BJR artificial intelligence最新文献

Artificial intelligence (AI) and its computer algorithms, including those of machine learning and deep learning, have been transforming the oncology field. Generative AI (Gen AI) and general AI technologies, which can be adapted to various tasks based on diverse inputs, are being increasingly utilized in healthcare applications. Foundation models, driven by Gen AI and general AI technologies, have enthralled the medical world, representing the next generation of large-scale AI tools in medicine that are trained on massive amounts of data and will make a major impact due to their versatility, high performance, personalization enhancement, and assistance of healthcare workers. That said, it is undeniable that there are multiple obstacles that these algorithms must overcome to be successfully utilized in the field. In this review, we first present a brief background and a history of the recent rise of Gen AI and foundation models. Next, we explore their applications in the diagnosis and treatment of cancer, specifically in radiology, pathology, precision medicine, personalization of care, and surgical oncology. Then, we discuss some of the limitations that could hinder general and Gen AI's clinical translation.

Objectives: This study aims to evaluate the impact of image preprocessing methods, including traditional and artificial intelligence (AI)-based techniques, on the performance of MRI-based radiomics for predicting tumour aggressiveness in papillary thyroid carcinoma (PTC).

Methods: We retrospectively analysed MRI data from 69 patients with PTC, acquired between January 2011 and April 2023, alongside corresponding histopathology. MRI scans underwent N4 bias field correction and resampling using 10 traditional methods and an AI-based technique, synthetic multi-orientation resolution enhancement (SMORE). Radiomic features were extracted from the original and preprocessed images. Recursive feature elimination with random forests was used for feature selection, and predictive models were developed using XGBoost. The performance of the model was assessed by calculating the area under the receiver operating characteristic curve (AUC) across 1000 iterations.

Results: The combination of the correction of the bias field of N4 with SMORE resampling produced the highest mean AUC (0.75), significantly outperforming all traditional resampling methods ( $P<.001$ ). The lowest mean AUC (0.66) was observed with nearest neighbour resampling. Texture-based radiomic features, particularly the standard deviation of the grey-level co-occurrence matrix autocorrelation, were frequently selected in models using SMORE-resampled images.

Conclusions: Preprocessing techniques critically influence the predictive performance of MRI-based radiomics in PTC. The combination of N4 bias field correction and SMORE resampling enhances accuracy, highlighting the necessity of optimizing preprocessing pipelines.

Advances in knowledge: This study demonstrates the superiority of AI-driven preprocessing techniques, such as SMORE, in improving MRI radiomic models, paving the way for enhanced clinical decision-making in PTC management.

Objectives: To assess whether use of an artificial intelligence (AI) model for mammography could result in more longitudinally consistent breast density assessments compared with interpreting radiologists.

Methods: The AI model was evaluated retrospectively on a large mammography dataset including 50 sites across the United States from an outpatient radiology practice. Examinations were acquired on Hologic imaging systems between 2016 and 2021 and were interpreted by 39 radiologists (36% fellowship trained; years of experience: 2-37 years). Longitudinal patterns in 4-category breast density and binary breast density (non-dense vs. dense) were characterized for all women with at least 3 examinations (61 177 women; 214 158 examinations) as constant, descending, ascending, or bi-directional. Differences in longitudinal density patterns were assessed using paired proportion hypothesis testing.

Results: The AI model produced more constant (P < .001) and fewer bi-directional (P < .001) longitudinal density patterns compared to radiologists (AI: constant 81.0%, bi-directional 4.9%; radiologists: constant 56.8%, bi-directional 15.3%). The AI density model also produced more constant (P < .001) and fewer bi-directional (P < .001) longitudinal patterns for binary breast density. These findings held in various subset analyses, which minimize (1) change in breast density (post-menopausal women, women with stable image-based BMI), (2) inter-observer variability (same radiologist), and (3) variability by radiologist's training level (fellowship-trained radiologists).

Conclusions: AI produces more longitudinally consistent breast density assessments compared with interpreting radiologists.

Advances in knowledge: Our results extend the advantages of AI in breast density evaluation beyond automation and reproducibility, showing a potential path to improved longitudinal consistency and more consistent downstream care for screened women.

This review examines the use of large language models (LLMs) in cancer, analysing articles sourced from PubMed, Embase, and Ovid Medline, published between 2017 and 2024. Our search strategy included terms related to LLMs, cancer research, risks, safeguards, and ethical issues, focusing on studies that utilized text-based data. 59 articles were included in the review, categorized into 3 segments: quantitative studies on LLMs, chatbot-focused studies, and qualitative discussions on LLMs on cancer. Quantitative studies highlight LLMs' advanced capabilities in natural language processing (NLP), while chatbot-focused articles demonstrate their potential in clinical support and data management. Qualitative research underscores the broader implications of LLMs, including the risks and ethical considerations. Our findings suggest that LLMs, notably ChatGPT, have potential in data analysis, patient interaction, and personalized treatment in cancer care. However, the review identifies critical risks, including data biases and ethical challenges. We emphasize the need for regulatory oversight, targeted model development, and continuous evaluation. In conclusion, integrating LLMs in cancer research offers promising prospects but necessitates a balanced approach focusing on accuracy, ethical integrity, and data privacy. This review underscores the need for further study, encouraging responsible exploration and application of artificial intelligence in oncology.

Our opinion piece pays homage to the evolution of medical practices, tracing back to the era of Hippocrates, through significant historical milestones, and drawing parallels with the principles underpinning foundational artificial intelligence (AI) models. It emphasizes the shared ethos of both domains: a commitment to comprehensive care that values diverse data integration and individualized patient treatment. The excitement surrounding foundation models in medical imaging is understandable. However, a critical and cautious approach is crucial before widespread adoption. By addressing the present 4 major limitations (ie, data bias and generalizability, interpretability of AI models, data scarcity and diversity, and computational resources and infrastructure) and fostering a culture of rigorous research, we can unlock the true potential of these models and revolutionize medical care. This critique (opinion) paper highlights the need for a more measured approach in the field of foundation AI models for medicine in general and for medical imaging in particular. It emphasizes the importance of tackling core challenges before rushing toward clinical applications. By focusing on robust methodologies and addressing limitations, researchers can ensure the development of truly impactful and trustworthy models for the betterment of healthcare.

Breast cancer is one of the most common and deadly cancers in women. Triple-negative breast cancer (TNBC) accounts for approximately 10%-15% of breast cancer diagnoses and is an aggressive molecular breast cancer subtype associated with important challenges in its diagnosis, treatment, and prognostication. This poses an urgent need for developing more effective and personalized imaging biomarkers for TNBC. Towards this direction, artificial intelligence (AI) for radiologic imaging holds a prominent role, leveraging unique advantages of radiologic breast images, being used routinely for TNBC diagnosis, staging, and treatment planning, and offering high-resolution whole-tumour visualization, combined with the immense potential of AI to elucidate anatomical and functional properties of tumours that may not be easily perceived by the human eye. In this review, we synthesize the current state-of-the-art radiologic imaging applications of AI in assisting TNBC diagnosis, treatment, and prognostication. Our goal is to provide a comprehensive overview of radiomic and deep learning-based AI developments and their impact on advancing TNBC management over the last decade (2013-2024). For completeness of the review, we start with a brief introduction of AI, radiomics, and deep learning. Next, we focus on clinically relevant AI-based diagnostic, predictive, and prognostic models for radiologic breast images evaluated in TNBC. We conclude with opportunities and future directions for AI towards advancing diagnosis, treatment response predictions, and prognostic evaluations for TNBC.

Innovation in medical imaging artificial intelligence (AI)/machine learning (ML) demands extensive data collection, algorithmic advancements, and rigorous performance assessments encompassing aspects such as generalizability, uncertainty, bias, fairness, trustworthiness, and interpretability. Achieving widespread integration of AI/ML algorithms into diverse clinical tasks will demand a steadfast commitment to overcoming issues in model design, development, and performance assessment. The complexities of AI/ML clinical translation present substantial challenges, requiring engagement with relevant stakeholders, assessment of cost-effectiveness for user and patient benefit, timely dissemination of information relevant to robust functioning throughout the AI/ML lifecycle, consideration of regulatory compliance, and feedback loops for real-world performance evidence. This commentary addresses several hurdles for the development and adoption of AI/ML technologies in medical imaging. Comprehensive attention to these underlying and often subtle factors is critical not only for tackling the challenges but also for exploring novel opportunities for the advancement of AI in radiology.

Objectives: Auto-segmentation promises greater speed and lower inter-reader variability than manual segmentations in radiation oncology clinical practice. This study aims to implement and evaluate the accuracy of the auto-segmentation algorithm, "Masked Image modeling using the vision Transformers (SMIT)," for neck nodal metastases on longitudinal T₂-weighted (T₂w) MR images in oropharyngeal squamous cell carcinoma (OPSCC) patients.

Methods: This prospective clinical trial study included 123 human papillomaviruses (HPV-positive [+]) related OSPCC patients who received concurrent chemoradiotherapy. T₂w MR images were acquired on 3 T at pre-treatment (Tx), week 0, and intra-Tx weeks (1-3). Manual delineations of metastatic neck nodes from 123 OPSCC patients were used for the SMIT auto-segmentation, and total tumor volumes were calculated. Standard statistical analyses compared contour volumes from SMIT vs manual segmentation (Wilcoxon signed-rank test [WSRT]), and Spearman's rank correlation coefficients (ρ) were computed. Segmentation accuracy was evaluated on the test data set using the dice similarity coefficient (DSC) metric value. P-values <0.05 were considered significant.

Results: No significant difference in manual and SMIT delineated tumor volume at pre-Tx (8.68 ± 7.15 vs 8.38 ± 7.01 cm³, P = 0.26 [WSRT]), and the Bland-Altman method established the limits of agreement as -1.71 to 2.31 cm³, with a mean difference of 0.30 cm³. SMIT model and manually delineated tumor volume estimates were highly correlated (ρ = 0.84-0.96, P < 0.001). The mean DSC metric values were 0.86, 0.85, 0.77, and 0.79 at the pre-Tx and intra-Tx weeks (1-3), respectively.

Conclusions: The SMIT algorithm provides sufficient segmentation accuracy for oncological applications in HPV+ OPSCC.

Advances in knowledge: First evaluation of auto-segmentation with SMIT using longitudinal T₂w MRI in HPV+ OPSCC.

The adoption of artificial intelligence (AI) tools in medicine poses challenges to existing clinical workflows. This commentary discusses the necessity of context-specific quality assurance (QA), emphasizing the need for robust QA measures with quality control (QC) procedures that encompass (1) acceptance testing (AT) before clinical use, (2) continuous QC monitoring, and (3) adequate user training. The discussion also covers essential components of AT and QA, illustrated with real-world examples. We also highlight what we see as the shared responsibility of manufacturers or vendors, regulators, healthcare systems, medical physicists, and clinicians to enact appropriate testing and oversight to ensure a safe and equitable transformation of medicine through AI.