Osteosarcoma (OS) is the most common primary bone malignancy, and neoadjuvant chemotherapy (NAC) improves survival rates. However, OS heterogeneity results in variable treatment responses, highlighting the need for reliable, non-invasive tools to predict NAC response. Radiomics-based machine learning (ML) offers potential for identifying imaging biomarkers to predict treatment outcomes. This systematic review and meta-analysis evaluated the accuracy and reliability of radiomics models for predicting NAC response in OS.

A systematic search was conducted in PubMed, Embase, Scopus, and Web of Science up to November 2024. Studies using radiomics-based ML for NAC response prediction in OS were included. Pooled sensitivity, specificity, and AUC for training and validation cohorts were calculated using bivariate random-effects modeling, with clinical-combined models analyzed separately. Quality assessment was performed using the QUADAS-2 tool, radiomics quality score (RQS), and METRICS scores.

Sixteen studies were included, with 63 % using MRI and 37 % using CT. Twelve studies, comprising 1639 participants, were included in the meta-analysis. Pooled metrics for training cohorts showed an AUC of 0.93, sensitivity of 0.89, and specificity of 0.85. Validation cohorts achieved an AUC of 0.87, sensitivity of 0.81, and specificity of 0.82. Clinical-combined models outperformed radiomics-only models. The mean RQS score was 9.44 ± 3.41, and the mean METRICS score was 60.8 % ± 17.4 %.

Radiomics-based ML shows promise for predicting NAC response in OS, especially when combined with clinical indicators. However, limitations in external validation and methodological consistency must be addressed.

To assess the effect of different intensity standardisation techniques (ISTs) and ComBat batch sizes on radiomics survival model performance and stability in a heterogenous, multi-centre cohort of patients with glioblastoma (GBM).

Multi-centre pre-operative MRI acquired between 2014 and 2020 in patients with IDH-wildtype unifocal WHO grade 4 GBM were retrospectively evaluated. WhiteStripe (WS), Nyul histogram matching (HM), and Z-score (ZS) ISTs were applied before radiomic feature (RF) extraction. RFs were realigned using ComBat and minimum batch size (MBS) of 5, 10, or 15 patients. Cox proportional hazards models for overall survival (OS) prediction were produced using five different selection strategies and the impact of IST and MBS was evaluated using bootstrapping. Calibration, discrimination, relative explained variation, and model fit were assessed. Instability was evaluated using 95% confidence intervals (95% CIs), feature selection frequency and calibration curves across the bootstrap resamples.

One hundred ninety-five patients were included. Median OS = 13 (95% CI: 12-14) months. Twelve to fourteen unique MRI protocols were used per MRI sequence. HM and WS produced the highest relative increase in model discrimination, explained variation and model fit but IST choice did not greatly impact on stability, nor calibration. Larger ComBat batches improved discrimination, model fit, and explained variation but higher MBS (reduced sample size) reduced stability (across all performance metrics) and reduced calibration accuracy.

Heterogenous, real-world GBM data poses a challenge to the reproducibility of radiomics. ComBat generally improved model performance as MBS increased but reduced stability and calibration. HM and WS tended to improve model performance.

Question ComBat harmonisation of RFs and intensity standardisation of MRI have not been thoroughly evaluated in multicentre, heterogeneous GBM data. Findings The addition of ComBat and ISTs can improve discrimination, relative model fit, and explained variance but degrades the calibration and stability of survival models. Clinical relevance Radiomics risk prediction models in real-world, multicentre contexts could be improved by ComBat and ISTs, however, this degrades calibration and prediction stability and this must be thoroughly investigated before patients can be accurately separated into different risk groups.

The last decade has witnessed a surge in artificial intelligence (AI). With bone metastases becoming more prevalent, there is an increasing call for personalized treatment options, a domain where AI can greatly contribute. However, integrating AI into clinical settings has proven to be difficult. Therefore, we aimed to provide an overview of AI modalities for treating bone metastases and recommend implementation-worthy models based on TRIPOD, CLAIM, and UPM scores.

This systematic review included 92 studies on AI models in bone metastases between 2008 and 2024. Using three assessment tools we provided a reliable foundation for recommending AI modalities fit for clinical use (TRIPOD or CLAIM ≥ 70 % and UPM score ≥ 10).

Most models focused on survival prediction (44/92;48%), followed by imaging studies (37/92;40%). Median TRIPOD completeness was 70% (IQR 64-81%), CLAIM completeness was 57% (IQR 48-67%), and UPM score was 7 (IQR 5-9). In total, 10% (9/92) AI modalities were deemed fit for clinical use.

Transparent reporting, utilizing the aforementioned three evaluation tools, is essential for effectively integrating AI models into clinical practice, as currently, only 10% of AI models for bone metastases are deemed fit for clinical use. Such transparency ensures that both patients and clinicians can benefit from clinically useful AI models, potentially enhancing AI-driven personalized cancer treatment.

Advancements in magnetic resonance imaging (MRI) analysis over the past decades have significantly reshaped the field of neuroradiology. The ability to extract multiple quantitative measures from each MRI scan, alongside the development of extensive data repositories, has been fundamental to the emergence of advanced methodologies such as radiomics and artificial intelligence (AI). This educational review aims to delineate the importance of image analysis, highlight key paradigm shifts, examine their implications, and identify existing constraints that must be addressed to facilitate integration into clinical practice. Particular attention is given to aiding junior neuroradiologists in navigating this complex and evolving landscape.

A comprehensive review of the available analysis toolboxes was conducted, focusing on major technological advancements in MRI analysis, the evolution of data repositories, and the rise of AI and radiomics in neuroradiology. Stakeholders within the field were identified and their roles examined. Additionally, current challenges and barriers to clinical implementation were analyzed.

The analysis revealed several pivotal shifts, including the transition from qualitative to quantitative imaging, the central role of large datasets in developing AI tools, and the growing importance of interdisciplinary collaboration. Key stakeholders-including academic institutions, industry partners, regulatory bodies, and clinical practitioners-were identified, each playing a distinct role in advancing the field. However, significant barriers remain, particularly regarding standardization, data sharing, regulatory approval, and integration into clinical workflows.

While advancements in MRI analysis offer tremendous potential to enhance neuroradiology practice, realizing this potential requires overcoming technical, regulatory, and practical barriers. Education and structured support for junior neuroradiologists are essential to ensure they are well-equipped to participate in and drive future developments. A coordinated effort among stakeholders is crucial to facilitate the seamless translation of these technological innovations into everyday clinical practice.

This study evaluated the performance of a multi-stage Segmentation Residual Network (SegResNet)-based deep learning (DL) model for the automatic segmentation of cone-beam computed tomography (CBCT) images of patients with stage III and IV periodontitis.

Seventy pre-processed CBCT scans from patients undergoing periodontal rehabilitation were used for training and validation. The model was tested on 10 CBCT scans independent from the training dataset by comparing results with semi-automatic (SA) segmentations. Segmentation accuracy was assessed using the Dice similarity coefficient (DSC), Intersection over Union (IoU), and Hausdorff distance 95th percentile (HD95). Linear periodontal measurements were performed on four tooth surfaces to assess the validity of the DL segmentation in the periodontal region.

The DL model achieved a mean DSC of 0.9650 ± 0.0097, with an IoU of 0.9340 ± 0.0180 and HD95 of 0.4820 mm ± 0.1269 mm, showing strong agreement with SA segmentation. Linear measurements revealed high statistical correlations between the mesial, distal, and lingual surfaces, with intraclass correlation coefficients (ICC) of 0.9442 (p<0.0001), 0.9232 (p<0.0001), and 0.9598(p<0.0001), respectively, while buccal measurements revealed lower consistency, with an ICC of 0.7481 (p<0.0001). The DL method reduced the segmentation time by 47 times compared to the SA method.

Acquired 3D models may enable precise treatment planning in cases where conventional diagnostic modalities are insufficient. However, the robustness of the model must be increased to improve its general reliability and consistency at the buccal aspect of the periodontal region.

This study presents a DL model for the CBCT-based segmentation of periodontal defects, demonstrating high accuracy and a 47-fold time reduction compared to SA methods, thus improving the feasibility of 3D diagnostics for advanced periodontitis.

Background/Objectives: Large language models (LLMs), such as ChatGPT, have emerged as potential clinical support tools to enhance precision in personalized patient care, but their reliability in radiological image interpretation remains uncertain. The primary aim of our study was to evaluate the diagnostic accuracy of ChatGPT-4o in interpreting chest X-rays (CXRs) and abdominal X-rays (AXRs) by comparing its performance to expert radiology findings, whilst secondary aims were diagnostic confidence and patient safety. Methods: A total of 500 X-rays, including 257 CXR (51.4%) and 243 AXR (48.5%), were analyzed. Diagnoses made by ChatGPT-4o were compared to expert interpretations. Confidence scores (1-4) were assigned and responses were evaluated for patient safety. Results: ChatGPT-4o correctly identified 345 of 500 (69%) pathologies (95% CI: 64.81-72.9). For AXRs 175 of 243 (72.02%) pathologies were correctly diagnosed (95% CI: 66.06-77.28), while for CXRs 170 of 257 (66.15%) were accurate (95% CI: 60.16-71.66). The highest detection rates among CXRs were observed for pulmonary edema, tumor, pneumonia, pleural effusion, cardiomegaly, and emphysema, and lower rates were observed for pneumothorax, rib fractures, and enlarged mediastinum. AXR performance was highest for intestinal obstruction and foreign bodies, and weaker for pneumoperitoneum, renal calculi, and diverticulitis. Confidence scores were higher for AXRs (mean 3.45 ± 1.1) than CXRs (mean 2.48 ± 1.45). All responses (100%) were considered to be safe for the patient. Interobserver agreement was high (kappa = 0.920), and reliability (second prompt) was moderate (kappa = 0.750). Conclusions: ChatGPT-4o demonstrated moderate accuracy for the interpretation of X-rays, being higher for AXRs compared to CXRs. Improvements are required for its use as efficient clinical support tool.

Artificial intelligence (AI) is a revolutionary tool yet to be fully integrated into several health care sectors, including medical imaging. AI can transform how medical imaging is conducted and interpreted, especially in cardio-oncology.

This study aims to systematically review the available literature on the use of AI in cardio-oncology imaging to predict cardiotoxicity and describe the possible improvement of different imaging modalities that can be achieved if AI is successfully deployed to routine practice.

We conducted a database search in PubMed, Ovid MEDLINE, Cochrane Library, CINAHL, and Google Scholar from inception to 2023 using the AI research assistant tool (Elicit) to search for original studies reporting AI outcomes in adult patients diagnosed with any cancer and undergoing cardiotoxicity assessment. Outcomes included incidence of cardiotoxicity, left ventricular ejection fraction, risk factors associated with cardiotoxicity, heart failure, myocardial dysfunction, signs of cancer therapy-related cardiovascular toxicity, echocardiography, and cardiac magnetic resonance imaging. Descriptive information about each study was recorded, including imaging technique, AI model, outcomes, and limitations.

The systematic search resulted in 7 studies conducted between 2018 and 2023, which are included in this review. Most of these studies were conducted in the United States (71%), included patients with breast cancer (86%), and used magnetic resonance imaging as the imaging modality (57%). The quality assessment of the studies had an average of 86% compliance in all of the tool's sections. In conclusion, this systematic review demonstrates the potential of AI to enhance cardio-oncology imaging for predicting cardiotoxicity in patients with cancer.

Our findings suggest that AI can enhance the accuracy and efficiency of cardiotoxicity assessments. However, further research through larger, multicenter trials is needed to validate these applications and refine AI technologies for routine use, paving the way for improved patient outcomes in cancer survivors at risk of cardiotoxicity.

Privacy concerns, such as identifiable facial features within brain scans, have hindered the availability of pediatric neuroimaging data sets for research. Consequently, pediatric neuroscience research lags adult counterparts, particularly in rare disease and under-represented populations. The removal of face regions (image defacing) can mitigate this; however, existing defacing tools often fail with pediatric cases and diverse image types, leaving a critical gap in data accessibility. Given recent National Institutes of Health data sharing mandates, novel solutions are a critical need.

To develop an artificial intelligence (AI)-powered tool for automatic defacing of pediatric brain MRIs, deep learning methodologies (nnU-Net) were used by using a large, diverse multi-institutional data set of clinical radiology images. This included multiparametric MRIs (T1-weighted [T1W], T1W-contrast-enhanced, T2-weighted [T2W], T2W-FLAIR) with 976 total images from 208 patients with brain tumor (Children's Brain Tumor Network, CBTN) and 36 clinical control patients (Scans with Limited Imaging Pathology, SLIP) ranging in age from 7 days to 21 years old.

Face and ear removal accuracy for withheld testing data were the primary measure of model performance. Potential influences of defacing on downstream research usage were evaluated with standard image processing and AI-based pipelines. Group-level statistical trends were compared between original (nondefaced) and defaced images. Across image types, the model had high accuracy for removing face regions (mean accuracy, 98%; n=98 subjects/392 images), with lower performance for removal of ears (73%). Analysis of global and regional brain measures (SLIP cohort) showed minimal differences between original and defaced outputs (mean r S = 0.93, all P < .0001). AI-generated whole brain and tumor volumes (CBTN cohort) and temporalis muscle metrics (volume, cross-sectional area, centile scores; SLIP cohort) were not significantly affected by image defacing (all r S > 0.9, P < .0001).

The defacing model demonstrates efficacy in removing facial regions across multiple MRI types and exhibits minimal impact on downstream research usage. A software package with the trained model is freely provided for wider use and further development (pediatric-auto-defacer; https://github.com/d3b-center/pediatric-auto-defacer-public). By offering a solution tailored to pediatric cases and multiple MRI sequences, this defacing tool will expedite research efforts and promote broader adoption of data sharing practices within the neuroscience community.

High dose rate (HDR) prostate brachytherapy (BT) procedure requires image-guided needle insertion. Given that general anesthesia is often employed during the procedure, minimizing overall planning time is crucial. In this study, we explore the clinical feasibility and time-saving potential of artificial intelligence (AI)-driven auto-reconstruction of transperineal needles in the context of ultrasound (US)-guided prostate BT planning.

This study included a total of 102 US-planned BT images from a single institution and split into 3 groups: 50 for model training and validation, 11 to evaluate reconstruction accuracy (test set), and 41 to evaluate the AI tool in a clinical implementation (clinical set). Reconstruction accuracy for the test set was evaluated by comparing the performance of AI-derived and manually reconstructed needles from 5 medical physicists on the 3D-US scans after treatment. The needle total reconstruction time for the clinical set was defined as the timestamp difference from scan acquisition to the start of dose calculations and was compared with values recorded before the clinical implementation of the AI-assisted tool.

A mean error of (0.44 ± 0.32) mm was found between the AI-reconstructed and the human consensus needle positions in the test set, with 95.0% of AI needle points falling below 1 mm from their human-made counterparts. Post-hoc analysis showed that only one of the human observers' reconstructions were significantly different from the others including the AIs. In the clinical set, the AI algorithm achieved a true positive reconstruction rate of 93.4% with only 4.5% of these needles requiring manual corrections from the planner before dosimetry. Total time required to perform AI-assisted catheter reconstruction on clinical cases was on average 15.2 min lower (P < .01) compared with procedure without AI assistance.

This study demonstrates the feasibility of an AI-assisted needle reconstructing tool for 3D-US-based HDR prostate BT. This is a step toward treatment planning automation and increased efficiency in HDR prostate BT.

Purpose: To evaluate the accuracy of a Bidirectional Encoder Representations for Transformers (BERT) Natural Language Processing (NLP) model for automating triage and protocol selection of cross-sectional image requisitions. Methods: A retrospective study was completed using 222 392 CT and MRI studies from a single Canadian university hospital database (January 2018-September 2022). Three hundred unique protocols (116 CT and 184 MRI) were included. A BERT model was trained, validated, and tested using an 80%-10%-10% stratified split. Naive Bayes (NB) and Support Vector Machine (SVM) machine learning models were used as comparators. Models were assessed using F1 score, precision, recall, and area under the receiver operating characteristic curve (AUROC). The BERT model was also assessed for multi-class protocol suggestion and subgroups based on referral location, modality, and imaging section. Results: BERT was superior to SVM for protocol selection (F1 score: BERT-0.901 vs SVM-0.881). However, was not significantly different from SVM for triage prediction (F1 score: BERT-0.844 vs SVM-0.845). Both models outperformed NB for protocol and triage. BERT had superior performance on minority classes compared to SVM and NB. For multiclass prediction, BERT accuracy was up to 0.991 for top-5 protocol suggestion, and 0.981 for top-2 triage suggestion. Emergency department patients had the highest F1 scores for both protocol (0.957) and triage (0.986), compared to inpatients and outpatients. Conclusion: The BERT NLP model demonstrated strong performance in automating the triage and protocol selection of radiology studies, showing potential to enhance radiologist workflows. These findings suggest the feasibility of using advanced NLP models to streamline radiology operations.

Despite growing awareness of problems with fairness in artificial intelligence (AI) models in radiology, evaluation of algorithmic biases, or AI biases, remains challenging due to various complexities. These include incomplete reporting of demographic information in medical imaging datasets, variability in definitions of demographic categories, and inconsistent statistical definitions of bias. To guide the appropriate evaluation of AI biases in radiology, this article summarizes the pitfalls in the evaluation and measurement of algorithmic biases. These pitfalls span the spectrum from the technical (eg, how different statistical definitions of bias impact conclusions about whether an AI model is biased) to those associated with social context (eg, how different conventions of race and ethnicity impact identification or masking of biases). Actionable best practices and future directions to avoid these pitfalls are summarized across three key areas: (a) medical imaging datasets, (b) demographic definitions, and (c) statistical evaluations of bias. Although AI bias in radiology has been broadly reviewed in the recent literature, this article focuses specifically on underrecognized potential pitfalls related to the three key areas. By providing awareness of these pitfalls along with actionable practices to avoid them, exciting AI technologies can be used in radiology for the good of all people.

Gynecological cancer is the most common cancer that affects women's quality of life and well-being. Artificial intelligence (AI) technology enables us to exploit high-dimensional imaging data for precision oncology. Tremendous progress has been made with AI radiomics in cancers such as lung and breast cancers. Herein, we performed a literature review on AI radiomics in the management of gynecological cancer.

A search was performed in the databases of PubMed, Embase, and Web of Science for original articles written in English up to 10 September 2024, using the terms "gynecological cancer", "cervical cancer", "endometrial cancer", "ovarian cancer", AND "artificial intelligence", "AI", AND "radiomics". The included studies mainly focused on the current landscape of AI radiomics in the diagnosis, treatment, and prognosis of gynecological cancer.

A total of 128 studies were included, with 86 studies focusing on tumor diagnosis (n=23) and characterization (n=63), 15 on treatment response prediction, and 27 on recurrence and survival prediction. AI radiomics has shown potential value in tumor diagnosis and characterization [tumor staging, histological subtyping, lymph node metastasis (LNM), lymphovascular space invasion (LVSI), myometrial invasion (MI), and other molecular or clinicopathological factors], chemotherapy or chemoradiotherapy response evaluation, and prognosis (disease recurrence or metastasis, and survival) prediction. However, most included studies were single-center and retrospective. There was substantial heterogeneity in methodology and results reporting.

AI radiomics has been increasingly adopted in the management of gynecological cancer. Further validation in large-scale datasets is needed before clinical translation.

Machine learning (ML) in surgical video analysis offers promising prospects for training and decision support in surgery. The past decade has seen key advances in ML-based operative workflow analysis, though existing applications mostly feature shorter surgeries (<2 hours) with limited scene changes. The aim of this study was to develop and evaluate a ML model capable of automated operative workflow recognition for retrosigmoid vestibular schwannoma (VS) resection. In doing so, this project furthers previous research by applying workflow prediction platforms to lengthy (median >5 hours duration), data-heavy surgeries, using VS resection as an exemplar.

A video dataset of 21 microscopic retrosigmoid VS resections was collected at a single institution over 3 years and underwent workflow annotation according to a previously agreed expert consensus (Approach, Excision, and Closure phases; and Debulking or Dissection steps within the Excision phase). Annotations were used to train a ML model consisting of a convolutional neural network and a recurrent neural network. 5-fold cross-validation was used, and performance metrics (accuracy, precision, recall, F1 score) were assessed for phase and step prediction.

Median operative video time was 5 hours 18 minutes (IQR 3 hours 21 minutes-6 hours 1 minute). The "Tumor Excision" phase accounted for the majority of each case (median 4 hours 23 minutes), whereas "Approach and Exposure" (28 minutes) and "Closure" (17 minutes) comprised shorter phases. The ML model accurately predicted operative phases (accuracy 81%, weighted F1 0.83) and dichotomized steps (accuracy 86%, weighted F1 0.86).

This study demonstrates that our ML model can accurately predict the surgical phases and intraphase steps in retrosigmoid VS resection. This demonstrates the successful application of ML in operative workflow recognition on low-volume, lengthy, data-heavy surgical videos. Despite this, there remains room for improvement in individual step classification. Future applications of ML in low-volume high-complexity operations should prioritize collaborative video sharing to overcome barriers to clinical translation.

Partial hepatectomy continues to be the primary treatment approach for liver tumors, and post-hepatectomy liver failure (PHLF) remains the most critical life-threatening complication following surgery.

To comprehensively review the PHLF prognostic models developed in recent years and objectively assess the risk of bias in these models.

This review followed the Checklist for Critical Appraisal and Data Extraction for Systematic Reviews of Prediction Modelling Studies and the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guideline. Three databases were searched from November 2019 to December 2022, and references as well as cited literature in all included studies were manually screened in March 2023. Based on the defined inclusion criteria, articles on PHLF prognostic models were selected, and data from all included articles were extracted by two independent reviewers. The PROBAST was used to evaluate the quality of each included article.

A total of thirty-four studies met the eligibility criteria and were included in the analysis. Nearly all of the models (32/34, 94.1%) were developed and validated exclusively using private data sources. Predictive variables were categorized into five distinct types, with the majority of studies (32/34, 94.1%) utilizing multiple types of data. The area under the curve for the training models included ranged from 0.697 to 0.956. Analytical issues resulted in a high risk of bias across all studies included.

The validation performance of the existing models was substantially lower compared to the development models. All included studies were evaluated as having a high risk of bias, primarily due to issues within the analytical domain. The progression of modeling technology, particularly in artificial intelligence modeling, necessitates the use of suitable quality assessment tools.

Digital medicine and smart healthcare will not be realised without the cognizant participation of clinicians. Artificial intelligence (AI) today primarily involves computers or machines designed to simulate aspects of human intelligence using mathematically designed neural networks, although early AI systems relied on a variety of non-neural network techniques. With the increased complexity of the neural layers, deep machine learning (ML) can self-learn and augment many human tasks that require decision-making on the basis of multiple sources of data. Clinicians are important stakeholders in the use of AI and ML tools. The review questions are as follows: What is the typical process of AI tool development in the full cycle? What are the important concepts and technical aspects of each step? This review synthesises a targeted literature review and reports and summarises online structured materials to present a succinct explanation of the whole development process of AI tools. The development of AI tools in healthcare involves a series of cyclical processes: (1) identifying clinical problems suitable for AI solutions, (2) forming project teams or collaborating with experts, (3) organising and curating relevant data, (4) establishing robust physical and virtual infrastructure, and computer systems' architecture that support subsequent stages, (5) exploring AI neural networks on open access platforms before making a new decision, (6) validating AI/ML models, (7) registration, (8) clinical deployment and continuous performance monitoring and (9) improving the AI ecosystem ensures its adaptability to evolving clinical needs. A sound understanding of this would help clinicians appreciate the development of AI tools and engage in codesigning, evaluating and monitoring the tools. This would facilitate broader use and closer regulation of AI/ML tools in healthcare settings.

We created and validated an open-access AI algorithm (AIc) for assessing image segmentation and patient centering in a multi-body-region, multi-center, and multi-scanner study. Our study included 825 head, chest, and abdomen-pelvis CT from 275 patients (153 females, 128 males; mean age 67 ± 14 years) scanned at five academic and community hospitals. CT images were processed with the AIc to determine vertical and horizontal centering at the skull base (head CT), carina (chest CT), and L2-L3 disc (abdomen CT). We manually measured the vertical and horizontal off-centering. We found strong correlations between AIc and manual estimate of off-centering in both the vertical (head, r = 0.93; chest, r = 0.94; abdomen, and r = 0.95) and horizontal directions (head CT, r = 0.85; chest, r = 0.85; abdomen, r = 0.8) and across age groups (r = 0.70-0.97), gender (r = 0.81-0.96), and multiple scanners from the five sites (r = 0.74-0.99). The AIc area under the receiver operating characteristic curve for centered and off-centered CT exams ranged from 0.72 (head) to 0.99 (chest). Therefore, our study showed that positron-emission tomography/CT (PET/CT) examinations commonly exhibit significant off-centering, particularly with vertical deviations often exceeding 30 mm and horizontal deviations between 10 and 30 mm. In addition, it demonstrated that our AI model can effectively assess both vertical and horizontal off-centering, although it performs better at estimating vertical off-centering.

To comprehensively evaluate the effectiveness of FDG-PET/MRI multiparametric analysis in predicting myocardial wall motion recovery following revascularization of chronic coronary total occlusions (CTO), incorporating both traditional and machine learning approaches.

This retrospective study assessed fluorine-18 fluorodeoxyglucose uptake (FDG), late gadolinium enhanced MR imaging (LGE), and MR wall motion abnormalities (WMA) of the left ventricle walls of a clinical cohort with 21 CTO patients (62 ± 9 years, 20 men). All patients were examined using a PET/MRI prior to revascularization and a follow-up cardiac MRI six months later. Prediction models for wall motion recovery after perfusion restoration were developed using linear and nonlinear algorithms as well as multiparametric variables. Performance and prediction explainability were evaluated in a 5x2 cross-validation framework, using ROC AUC and McNemar's test modified for clustered matched-pair data, and Shapley values.

Based on 79 CTO-subtended myocardial wall segments with wall motion abnormalities at baseline, the reference logistic regression model LGE + FDG obtained 0.55(SE = 0.07) in the clustered ROC AUC (cROC AUC) and 0.17(0.05) in the Global Absolute Shapley value. The reference outperformed FDG standalone in cROC AUC (-35(17) %, p < 0.0001), but not LGE standalone (11(12) %, p > 0.05). There were no statistically significant differences between the marginal probabilities of success of these three models. Moreover, no significant improvements (differences < 10 % in cROC AUC, and < 20 % in Global Absolute Shapley, p > 0.05) were found when using mixed effects logistic regression, decision tree, k-nearest neighbor, Naive Bayes, random forest, and support vector machine, with multiparametric combinations of FDG, LGE, and/or WMA.

In this clinical cohort, adding more complex interactions between PET/MRI imaging of cardiac function, infarct extension, and/or metabolism did not enhance the prediction of wall motion recovery after perfusion restoration. This finding raises the question whether multiparametric FDG-PET/MRI analysis has demonstrable benefits in risk stratification for CTO revascularization. Further studies with larger cohorts and external validation datasets are crucial to clarify this question and refine the role of multiparametric imaging in this context.

Pediatric respiratory diseases, including asthma and pneumonia, are major causes of morbidity and mortality in children. Auscultation of lung sounds is a key diagnostic tool but is prone to subjective variability. The integration of artificial intelligence (AI) and machine learning (ML) with electronic stethoscopes offers a promising approach for automated and objective lung sound.

This systematic review and meta-analysis assess the performance of ML models in pediatric lung sound analysis. The study evaluates the methodologies, model performance, and database characteristics while identifying limitations and future directions for clinical implementation.

A systematic search was conducted in Medline via PubMed, Embase, Web of Science, OVID, and IEEE Xplore for studies published between January 1, 1990, and December 16, 2024. Inclusion criteria are as follows: studies developing ML models for pediatric lung sound classification with a defined database, physician-labeled reference standard, and reported performance metrics. Exclusion criteria are as follows: studies focusing on adults, cardiac auscultation, validation of existing models, or lacking performance metrics. Risk of bias was assessed using a modified Quality Assessment of Diagnostic Accuracy Studies (version 2) framework. Data were extracted on study design, dataset, ML methods, feature extraction, and classification tasks. Bivariate meta-analysis was performed for binary classification tasks, including wheezing and abnormal lung sound detection.

A total of 41 studies met the inclusion criteria. The most common classification task was binary detection of abnormal lung sounds, particularly wheezing. Pooled sensitivity and specificity for wheeze detection were 0.902 (95% CI 0.726-0.970) and 0.955 (95% CI 0.762-0.993), respectively. For abnormal lung sound detection, pooled sensitivity was 0.907 (95% CI 0.816-0.956) and specificity 0.877 (95% CI 0.813-0.921). The most frequently used feature extraction methods were Mel-spectrogram, Mel-frequency cepstral coefficients, and short-time Fourier transform. Convolutional neural networks were the predominant ML model, often combined with recurrent neural networks or residual network architectures. However, high heterogeneity in dataset size, annotation methods, and evaluation criteria were observed. Most studies relied on small, single-center datasets, limiting generalizability.

ML models show high accuracy in pediatric lung sound analysis, but face limitations due to dataset heterogeneity, lack of standard guidelines, and limited external validation. Future research should focus on standardized protocols and the development of large-scale, multicenter datasets to improve model robustness and clinical implementation.

The quality of clinical PET/CT images is critical for both accurate diagnosis and image-based research. However, current image quality assessment (IQA) methods predominantly rely on handcrafted features and region-specific analyses, thereby limiting automation in whole-body and multicenter evaluations. This study aims to develop an expert-perceptive deep learning-based IQA system for [18F]FDG PET/CT to tackle the lack of automated, interpretable assessments of clinical whole-body PET/CT image quality.

This retrospective multicenter study included clinical whole-body [18F]FDG PET/CT scans from 718 patients. Automated identification and localization algorithms were applied to select predefined pairs of PET and CT slices from whole-body images. Fifteen experienced experts, trained to conduct blinded slice-level subjective assessments, provided average visual scores as reference standards. Using the MANIQA framework, the developed IQA model integrates the Vision Transformer, Transposed Attention, and Scale Swin Transformer Blocks to categorize PET and CT images into five quality classes. The model's correlation, consistency, and accuracy with expert evaluations on both PET and CT test sets were statistically analysed to assess the system's IQA performance. Additionally, the model's ability to distinguish high-quality images was evaluated using receiver operating characteristic (ROC) curves.

The IQA model demonstrated high accuracy in predicting image quality categories and showed strong concordance with expert evaluations of PET/CT image quality. In predicting slice-level image quality across all body regions, the model achieved an average accuracy of 0.832 for PET and 0.902 for CT. The model's scores showed substantial agreement with expert assessments, achieving average Spearman coefficients (ρ) of 0.891 for PET and 0.624 for CT, while the average Intraclass Correlation Coefficient (ICC) reached 0.953 for PET and 0.92 for CT. The PET IQA model demonstrated strong discriminative performance, achieving an area under the curve (AUC) of ≥ 0.88 for both the thoracic and abdominal regions.

This fully automated IQA system provides a robust and comprehensive framework for the objective evaluation of clinical image quality. Furthermore, it demonstrates significant potential as an impartial, expert-level tool for standardised multicenter clinical IQA.

Deep learning-based synthetic super-resolution magnetic resonance imaging (SynthMRI) may improve the quantitative lesion performance of portable low-field strength magnetic resonance imaging (LF-MRI). The aim of this study is to evaluate whether SynthMRI improves the diagnostic performance of LF-MRI in assessing ischemic lesions.

We retrospectively included 178 stroke patients and 104 healthy controls with both LF-MRI and high-field strength magnetic resonance imaging (HF-MRI) examinations. Using HF-MRI as the ground truth, the deep learning-based super-resolution framework (SCUNet) was pretrained using large-scale open-source data sets to generate SynthMRI images from LF-MRI images. Participants were split into a training set (64.2%) to fine-tune the pretrained SCUNet, and a testing set (35.8%) to evaluate the performance of SynthMRI. Sensitivity and specificity of LF-MRI and SynthMRI were assessed. Agreement with HF-MRI for Alberta Stroke Program Early Computed Tomography Score in the anterior and posterior circulation (diffusion-weighted imaging-Alberta Stroke Program Early Computed Tomography Score and diffusion-weighted imaging-posterior circulation Alberta Stroke Program Early Computed Tomography Score) was evaluated using intraclass correlation coefficients (ICCs). Agreement with HF-MRI for lesion volume and mean apparent diffusion coefficient (ADC) within lesions was assessed using both ICCs and Pearson correlation coefficients.

SynthMRI demonstrated significantly higher sensitivity and specificity than LF-MRI (89.0% [83.3%-94.6%] versus 77.1% [69.5%-84.7%]; P<0.001 and 91.3% [84.7%-98.0%] versus 71.0% [60.3%-81.7%]; P<0.001, respectively). The ICCs of diffusion-weighted imaging-Alberta Stroke Program Early Computed Tomography Score between SynthMRI and HF-MRI were also better than that between LF-MRI and HF-MRI (0.952 [0.920-0.972] versus 0.797 [0.678-0.876], P<0.001). For lesion volume and mean apparent diffusion coefficient within lesions, SynthMRI showed significantly higher agreement (P<0.001) with HF-MRI (ICC>0.85, r>0.78) than LF-MRI (ICC>0.45, r>0.35). Furthermore, for lesions during various poststroke phases, SynthMRI exhibited significantly higher agreement with HF-MRI than LF-MRI during the early hyperacute and subacute phases.

SynthMRI demonstrates high agreement with HF-MRI in detecting and quantifying ischemic lesions and is better than LF-MRI, particularly for lesions during the early hyperacute and subacute phases.

Prior work has demonstrated that AI access can help residents more accurately detect pediatric fractures. We wished to evaluate the effectiveness of an unsupervised AI-based training module as a pediatric fracture detection educational tool.

Two hundred forty radiographic examinations from throughout the pediatric upper extremity were split into two groups of 120 examinations. A previously developed open-source deep learning fracture detection algorithm ( www.childfx.com ) was used to annotate radiographs. Four medical students and four PGY-2 radiology residents first evaluated 120 examinations for fracture without AI assistance and subsequently reviewed AI annotations on these cases via a training module. They then interpreted 120 different examinations without AI assistance. Pre- and post-intervention fracture detection accuracy was evaluated using a chi-squared test.

Overall resident fracture detection accuracy significantly improved from 71.3% pre-intervention to 77.5% post-intervention (p = 0.032). Medical student fracture detection accuracy was not significantly changed from 56.3% pre-intervention to 57.3% post-intervention (p = 0.794). Eighty-eight percent of responding participants (7/8) would recommend this model of learning.

We found that a tailored AI-based training module increased resident accuracy for detecting pediatric fractures by 6.2%. Medical student accuracy was not improved, likely due to their limited background familiarity with the task. AI offers a scalable method for automatically generating annotated teaching cases covering varied pathology, allowing residents to efficiently learn from simulated experience.

An extensive library of symptom inventories has been developed over time to measure clinical symptoms of traumatic brain injury (TBI), but this variety has led to several long-standing issues. Most notably, results drawn from different settings and studies are not comparable. This creates a fundamental problem in TBI diagnostics and outcome prediction, namely that it is not possible to equate results drawn from distinct tools and symptom inventories. Here, we present an approach using semantic textual similarity (STS) to link symptoms and scores across previously incongruous symptom inventories by ranking item text similarities according to their conceptual likeness. We tested the ability of four pretrained deep learning models to screen thousands of symptom description pairs for related content-a challenging task typically requiring expert panels. Models were tasked to predict symptom severity across four different inventories for 6,607 participants drawn from 16 international data sources. The STS approach achieved 74.8% accuracy across five tasks, outperforming other models tested. Correlation and factor analysis found the properties of the scales were broadly preserved under conversion. This work suggests that incorporating contextual, semantic information can assist expert decision-making processes, yielding broad gains for the harmonization of TBI assessment.

To evaluate a deep learning (DL) model for detecting keratinized gingiva (KG) in dental photographs and validate its clinical applicability using reference retainers for calibration.

A total of 576 sextant photographs were selected from 32 subjects, each with three sets of photographs: iodine-stained, unstained and line-marked retainers. Relative keratinized gingiva width (rKGW) was measured using visual, functional and histochemical staining methods with reference retainers. A pre-trained DeepLabv3 model with ResNet50 backbone was fine-tuned to predict KG areas, which were then applied to the photographs with line-marked retainers for subsequent rKGW measurement.

The AI model achieved a Dice coefficient of 93.30% and an accuracy of 93.32%. Using histochemical measurements as gold standards, the absolute differences in rKGW of AI measurements were statistically insignificant with visual (p = 0.935) and functional (p = 0.979) measurements. The adjusted difference between AI and histochemical measurements was 0.377 mm. AI closely matched histochemical measurements in the maxillary anterior region (0.011 mm, p = 0.903) but was significantly higher in the maxillary posterior region (0.327 mm, p < 0.05).

The proposed AI model is the first to reliably identify full-mouth KG, validated thoroughly using reference retainers. However, predictions for posterior teeth warrant further improvement.

The objectives of this study are to synthesize findings from recent research of retrieval-augmented generation (RAG) and large language models (LLMs) in biomedicine and provide clinical development guidelines to improve effectiveness.

We conducted a systematic literature review and a meta-analysis. The report was created in adherence to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses 2020 analysis. Searches were performed in 3 databases (PubMed, Embase, PsycINFO) using terms related to "retrieval augmented generation" and "large language model," for articles published in 2023 and 2024. We selected studies that compared baseline LLM performance with RAG performance. We developed a random-effect meta-analysis model, using odds ratio as the effect size.

Among 335 studies, 20 were included in this literature review. The pooled effect size was 1.35, with a 95% confidence interval of 1.19-1.53, indicating a statistically significant effect (P = .001). We reported clinical tasks, baseline LLMs, retrieval sources and strategies, as well as evaluation methods.

Building on our literature review, we developed Guidelines for Unified Implementation and Development of Enhanced LLM Applications with RAG in Clinical Settings to inform clinical applications using RAG.

Overall, RAG implementation showed a 1.35 odds ratio increase in performance compared to baseline LLMs. Future research should focus on (1) system-level enhancement: the combination of RAG and agent, (2) knowledge-level enhancement: deep integration of knowledge into LLM, and (3) integration-level enhancement: integrating RAG systems within electronic health records.

Alzheimer's disease (AD) is a complex neurodegenerative disorder characterized by progressive cognitive decline and is traditionally associated with grey matter pathology. Recent research highlights the significance of white matter and myelin damage in AD, presenting a paradigm shift in understanding the disease. The aim of this study was to summarize current advancements in magnetic resonance imaging (MRI) techniques and their applications in assessing myelin and brain pathology in AD with a special focus on ultrashort echo time (UTE) based techniques, alongside the role of artificial intelligence (AI) in enhancing diagnostic accuracy.

Between April and May 2024, we conducted a literature search using Google Scholar, Web of Science, and PubMed, focusing on publications from 1990 to 2024. Search terms included "Quantitative imaging", "Alzheimer's MRI", "T1ρ Alzheimer's", "MT imaging Alzheimer's", and "myelin water fraction Alzheimer's". We included quantitative MRI studies involving AD brains and excluded volumetric analyses, non-quantitative studies, non-English reports, non-peer-reviewed studies, and animal research.

Quantitative MRI techniques, including T1, T1ρ, magnetization transfer ratio (MTR), T2, T2*, susceptibility, myelin water fraction (MWF), and non-aqueous myelin proton density (PD) were described. These biomarkers represent different pathophysiological elements of brain damage and may have distinct functions at different phases of the disease. The role of AI in enhancing diagnostic accuracy is also discussed.

In conclusion, integrating advanced MRI techniques and AI offers promising avenues for understanding and diagnosing AD. The focus on myelin damage and white matter integrity underscores the importance of comprehensive imaging approaches. Continued research and development are essential to address current challenges and improve clinical practice in AD diagnostics.

Despite a global decrease in the number of COVID-19 patients, early prediction of the clinical course for optimal patient care remains challenging. Recently, the usefulness of image generation for medical images has been investigated. This study aimed to generate short-term follow-up chest CT images using a latent diffusion model in patients with COVID-19.

We retrospectively enrolled 505 patients with COVID-19 for whom the clinical parameters (patient background, clinical symptoms, and blood test results) upon admission were available and chest CT imaging was performed. Subject datasets (n = 505) were allocated for training (n = 403), and the remaining (n = 102) were reserved for evaluation. The image underwent variational autoencoder (VAE) encoding, resulting in latent vectors. The information consisting of initial clinical parameters and radiomic features were formatted as a table data encoder. Initial and follow-up latent vectors and the initial table data encoders were utilized for training the diffusion model. The evaluation data were used to generate prognostic images. Then, similarity of the prognostic images (generated images) and the follow-up images (real images) was evaluated by zero-mean normalized cross-correlation (ZNCC), peak signal-to-noise ratio (PSNR), and structural similarity (SSIM). Visual assessment was also performed using a numerical rating scale.

Prognostic chest CT images were generated using the diffusion model. Image similarity showed reasonable values of 0.973 ± 0.028 for the ZNCC, 24.48 ± 3.46 for the PSNR, and 0.844 ± 0.075 for the SSIM. Visual evaluation of the images by two pulmonologists and one radiologist yielded a reasonable mean score.

The similarity and validity of generated predictive images for the course of COVID-19-associated pneumonia using a diffusion model were reasonable. The generation of prognostic images may suggest potential utility for early prediction of the clinical course in COVID-19-associated pneumonia and other respiratory diseases.

Artificial intelligence (AI) has the potential to transform cancer diagnosis, ultimately leading to better patient outcomes.

We performed an umbrella review to summarize and critically evaluate the evidence for the AI-based imaging diagnosis of cancers.

PubMed, Embase, Web of Science, Cochrane, and IEEE databases were searched for relevant systematic reviews from inception to June 19, 2024. Two independent investigators abstracted data and assessed the quality of evidence, using the Joanna Briggs Institute (JBI) Critical Appraisal Checklist for Systematic Reviews and Research Syntheses. We further assessed the quality of evidence in each meta-analysis by applying the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) criteria. Diagnostic performance data were synthesized narratively.

In a comprehensive analysis of 158 included studies evaluating the performance of AI algorithms in noninvasive imaging diagnosis across 8 major human system cancers, the accuracy of the classifiers for central nervous system cancers varied widely (ranging from 48% to 100%). Similarities were observed in the diagnostic performance for cancers of the head and neck, respiratory system, digestive system, urinary system, female-related systems, skin, and other sites. Most meta-analyses demonstrated positive summary performance. For instance, 9 reviews meta-analyzed sensitivity and specificity for esophageal cancer, showing ranges of 90%-95% and 80%-93.8%, respectively. In the case of breast cancer detection, 8 reviews calculated the pooled sensitivity and specificity within the ranges of 75.4%-92% and 83%-90.6%, respectively. Four meta-analyses reported the ranges of sensitivity and specificity in ovarian cancer, and both were 75%-94%. Notably, in lung cancer, the pooled specificity was relatively low, primarily distributed between 65% and 80%. Furthermore, 80.4% (127/158) of the included studies were of high quality according to the JBI Critical Appraisal Checklist, with the remaining studies classified as medium quality. The GRADE assessment indicated that the overall quality of the evidence was moderate to low.

Although AI shows great potential for achieving accelerated, accurate, and more objective diagnoses of multiple cancers, there are still hurdles to overcome before its implementation in clinical settings. The present findings highlight that a concerted effort from the research community, clinicians, and policymakers is required to overcome existing hurdles and translate this potential into improved patient outcomes and health care delivery.

PROSPERO CRD42022364278; https://www.crd.york.ac.uk/PROSPERO/view/CRD42022364278.

Hip dysplasia (HD) is a prevalent cause of non-traumatic hip pain, which may result in osteoarthritis. Radiological measurements of HD exhibit variability based on reader and imaging modality, why it is important to know the agreement between different measurement methods.

To estimate agreement between measurements of lateral center edge angle (LCEA) and acetabular inclination angle (AIA) made, respectively, on Computed Tomography (CT) scans by humans and radiographs analyzed by an algorithm. To estimate impact of pelvic rotation on agreement between CT and radiographic measurements.

CT measurements were retrospectively extracted from 172 radiology reports. Radiographs were analyzed using an algorithm. Bland-Altman analysis assessed agreement between CT and radiographic measurements. Regression analyses estimated impact of pelvic rotation on inter-modality agreement.

Mean measured bias (95% confidence interval [CI]) between CT and radiographs for LCEA of right/left hip was 5.53° (95% CI: 4.81 to 6.24) and 5.13 (95% CI: 4.43 to 5.83), respectively. Corresponding values for right/left AIA were 1.08 (95% CI: 0.49 to 1.67) and -0.03 (95% CI: -0.60 to 0.05). Pelvic rotation affected right LCEA and AIA measurements, with a change in obturator foramen index of, respectively, 0.35 and 0.6 resulting in approximately 2° change in values.

There was a significant difference in agreement of 5° between CT and radiographs for the LCEA bilaterally. The difference for the AIA was between 0 and 1°, probably of little clinical significance. Pelvic rotation slightly affected bias of the right LCEA, suggesting minimal clinical impact of a slightly rotated pelvis.

Soft-tissue and bone tumours (STBT) are rare, diagnostically challenging lesions with variable clinical behaviours and treatment approaches. This systematic review aims to provide an overview of Artificial Intelligence (AI) methods using radiological imaging for diagnosis and prognosis of these tumours, highlighting challenges in clinical translation, and evaluating study alignment with the Checklist for AI in Medical Imaging (CLAIM) and the FUTURE-AI international consensus guidelines for trustworthy and deployable AI to promote the clinical translation of AI methods.

The systematic review identified literature from several bibliographic databases, covering papers published before 17/07/2024. Original research published in peer-reviewed journals, focused on radiology-based AI for diagnosis or prognosis of primary STBT was included. Exclusion criteria were animal, cadaveric, or laboratory studies, and non-English papers. Abstracts were screened by two of three independent reviewers to determine eligibility. Included papers were assessed against the two guidelines by one of three independent reviewers. The review protocol was registered with PROSPERO (CRD42023467970).

The search identified 15,015 abstracts, from which 325 articles were included for evaluation. Most studies performed moderately on CLAIM, averaging a score of 28.9 ± 7.5 out of 53, but poorly on FUTURE-AI, averaging 5.1 ± 2.1 out of 30.

Imaging-AI tools for STBT remain at the proof-of-concept stage, indicating significant room for improvement. Future efforts by AI developers should focus on design (e.g. defining unmet clinical need, intended clinical setting and how AI would be integrated in clinical workflow), development (e.g. building on previous work, training with data that reflect real-world usage, explainability), evaluation (e.g. ensuring biases are evaluated and addressed, evaluating AI against current best practices), and the awareness of data reproducibility and availability (making documented code and data publicly available). Following these recommendations could improve clinical translation of AI methods.

Hanarth Fonds, ICAI Lab, NIHR, EuCanImage.

The aim of this paper is to contextualize and review reporting guidelines available at the EQUATOR Network that are most relevant to radiology-specific investigations. Eight EQUATOR Network reporting guidelines for the clinical area of radiology, not including the subspecialized areas of imaging of the cardiovascular, neurologic, and oncologic diseases are reviewed and discussed. The reporting guidelines are for diagnostic and therapeutic clinical research. Why the reporting guideline was development, by whom, their aims and what they hope to achieve are discussed. A table summarizes what the reporting guideline is provided for; an acronym if present is given; a full bibliographic reference with PMID number; the reporting guideline website URL or link; the study design and section of the report that the guideline applies to; and the date that the reporting guideline was last updated.

To perform a systematic review and meta-analysis that assesses the diagnostic performance of deep learning algorithms applied to breast MRI for predicting axillary lymph nodes metastases in patients of breast cancer.

A systematic literature search in PubMed, MEDLINE, and Embase databases for articles published from January 2004 to February 2025. Inclusion criteria were: patients with breast cancer; deep learning using MRI images was applied to predict axillary lymph nodes metastases; sufficient data were present; original research articles. Quality Assessment of Diagnostic Accuracy Studies-AI and Checklist for Artificial Intelligence in Medical Imaging was used to assess the quality. Statistical analysis included pooling of diagnostic accuracy and investigating between-study heterogeneity. A summary receiver operating characteristic curve (SROC) was performed. R statistical software (version 4.4.0) was used for statistical analyses.

A total of 10 studies were included. The pooled sensitivity and specificity were 0.76 (95% CI, 0.67-0.83) and 0.81 (95% CI, 0.74-0.87), respectively, with both measures having moderate between-study heterogeneity (I2 = 61% and 60%, respectively; p < 0.01). The SROC analysis yielded a weighted AUC of 0.788.

This meta-analysis demonstrates that deep learning algorithms applied to breast MRI offer promising diagnostic performance for predicting axillary lymph node metastases in breast cancer patients. Incorporating deep learning into clinical practice may enhance decision-making by providing a non-invasive method to more accurately predict lymph node involvement, potentially reducing unnecessary surgeries.

Distinguishing between benign and malignant testicular lesions on clinical magnetic resonance imaging (MRI) is crucial for guiding treatment planning. However, conventional MRI-based radiomics to identify testicular cancer requires expert machine learning knowledge. This study aims to investigate the potential of utilizing automatic machine learning (AutoML) based on MRI to diagnose testicular lesions without the need for expert algorithm optimization.

Retrospective preoperative MRI scans from 115 patients diagnosed with testicular disease through pathology were obtained. A total of 1781 radiomics features were extracted from each lesion on the T2-weighted images. Intraclass and interclass correlation coefficients were used to evaluate the intra-observer and interobserver agreements for each radiomics feature. We developed an AutoML method based on the tree-based pipeline optimization tool (TPOT) algorithm to construct a discriminant model. The best pipeline was determined through 100 repeated operations using a 5-fold cross-validation algorithm in TPOT. The model was evaluated for accuracy, sensitivity, and specificity using the area under the curve (AUC) value of the receiver operating characteristic (ROC) curve. Shapley Additive exPlanations were used to illustrate the optimization results.

Utilizing the TPOT method, 100 diagnostic models were developed to identify testicular lesions. The best model was determined based on the highest AUC in the training cohort. The prediction model yielded AUC values of 0.989 (95% confidence interval [CI]: 0.985-0.993) and 0.909 (95% CI: 0.893-0.923) in the training and testing cohorts, respectively.

AutoML, based on the TPOT algorithm, holds potential as a noninvasive method for effectively discriminating between benign and malignant testicular lesions.

Background: Approximately 50% of all oncological patients undergo radiation therapy, where personalized planning of treatment relies on gross tumor volume (GTV) delineation. Manual delineation of GTV is time-consuming, operator-dependent, and prone to variability. An increasing number of studies apply artificial intelligence (AI) techniques to automate such delineation processes. Methods: To perform a systematic review comparing the performance of AI models in tumor delineations within the body (thoracic cavity, esophagus, abdomen, and pelvis, or soft tissue and bone). A retrospective search of five electronic databases was performed between January 2017 and February 2025. Original research studies developing and/or validating algorithms delineating GTV in CT, MRI, and/or PET were included. The Checklist for Artificial Intelligence in Medical Imaging (CLAIM) and Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis statement and checklist (TRIPOD) were used to assess the risk, bias, and reporting adherence. Results: After screening 2430 articles, 48 were included. The pooled diagnostic performance from the use of AI algorithms across different tumors and topological areas ranged 0.62-0.92 in dice similarity coefficient (DSC) and 1.33-47.10 mm in Hausdorff distance (HD). The algorithms with the highest DSC deployed an encoder-decoder architecture. Conclusions: AI algorithms demonstrate a high level of concordance with clinicians in GTV delineation. Translation to clinical settings requires the building of trust, improvement in performance and robustness of results, and testing in prospective studies and randomized controlled trials.

Acute ischemic stroke (AIS) is a major cause of mortality and morbidity, with hemorrhagic transformation (HT) as a severe complication. Accurate prediction of HT is essential for optimizing treatment strategies. This review assesses the accuracy and utility of deep learning (DL) and radiomics in predicting HT through imaging, regarding clinical decision-making for AIS patients. A literature search was conducted across five databases (Pubmed, Scopus, Web of Science, Embase, IEEE) up to January 23, 2025. Studies involving DL or radiomics-based ML models for predicting HT in AIS patients were included. Data from training, validation, and clinical-combined models were extracted and analyzed separately. Pooled sensitivity, specificity, and AUC were calculated with a random-effects bivariate model. For the quality assessment of studies, the Methodological Radiomics Score (METRICS) and QUADAS-2 tool were used. 16 studies consisting of 3,083 individual participants were included in the meta-analysis. The pooled AUC for training cohorts was 0.87, sensitivity 0.80, and specificity 0.85. For validation cohorts, AUC was 0.87, sensitivity 0.81, and specificity 0.86. Clinical-combined models showed an AUC of 0.93, sensitivity 0.84, and specificity 0.89. Moderate to severe heterogeneity was noted and addressed. Deep-learning models outperformed radiomics models, while clinical-combined models outperformed deep learning-only and radiomics-only models. The average METRICS score was 62.85%. No publication bias was detected. DL and radiomics models showed great potential in predicting HT in AIS patients. However, addressing methodological issues-such as inconsistent reference standards and limited external validation-is essential for the clinical implementation of these models.