184 images were produced for evaluation of 391 characters. All images were independently analyzed by 3 reviewers for apparent gender and skin tone. Results: Collectively (individual and group characters) (n = 391), 60.6% were male and 87.7% were of a light skin tone. DALL-E 3 (65.6%), Midjourney 5.2 (76.7%), and Stable Diffusion 2.1 (56.2%) had a statistically higher representation of men than Firefly 2 (42.9%) (P < 0.0001). With Firefly 2, 70.3% of characters had light skin tones, which was statistically lower (P < 0.0001) than for Stable Diffusion 2.1 (84.8%), Midjourney 5.2 (100%), and DALL-E 3 (94.8%). Overall, image quality metrics were average or better in 87.2% for DALL-E 3 and 86.2% for Midjourney 5.2, whereas 50.9% were inadequate or poor for Firefly 2 and 86.0% for Stable Diffusion 2.1. Conclusion: Generative AI text-to-image generation using DALL-E 3 via GPT-4 has the best overall quality compared with Firefly 2, Midjourney 5.2, and Stable Diffusion 2.1. Nonetheless, DALL-E 3 includes inherent biases associated with gender and ethnicity that demand more critical evaluation.

Exploring the role of artificial intelligence in oral cancer diagnosis: Review

Oral cancer (OC) is the most common forms of head and neck cancer. Even with advancing research it still continues to have lowest survival rate globally. Histopathologiy still remains the gold standard for oral cancer diagnosis. But, it has few drawbacks being prone to errors (variability) and is time consuming leading to delay in diagnosis. In order to overcome these pitfalls there is a need for alternative diagnostic approaches which are faster and accurate at the same time. Considerable efforts have been focused on exploring the role of artificial intelligence (AI) in medical diagnostics in recent times. The role of AI in medical imaging and diagnosis seems promising. Key benefit of AI is better accuracy, rapid diagnosis and reduced manual visualization of slides. This review article focuses on the role of AI in oral cancer detection with emphasis on machine learning and deep learning methods, decision support systems. Future challenges, benefits and limitations of AI have also been discussed.

Enhanced MRI brain tumor detection and classification via topological data analysis and low-rank tensor decomposition

The advent of artificial intelligence in medical imaging has paved the way for significant advancements in the diagnosis of brain tumors. This study presents a novel ensemble approach that uses magnetic resonance imaging (MRI) to identify and categorize common brain cancers, such as pituitary, meningioma, and glioma. The proposed workflow is composed of a two-fold approach: firstly, it employs non-trivial image enhancement techniques in data preprocessing, low-rank Tucker decomposition for dimensionality reduction, and machine learning (ML) classifiers to detect and predict the type of brain tumor. Secondly, persistent homology (PH), a topological data analysis (TDA) technique, is exploited to extract potential critical areas in MRI scans. When paired with the ML classifier output, this additional information can help domain experts to identify areas of interest that might contain tumor signatures, improving the interpretability of ML predictions. When compared to automated diagnoses, this transparency adds another level of confidence and is essential for clinical acceptance. The performance of the system was quantitatively evaluated on a well-known MRI dataset, with an overall classification accuracy of 97.28% using an extremely randomized trees model. The promising results show that the integration of TDA, ML, and low-rank approximation methods is a successful approach for brain tumor identification and categorization, providing a solid foundation for further study and clinical application.

Australian healthcare workers’ views on artificial intelligence in BreastScreen: Results of a mixed method survey study

IntroductionThe evolving role of Artificial Intelligence (AI) in medicine, particularly in radiology and population-based breast cancer screening programs, offers potential accuracy gains and efficiency improvements. However, successful implementation requires understanding of healthcare workers’ views on AI, which this study aims to explore within the Australian BreastScreen program. MethodsAn online survey was distributed to clinical staff involved in breast imaging, collecting responses from November 2022 to April 2023. The survey encompassed demographic information, opinions, and experiences with AI in medical imaging, with questions covering various scenarios of AI integration in BreastScreen ResultsOut of an estimated 350 professionals contacted, 95 responded, with 84.2 % (80/95) being radiologists. Less than half of respondents (44.9 %, 40/89) had worked with artificial intelligence for image classification previously. The majority of radiologists 74.2 % (46/62) thought that the use of AI in reading mammograms for BreastScreen would improve workflow. However, radiologists thought they would behave with increasing caution with scenarios where AI was more autonomous, with the majority of radiologists (63.3 %, 38/60) uncomfortable with holding accountability when the AI was used to triage and remove cases from the workflow. Notably, 60 % of radiologists (36/60) expressed concerns about accountability. DiscussionThe findings suggest an optimistic attitude towards AI among Australian healthcare workers, although when given hypothetical scenarios for the way AI could be integrated into BreastScreen, there was increasing caution with scenarios where AI was more autonomous. This study highlights understanding and concerns of healthcare professionals working in population screening which are important to address when implementing AI into the healthcare system.

Pixels and Pitfalls: Building Robust Artificial Intelligence for Medical Imaging

Pixels and Pitfalls: Building Robust Artificial Intelligence for Medical Imaging

Trends and Public Perception of Artificial Intelligence in Medical Imaging: A Social Media Analysis.

The rapid advancement of artificial intelligence (AI) in medical imaging has generated significant interest and debate among healthcare professionals, researchers, and the general public. This study aims to explore trends and public perception of AI in medical imaging by analyzing social media discussions. Using a retrospective content analysis approach, social media posts from X (formerly known as Twitter) and Reddit were collected, covering discussions from 2019 to 2024. A total of 1,022 posts were analyzed after data cleaning, employing both qualitative and quantitative methods to examine sentiment, themes, and keyword frequencies. The sentiment analysis revealed that 55% of the comments expressed positive sentiments towards AI in medical imaging, emphasizing its potential to enhance diagnostic accuracy and efficiency. Neutral sentiments accounted for 35% of the posts, while 10% expressed negative sentiments, primarily focusing on concerns related to job displacement, ethical issues, and data privacy. Thematic analysis identified four primary themes: ethical and privacy concerns, job displacement, trust and reliability, and workflow efficiency. Keyword frequency analysis highlighted significant discussions around AI, imaging, and radiology. The results underscore both the optimism and concerns associated with AI in medical imaging, emphasizing the need for ongoing dialogue among technology developers, healthcare providers, and the public. Addressing ethical and privacy concerns, and integrating AI responsibly into clinical workflows, is crucial for maximizing its benefits and minimizing potential risks. These findings provide valuable insights into public perceptions and inform strategies for the effective and ethical implementation of AI technologies in healthcare.

The Role of Radiomics in the Prediction of Clinically Significant Prostate Cancer in the PI-RADS v2 and v2.1 Era: A Systematic Review.

Early detection of clinically significant prostate cancer (csPCa) has substantially improved with the latest PI-RADS versions. However, there is still an overdiagnosis of indolent lesions (iPCa), and radiomics has emerged as a potential solution. The aim of this systematic review is to evaluate the role of handcrafted and deep radiomics in differentiating lesions with csPCa from those with iPCa and benign lesions on prostate MRI assessed with PI-RADS v2 and/or 2.1. The literature search was conducted in PubMed, Cochrane, and Web of Science databases to select relevant studies. Quality assessment was carried out with Quality Assessment of Diagnostic Accuracy Studies 2 (QUADAS-2), Radiomic Quality Score (RQS), and Checklist for Artificial Intelligence in Medical Imaging (CLAIM) tools. A total of 14 studies were deemed as relevant from 411 publications. The results highlighted a good performance of handcrafted and deep radiomics methods for csPCa detection, but without significant differences compared to radiologists (PI-RADS) in the few studies in which it was assessed. Moreover, heterogeneity and restrictions were found in the studies and quality analysis, which might induce bias. Future studies should tackle these problems to encourage clinical applicability. Prospective studies and comparison with radiologists (PI-RADS) are needed to better understand its potential.

Evaluating artificial intelligence for medical imaging: a primer for clinicians.

Artificial intelligence has the potential to transform medical imaging. The effective integration of artificial intelligence into clinical practice requires a robust understanding of its capabilities and limitations. This paper begins with an overview of key clinical use cases such as detection, classification, segmentation and radiomics. It highlights foundational concepts in machine learning such as learning types and strategies, as well as the training and evaluation process. We provide a broad theoretical framework for assessing the clinical effectiveness of medical imaging artificial intelligence, including appraising internal validity and generalisability of studies, and discuss barriers to clinical translation. Finally, we highlight future directions of travel within the field including multi-modal data integration, federated learning and explainability. By having an awareness of these issues, clinicians can make informed decisions about adopting artificial intelligence for medical imaging, improving patient care and clinical outcomes.

Artificial intelligence/machine learning for neuroimaging to predict hemorrhagic transformation: Systematic review/meta-analysis.

Early and reliable prediction of hemorrhagic transformation (HT) in patients with acute ischemic stroke (AIS) is crucial for treatment decisions and early intervention. The purpose of this study was to conduct a systematic review and meta-analysis on the performance of artificial intelligence (AI) and machine learning (ML) models that utilize neuroimaging to predict HT. A systematic search of PubMed, EMBASE, and Web of Science was conducted until February 19, 2024. Inclusion criteria were as follows: patients with AIS who received reperfusion therapy; AI/ML algorithm using imaging to predict HT; or presence of sufficient data on the predictive performance. Exclusion criteria were as follows: articles with less than 20 patients; articles lacking algorithms that operate solely on images; or articles not detailing the algorithm used. The quality of eligible studies was assessed using the Quality Assessment of Diagnostic Accuracy Studies-2 and Checklist for Artificial Intelligence in Medical Imaging. Pooled sensitivity, specificity, and diagnostic odds ratio (DOR) were calculated using a random-effects model, and a summary receiver operating characteristic curve was constructed using the Reitsma method. We identified six eligible studies, which included 1640 patients. Aside from an unclear risk of bias regarding flow and timing identified in two of the studies, all studies showed low risk of bias and applicability concerns in all categories. Pooled sensitivity, specificity, and DOR were.849,.878, and 45.598, respectively. AI/ML models can reliably predict the occurrence of HT in AIS patients. More prospective studies are needed for subgroup analyses and higher clinical certainty and usefulness.

Application of Medical Image Artificial Intelligence in Medical Diagnosis and Treatment

Application of Medical Image Artificial Intelligence in Medical Diagnosis and Treatment

Deep learning in pulmonary nodule detection and segmentation: a systematic review.

The accurate detection and precise segmentation of lung nodules on computed tomography are key prerequisites for early diagnosis and appropriate treatment of lung cancer. This study was designed to compare detection and segmentation methods for pulmonary nodules using deep-learning techniques to fill methodological gaps and biases in the existing literature. This study utilized a systematic review with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines, searching PubMed, Embase, Web of Science Core Collection, and the Cochrane Library databases up to May 10, 2023. The Quality Assessment of Diagnostic Accuracy Studies 2 criteria was used to assess the risk of bias and was adjusted with the Checklist for Artificial Intelligence in Medical Imaging. The study analyzed and extracted model performance, data sources, and task-focus information. After screening, we included nine studies meeting our inclusion criteria. These studies were published between 2019 and 2023 and predominantly used public datasets, with the Lung Image Database Consortium Image Collection and Image Database Resource Initiative and Lung Nodule Analysis 2016 being the most common. The studies focused on detection, segmentation, and other tasks, primarily utilizing Convolutional Neural Networks for model development. Performance evaluation covered multiple metrics, including sensitivity and the Dice coefficient. This study highlights the potential power of deep learning in lung nodule detection and segmentation. It underscores the importance of standardized data processing, code and data sharing, the value of external test datasets, and the need to balance model complexity and efficiency in future research. Deep learning demonstrates significant promise in autonomously detecting and segmenting pulmonary nodules. Future research should address methodological shortcomings and variability to enhance its clinical utility. Deep learning shows potential in the detection and segmentation of pulmonary nodules. There are methodological gaps and biases present in the existing literature. Factors such as external validation and transparency affect the clinical application.

Managing The Ethical Terrain Of AI In Radiography: A Cross-Sectional Investigation Of Radiographers' Viewpoints

Artificial intelligence (AI) in radiography raises difficult ethical issues and offers revolutionary possibilities for diagnostic imaging. This cross-sectional study set out to find out how radiographers felt about the ethical ramifications of AI in their line of work, as well as to pinpoint the main issues and possible solutions.Methods: A structured questionnaire was given to a wide range of Saudi Arabian radiologists. Questions about ethical AI concerns, the perceived effect on clinical practice, and recommendations for moral AI integration in radiography were all covered in the survey. Both quantitative and qualitative techniques were used to analyze the data in order to capture a wide variety of viewpoints.Conclusions: This study uncovered a complicated ethical environment surrounding the use of AI in radiography, one that is marked by professionals' excitement and trepidation. It emphasizes how ethical frameworks, instruction, and the creation of policies are required to direct the application of AI in radiography. These findings add to the current discussion about AI in medical imaging and offer guidance to practitioners, educators, and policymakers on how to handle the moral dilemmas associated with the use of AI in healthcare.

Bias in artificial intelligence for medical imaging: fundamentals, detection, avoidance, mitigation, challenges, ethics, and prospects.

Although artificial intelligence (AI) methods hold promise for medical imaging-based prediction tasks, their integration into medical practice may present a double-edged sword due to bias (i.e., systematic errors). AI algorithms have the potential to mitigate cognitive biases in human interpretation, but extensive research has highlighted the tendency of AI systems to internalize biases within their model. This fact, whether intentional or not, may ultimately lead to unintentional consequences in the clinical setting, potentially compromising patient outcomes. This concern is particularly important in medical imaging, where AI has been more progressively and widely embraced than any other medical field. A comprehensive understanding of bias at each stage of the AI pipeline is therefore essential to contribute to developing AI solutions that are not only less biased but also widely applicable. This international collaborative review effort aims to increase awareness within the medical imaging community about the importance of proactively identifying and addressing AI bias to prevent its negative consequences from being realized later. The authors began with the fundamentals of bias by explaining its different definitions and delineating various potential sources. Strategies for detecting and identifying bias were then outlined, followed by a review of techniques for its avoidance and mitigation. Moreover, ethical dimensions, challenges encountered, and prospects were discussed.

Artificial Intelligence in Medical Imaging: Applications of Deep Learning for Disease Detection and Diagnosis

The integration of artificial intelligence (AI) and deep learning techniques into medical imaging has revolutionized disease detection and diagnosis. This paper provides a comprehensive overview of the applications of deep learning in medical imaging and its impact on healthcare. The paper begins with an introduction to the fundamentals of deep learning, emphasizing convolutional neural networks (CNNs) and their relevance in analyzing medical images. It then explores various applications of deep learning in medical imaging, including automated disease detection and classification, image segmentation for precise anatomical localization, quantitative analysis for predictive modeling, personalized medicine, and workflow optimization. Case studies and examples from different medical specialties, such as oncology, cardiology, and neurology, are presented to illustrate the practical implementation and effectiveness of AI-driven approaches.

Towards objective and systematic evaluation of bias in artificial intelligence for medical imaging.

Artificial intelligence (AI) models trained using medical images for clinical tasks often exhibit bias in the form of subgroup performance disparities. However, since not all sources of bias in real-world medical imaging data are easily identifiable, it is challenging to comprehensively assess their impacts. In this article, we introduce an analysis framework for systematically and objectively investigating the impact of biases in medical images on AI models. Our framework utilizes synthetic neuroimages with known disease effects and sources of bias. We evaluated the impact of bias effects and the efficacy of 3 bias mitigation strategies in counterfactual data scenarios on a convolutional neural network (CNN) classifier. The analysis revealed that training a CNN model on the datasets containing bias effects resulted in expected subgroup performance disparities. Moreover, reweighing was the most successful bias mitigation strategy for this setup. Finally, we demonstrated that explainable AI methods can aid in investigating the manifestation of bias in the model using this framework. The value of this framework is showcased in our findings on the impact of bias scenarios and efficacy of bias mitigation in a deep learning model pipeline. This systematic analysis can be easily expanded to conduct further controlled in silico trials in other investigations of bias in medical imaging AI. Our novel methodology for objectively studying bias in medical imaging AI can help support the development of clinical decision-support tools that are robust and responsible.

The limits of fair medical imaging AI in real-world generalization

As artificial intelligence (AI) rapidly approaches human-level performance in medical imaging, it is crucial that it does not exacerbate or propagate healthcare disparities. Previous research established AI’s capacity to infer demographic data from chest X-rays, leading to a key concern: do models using demographic shortcuts have unfair predictions across subpopulations? In this study, we conducted a thorough investigation into the extent to which medical AI uses demographic encodings, focusing on potential fairness discrepancies within both in-distribution training sets and external test sets. Our analysis covers three key medical imaging disciplines—radiology, dermatology and ophthalmology—and incorporates data from six global chest X-ray datasets. We confirm that medical imaging AI leverages demographic shortcuts in disease classification. Although correcting shortcuts algorithmically effectively addresses fairness gaps to create ‘locally optimal’ models within the original data distribution, this optimality is not true in new test settings. Surprisingly, we found that models with less encoding of demographic attributes are often most ‘globally optimal’, exhibiting better fairness during model evaluation in new test environments. Our work establishes best practices for medical imaging models that maintain their performance and fairness in deployments beyond their initial training contexts, underscoring critical considerations for AI clinical deployments across populations and sites.

Reconciling privacy and accuracy in AI for medical imaging

Artificial intelligence (AI) models are vulnerable to information leakage of their training data, which can be highly sensitive, for example, in medical imaging. Privacy-enhancing technologies, such as differential privacy (DP), aim to circumvent these susceptibilities. DP is the strongest possible protection for training models while bounding the risks of inferring the inclusion of training samples or reconstructing the original data. DP achieves this by setting a quantifiable privacy budget. Although a lower budget decreases the risk of information leakage, it typically also reduces the performance of such models. This imposes a trade-off between robust performance and stringent privacy. Additionally, the interpretation of a privacy budget remains abstract and challenging to contextualize. Here we contrast the performance of artificial intelligence models at various privacy budgets against both theoretical risk bounds and empirical success of reconstruction attacks. We show that using very large privacy budgets can render reconstruction attacks impossible, while drops in performance are negligible. We thus conclude that not using DP at all is negligent when applying artificial intelligence models to sensitive data. We deem our results to lay a foundation for further debates on striking a balance between privacy risks and model performance.