Bone age estimation (BAE) is based on skeletal maturity and degenerative process of the skeleton. The clinical importance of BAE is in understanding the pediatric and growth-related disorders; whereas medicolegally it is important in determining criminal responsibility and establishing identification. Artificial Intelligence (AI) has been used in the field of the field of medicine and specifically in diagnostics using medical images. AI can greatly benefit the BAE techniques by decreasing the intra observer and inter observer variability as well as by reducing the analytical time. The AI techniques rely on object identification, feature extraction and segregation. Bone age assessment is the classical example where the concepts of AI such as object recognition and segregation can be used effectively. The paper describes various AI based algorithms developed for the purpose of radiologic BAE and the performances of the models. In the current paper we have also carried out qualitative analysis using Strength, Weakness, Opportunities and Challenges (SWOC) to examine critical factors that contribute to the application of AI in BAE. To best of our knowledge, the SWOC analysis is being carried out for the first time to assess the applicability of AI in BAE. Based on the SWOC analysis we have provided strategies for successful implementation of AI in BAE in forensic and medicolegal context.

Age at death is one of the key elements of the "biological profile" prepared when analysing unidentified human remains. Biological age is determined according to physiological indicators and developmental stage, which can be determined by bone assessment. It is worth remembering that the researcher must interpret each case individually and in accordance with the current state of knowledge. One of the most developed tools for analysing human remains is postmortem computed tomography. This allows for the visualization not only of bones without maceration but also of the entire body under various altered states, including corpses in advanced stages of decomposition and burnt bodies. The aim of this review is to present the current methods for age estimation based on postmortem computed tomography evaluation, comparing the results presented in 18 research projects published between 2013 and 2023 on foetuses, children, and adults from contemporary populations. Recent literature includes assessment of bones and characteristics such as skulls, teeth, vertebrae, pelvises, and long bones to estimate age at death. We cover the methods used in this recent literature, including machine learning, and discuss the advantages and disadvantages of them.

Postmortem computed tomography allows the analysis of several areas of the body at the same time, which may not be possible in the case of clinical trials (where the examination area should be limited).Postmortem computed tomography may enable the collection of data from people whose clinical examinations are relatively rare (e.g., pregnant women, children).Artificial intelligence should increasingly be used in studies on age estimation.Further research on modern populations is necessary to verify and refine the methods used to estimate age at death.

To date, AI-supported programs for bone age (BA) determination for medical use in Europe have almost only been validated separately, according to Greulich and Pyle (G&P). Therefore, the current study aimed to compare the performance of three programs, namely BoneXpert, PANDA, and BoneView, on a single Central European population.

For this retrospective study, hand radiographs of 306 children aged 1-18 years, stratified by gender and age, were included. A subgroup consisting of the age group accounting for 90% of examinations in clinical practice was formed. The G&P BA was estimated by three human experts-as ground truth-and three AI-supported programs. The mean absolute deviation, the root mean squared error (RMSE), and dropouts by the AI were calculated.

The correlation between all programs and the ground truth was prominent (R2 ≥ 0.98). In the total group, BoneXpert had a lower RMSE than BoneView and PANDA (0.62 vs. 0.65 and 0.75 years) with a dropout rate of 2.3%, 20.3% and 0%, respectively. In the subgroup, there was less difference in RMSE (0.66 vs. 0.68 and 0.65 years, max. 4% dropouts). The standard deviation between the AI readers was lower than that between the human readers (0.54 vs. 0.62 years, p < 0.01).

All three AI programs predict BA after G&P in the main age range with similar high reliability. Differences arise at the boundaries of childhood.

Question There is a lack of comparative, independent validation for artificial intelligence-based bone age estimation in children. Findings Three commercially available programs estimate bone age after Greulich and Pyle with similarly high reliability in a central European cohort. Clinical relevance The comparative study will help the reader choose a software for bone age estimation approved for the European market depending on the targeted age group and economic considerations.

The potential intricacy of skull fractures as well as the complexity of underlying anatomy poses diagnostic hurdles for radiologists evaluating computed tomography (CT) scans. The necessity for automated diagnostic tools has been brought to light by the shortage of radiologists and the growing demand for rapid and accurate fracture diagnosis. Convolutional Neural Networks (CNNs) are a potential new class of medical imaging technologies that use deep learning (DL) to improve diagnosis accuracy. The objective of this systematic review and meta-analysis is to assess how well CNN models diagnose skull fractures on CT images.

PubMed, Scopus, and Web of Science were searched for studies published before February 2024 that used CNN models to detect skull fractures on CT scans. Meta-analyses were conducted for area under the receiver operating characteristic curve (AUC), sensitivity, specificity, and accuracy. Egger's and Begg's tests were used to assess publication bias.

Meta-analysis was performed for 11 studies with 20,798 patients. Pooled average AUC for implementing pre-training for transfer learning in CNN models within their training model's architecture was 0.96 ± 0.02. The pooled averages of the studies' sensitivity and specificity were 1.0 and 0.93, respectively. The accuracy was obtained 0.92 ± 0.04. Studies showed heterogeneity, which was explained by differences in model topologies, training models, and validation techniques. There was no significant publication bias detected.

CNN models perform well in identifying skull fractures on CT scans. Although there is considerable heterogeneity and possibly publication bias, the results suggest that CNNs have the potential to improve diagnostic accuracy in the imaging of acute skull trauma. To further enhance these models' practical applicability, future studies could concentrate on the utility of DL models in prospective clinical trials.

The purpose of this study was to evaluate the effectiveness of a deep learning model (DLM) in improving the sensitivity of neurosurgery residents to detect intracranial aneurysms on CT angiography (CTA) in patients with aneurysmal subarachnoid hemorrhage (aSAH).

In this diagnostic accuracy study, a set of 104 CTA scans of aSAH patients containing a total of 126 aneurysms were presented to three blinded neurosurgery residents (a first-year, third-year, and fifth-year resident), who individually assessed them for aneurysms. After the initial reading, the residents were given the predictions of a dedicated DLM previously established for automated detection and segmentation of intracranial aneurysms. The detection sensitivities for aneurysms of the DLM and the residents with and without the assistance of the DLM were compared.

The DLM had a detection sensitivity of 85.7%, while the residents showed detection sensitivities of 77.8%, 86.5%, and 87.3% without DLM assistance. After being provided with the DLM's results, the residents' individual detection sensitivities increased to 97.6%, 95.2%, and 98.4%, respectively, yielding an average increase of 13.2%. The DLM was particularly useful in detecting small aneurysms. In addition, interrater agreement among residents increased from a Fleiss κ of 0.394 without DLM assistance to 0.703 with DLM assistance.

The results of this pilot study suggest that deep learning models can help neurosurgeons detect aneurysms on CTA and make appropriate treatment decisions when immediate radiological consultation is not possible.

Bone age (BA) reflects skeletal maturity and is crucial in clinical and forensic contexts, particularly for growth assessment, adult height prediction, and managing conditions like short stature and precocious puberty, often using X-ray, MRI, CT, or ultrasound imaging. Traditional BA assessment methods, including the Greulich-Pyle and Tanner-Whitehouse techniques, compare morphological changes to reference atlases. Despite their effectiveness, factors like genetics and environment complicate evaluations, emphasizing the need for new methods that account for comprehensive variations in skeletal maturity. The limitations of classical BA assessment methods increase the demand for automated solutions. The first automated tool, HANDX, was introduced in 1989. Researchers now focus on developing reliable artificial intelligence (AI)-driven tools, utilizing machine learning and deep learning techniques to improve accuracy and efficiency in BA evaluations, addressing traditional methods' shortcomings. Recent reviews on BA assessment methods rarely compare AI-based approaches across imaging technologies. This article explores advancements in BA estimation, focusing on machine learning methods and their clinical implications while providing a historical context and highlighting each approach's benefits and limitations.

Detecting and segmenting intracranial aneurysms (IAs) from angiographic images is a laborious task.

To evaluates a novel deep-learning algorithm, named vessel attention (VA)-Unet, for the efficient detection and segmentation of IAs.

This retrospective study was conducted using head CT angiography (CTA) examinations depicting IAs from two hospitals in China between 2010 and 2021. Training included cases with subarachnoid hemorrhage (SAH) and arterial stenosis, common accompanying vascular abnormalities. Testing was performed in cohorts with reference-standard digital subtraction angiography (cohort 1), with SAH (cohort 2), acquired outside the time interval of training data (cohort 3), and an external dataset (cohort 4). The algorithm's performance was evaluated using sensitivity, recall, false positives per case (FPs/case), and Dice coefficient, with manual segmentation as the reference standard.

The study included 3190 CTA scans with 4124 IAs. Sensitivity, recall, and FPs/case for detection of IAs were, respectively, 98.58%, 96.17%, and 2.08 in cohort 1; 95.00%, 88.8%, and 3.62 in cohort 2; 96.00%, 93.77%, and 2.60 in cohort 3; and, 96.17%, 94.05%, and 3.60 in external cohort 4. The segmentation accuracy, as measured by the Dice coefficient, was 0.78, 0.71, 0.71, and 0.66 for cohorts 1-4, respectively. VA-Unet detection recall and FPs/case and segmentation accuracy were affected by several clinical factors, including aneurysm size, bifurcation aneurysms, and the presence of arterial stenosis and SAH.

VA-Unet accurately detected and segmented IAs in head CTA comparably to expert interpretation. The proposed algorithm has significant potential to assist radiologists in efficiently detecting and segmenting IAs from CTA images.

Our study aimed to assess the impact of inter- and intra-observer variations when utilizing an artificial intelligence (AI) system for bone age assessment (BAA) of preschool children.

A retrospective study was conducted involving a total sample of 53 female individuals and 41 male individuals aged 3-6 years in China. Radiographs were assessed by four mid-level radiology reviewers using the TW3 and RUS-CHN methods. Bone age (BA) was analyzed in two separate situations, with/without the assistance of AI. Following a 4-week wash-out period, radiographs were reevaluated in the same manner. Accuracy metrics, the correlation coefficient (ICC)and Bland-Altman plots were employed.

The accuracy of BAA by the reviewers was significantly improved with AI. The results of RMSE and MAE decreased in both methods (p < 0.001). When comparing inter-observer agreement in both methods and intra-observer reproducibility in two interpretations, the ICC results were improved with AI. The ICC values increased in both two interpretations for both methods and exceeded 0.99 with AI.

In the assessment of BA for preschool children, AI was found to be capable of reducing inter-observer variability and enhancing intra-observer reproducibility, which can be considered an important tool for clinical work by radiologists.

The RUS-CHN method is a special bone age method devised to be suitable for Chinese children. The preschool stage is a critical phase for children, marked by a high degree of variability that renders BA prediction challenging. The accuracy of BAA by the reviewers can be significantly improved with the aid of an AI model system. This study is the first to assess the impact of inter- and intra-observer variations when utilizing an AI model system for BAA of preschool children using both the TW3 and RUS-CHN methods.

Background: Assessment of skeletal maturity is a common clinical practice to investigate adolescent growth and endocrine disorders. The distal radius and ulna (DRU) maturity classification is a practical and easy-to-use scheme that was designed for adolescent idiopathic scoliosis clinical management and presents high sensitivity in predicting the growth peak and cessation among adolescents. However, time-consuming and error-prone manual assessment limits DRU in clinical application. Methods: In this study, we propose a multi-task learning framework with an attention mechanism for the joint segmentation and classification of the distal radius and ulna in hand X-ray images. The proposed framework consists of two sub-networks: an encoder-decoder structure with attention gates for segmentation and a slight convolutional network for classification. Results: With a transfer learning strategy, the proposed framework improved DRU segmentation and classification over the single task learning counterparts and previously reported methods, achieving an accuracy of 94.3% and 90.8% for radius and ulna maturity grading. Findings: Our automatic DRU assessment platform covers the whole process of growth acceleration and cessation during puberty. Upon incorporation into advanced scoliosis progression prognostic tools, clinical decision making will be potentially improved in the conservative and operative management of scoliosis patients.

To develop a deep learning-based cascaded HRNet model, in order to automatically measure X-ray imaging parameters of lumbar sagittal curvature and to evaluate its prediction performance.

A total of 3730 lumbar lateral digital radiography (DR) images were collected from picture archiving and communication system (PACS). Among them, 3150 images were randomly selected as the training dataset and validation dataset, and 580 images as the test dataset. The landmarks of the lumbar curve index (LCI), lumbar lordosis angle (LLA), sacral slope (SS), lumbar lordosis index (LLI), and the posterior edge tangent angle of the vertebral body (PTA) were identified and marked. The measured results of landmarks on the test dataset were compared with the mean values of manual measurement as the reference standard. Percentage of correct key-points (PCK), intra-class correlation coefficient (ICC), Pearson correlation coefficient (r), mean absolute error (MAE), mean square error (MSE), root-mean-square error (RMSE), and Bland-Altman plot were used to evaluate the performance of the cascade HRNet model.

The PCK of the cascaded HRNet model was 97.9-100% in the 3 mm distance threshold. The mean differences between the reference standard and the predicted values for LCI, LLA, SS, LLI, and PTA were 0.43 mm, 0.99°, 1.11°, 0.01 mm, and 0.23°, respectively. There were strong correlation and consistency of the five parameters between the cascaded HRNet model and manual measurements (ICC = 0.989-0.999, R = 0.991-0.999, MAE = 0.63-1.65, MSE = 0.61-4.06, RMSE = 0.78-2.01).

The cascaded HRNet model based on deep learning algorithm could accurately identify the sagittal curvature-related landmarks on lateral lumbar DR images and automatically measure the relevant parameters, which is of great significance in clinical application.

Artificial intelligence (AI) is capable of answering complex medical examination questions, offering the potential to revolutionize medical education and healthcare delivery. In this study we aimed to assess ChatGPT, a model that has demonstrated exceptional performance on standardized exams. Specifically, our focus was on evaluating ChatGPT's performance on the complete 2019 Orthopaedic In-Training Examination (OITE), including questions with an image component. Furthermore, we explored difference in performance when questions varied by text only or text with an associated image, including whether the image was described using AI or a trained orthopaedist.

Questions from the 2019 OITE were input into ChatGPT version 4.0 (GPT-4) using 3 response variants. As the capacity to input or interpret images is not publicly available in ChatGPT at the time of this study, questions with an image component were described and added to the OITE question using descriptions generated by Microsoft Azure AI Vision Studio or authors of the study.

ChatGPT performed equally on OITE questions with or without imaging components, with an average correct answer choice of 49% and 48% across all 3 input methods. Performance dropped by 6% when using image descriptions generated by AI. When using single answer multiple-choice input methods, ChatGPT performed nearly double the rate of random guessing, answering 49% of questions correctly. The performance of ChatGPT was worse than all resident classes on the 2019 exam, scoring 4% lower than PGY-1 residents.

ChatGT performed below all resident classes on the 2019 OITE. Performance on text only questions and questions with images was nearly equal if the image was described by a trained orthopaedic specialist but decreased when using an AI generated description. Recognizing the performance abilities of AI software may provide insight into the current and future applications of this technology into medical education.

Radiographic diagnosis of necrotizing enterocolitis (NEC) is challenging. Deep learning models may improve accuracy by recognizing subtle imaging patterns. We hypothesized it would perform with comparable accuracy to that of senior surgical residents.

This cohort study compiled 494 anteroposterior neonatal abdominal radiographs (214 images NEC, 280 other) and randomly divided them into training, validation, and test sets. Transfer learning was utilized to fine-tune a ResNet-50 deep convolutional neural network (DCNN) pre-trained on ImageNet. Gradient-weighted Class Activation Mapping (Grad-CAM) heatmaps visualized image regions of greatest relevance to the pretrained neural network. Senior surgery residents at a single institution examined the test set. Resident and DCNN ability to identify pneumatosis on radiographic images were measured via area under the receiver operating curves (AUROC) and compared using DeLong's method.

The pretrained neural network achieved AUROC of 0.918 (95% CI, 0.837-0.978) with an accuracy of 87.8% with five false negative and one false positive prediction. Heatmaps confirmed appropriate image region emphasis by the pretrained neural network. Senior surgical residents had a median area under the receiver operating curve of 0.896, ranging from 0.778 (95% CI 0.615-0.941) to 0.991 (95% CI 0.971-0.999) with zero to five false negatives and one to eleven false positive predictions. The deep convolutional neural network performed comparably to each surgical resident's performance (p > 0.05 for all comparisons).

A deep convolutional neural network trained to recognize pneumatosis can quickly and accurately assist clinicians in promptly identifying NEC in clinical practice.

III (study type: Study of Diagnostic Test, study of nonconsecutive patients without a universally applied "gold standard").

Cerebral aneurysms are abnormal dilations of blood vessels in the brain that have the potential to rupture, leading to subarachnoid hemorrhage and other serious complications. Early detection and prediction of aneurysm rupture are crucial for effective management and prevention of rupture-related morbidities and mortalities. This review aims to summarize the current knowledge on risk factors and predictive indicators of rupture in cerebral aneurysms. Morphological characteristics such as aneurysm size, shape, and location, as well as hemodynamic factors including blood flow patterns and wall shear stress, have been identified as important factors influencing aneurysm stability and rupture risk. In addition to these traditional factors, emerging evidence suggests that biological and genetic factors, such as inflammation, extracellular matrix remodeling, and genetic polymorphisms, may also play significant roles in aneurysm rupture. Furthermore, advancements in computational fluid dynamics and machine learning algorithms have enabled the development of novel predictive models for rupture risk assessment. However, challenges remain in accurately predicting aneurysm rupture, and further research is needed to validate these predictors and integrate them into clinical practice. By elucidating and identifying the various risk factors and predictive indicators associated with aneurysm rupture, we can enhance personalized risk assessment and optimize treatment strategies for patients with cerebral aneurysms.

This study aims to devise and assess an automated measurement framework for lumbar pedicle screw parameters leveraging preoperative computed tomography (CT) scans and a deep learning algorithm.

A deep learning model was constructed employing a dataset comprising 1410 axial preoperative CT images of lumbar pedicles sourced from 282 patients. The model was trained to predict several screw parameters, including the axial angle and width of pedicles, the length of pedicle screw paths, and the interpedicular distance. The mean values of these parameters, as determined by two radiologists and one spinal surgeon, served as the reference standard.

The deep learning model achieved high agreement with the reference standard for the axial angle of the left pedicle (ICC = 0.92) and right pedicle (ICC = 0.93), as well as for the length of the left pedicle screw path (ICC = 0.82) and right pedicle (ICC = 0.87). Similarly, high agreement was observed for pedicle width (left ICC = 0.97, right ICC = 0.98) and interpedicular distance (ICC = 0.91). Overall, the model's performance paralleled that of manual determination of lumbar pedicle screw parameters.

The developed deep learning-based model demonstrates proficiency in accurately identifying landmarks on preoperative CT scans and autonomously generating parameters relevant to lumbar pedicle screw placement. These findings suggest its potential to offer efficient and precise measurements for clinical applications.

In pediatric medicine, precise estimation of bone age is essential for skeletal maturity evaluation, growth disorder diagnosis, and therapeutic intervention planning. Conventional techniques for determining bone age depend on radiologists' subjective judgments, which may lead to non-negligible differences in the estimated bone age. This study proposes a deep learning-based model utilizing a fully connected convolutional neural network(CNN) to predict bone age from left-hand radiographs.

The data set used in this study, consisting of 473 patients, was retrospectively retrieved from the PACS (Picture Achieving and Communication System) of a single institution. We developed a fully connected CNN consisting of four convolutional blocks, three fully connected layers, and a single neuron as output. The model was trained and validated on 80% of the data using the mean-squared error as a cost function to minimize the difference between the predicted and reference bone age values through the Adam optimization algorithm. Data augmentation was applied to the training and validation sets yielded in doubling the data samples. The performance of the trained model was evaluated on a test data set (20%) using various metrics including, the mean absolute error (MAE), median absolute error (MedAE), root-mean-squared error (RMSE), and mean absolute percentage error (MAPE). The code of the developed model for predicting the bone age in this study is available publicly on GitHub at https://github.com/afiosman/deep-learning-based-bone-age-estimation .

Experimental results demonstrate the sound capabilities of our model in predicting the bone age on the left-hand radiographs as in the majority of the cases, the predicted bone ages and reference bone ages are nearly close to each other with a calculated MAE of 2.3 [1.9, 2.7; 0.95 confidence level] years, MedAE of 2.1 years, RMAE of 3.0 [1.5, 4.5; 0.95 confidence level] years, and MAPE of 0.29 (29%) on the test data set.

These findings highlight the usability of estimating the bone age from left-hand radiographs, helping radiologists to verify their own results considering the margin of error on the model. The performance of our proposed model could be improved with additional refining and validation.

This study aimed to evaluate the performance of artificial intelligence (AI) software in bone age (BA) assessment, according to the Greulich and Pyle (G&P) method in a German pediatric cohort.

Hand radiographs of 306 pediatric patients aged 1-18 years (153 boys, 153 girls, 18 patients per year of life)-including a subgroup of patients in the age group for which the software is declared (243 patients)-were analyzed retrospectively. Two pediatric radiologists and one endocrinologist made independent blinded BA reads. Subsequently, AI software estimated BA from the same images. Both agreements, accuracy, and interchangeability between AI and expert readers were assessed.

The mean difference between the average of three expert readers and AI software was 0.39 months with a mean absolute difference (MAD) of 6.8 months (1.73 months for the mean difference and 6.0 months for MAD in the intended use subgroup). Performance in boys was slightly worse than in girls (MAD 6.3 months vs. 5.6 months). Regression analyses showed constant bias (slope of 1.01 with a 95% CI 0.99-1.02). The estimated equivalence index for interchangeability was - 14.3 (95% CI -27.6 to - 1.1).

In terms of BA assessment, the new AI software was interchangeable with expert readers using the G&P method.

The use of AI software enables every physician to provide expert reader quality in bone age assessment.

• A novel artificial intelligence-based software for bone age estimation has not yet been clinically validated. • Artificial intelligence showed a good agreement and high accuracy with expert radiologists performing bone age assessment. • Artificial intelligence showed to be interchangeable with expert readers.

Artificial intelligence (AI) has emerged as a revolutionary technology with several applications across different dental fields, including pedodontics. This systematic review has the objective to catalog and explore the various uses of artificial intelligence in pediatric dentistry.

A thorough exploration of scientific databases was carried out to identify studies addressing the usage of AI in pediatric dentistry until December 2023 in the Embase, Scopus, PubMed, and Web of Science databases by two researchers, S.L.R. and A.L.G.

From a pool of 1301 articles, only 64 met the predefined criteria and were considered for inclusion in this review. From the data retrieved, it was possible to provide a narrative discussion of the potential implications of AI in the specialized area of pediatric dentistry. The use of AI algorithms and machine learning techniques has shown promising results in several applications of daily dental pediatric practice, including the following: (1) assisting the diagnostic and recognizing processes of early signs of dental pathologies, (2) enhancing orthodontic diagnosis by automating cephalometric tracing and estimating growth and development, (3) assisting and educating children to develop appropriate behavior for dental hygiene.

AI holds significant potential in transforming clinical practice, improving patient outcomes, and elevating the standards of care in pediatric patients. Future directions may involve developing cloud-based platforms for data integration and sharing, leveraging large datasets for improved predictive results, and expanding AI applications for the pediatric population.

Purpose To evaluate the robustness of an award-winning bone age deep learning (DL) model to extensive variations in image appearance. Materials and Methods In December 2021, the DL bone age model that won the 2017 RSNA Pediatric Bone Age Challenge was retrospectively evaluated using the RSNA validation set (1425 pediatric hand radiographs; internal test set in this study) and the Digital Hand Atlas (DHA) (1202 pediatric hand radiographs; external test set). Each test image underwent seven types of transformations (rotations, flips, brightness, contrast, inversion, laterality marker, and resolution) to represent a range of image appearances, many of which simulate real-world variations. Computational "stress tests" were performed by comparing the model's predictions on baseline and transformed images. Mean absolute differences (MADs) of predicted bone ages compared with radiologist-determined ground truth on baseline versus transformed images were compared using Wilcoxon signed rank tests. The proportion of clinically significant errors (CSEs) was compared using McNemar tests. Results There was no evidence of a difference in MAD of the model on the two baseline test sets (RSNA = 6.8 months, DHA = 6.9 months; P = .05), indicating good model generalization to external data. Except for the RSNA dataset images with an appended radiologic laterality marker (P = .86), there were significant differences in MAD for both the DHA and RSNA datasets among other transformation groups (rotations, flips, brightness, contrast, inversion, and resolution). There were significant differences in proportion of CSEs for 57% of the image transformations (19 of 33) performed on the DHA dataset. Conclusion Although an award-winning pediatric bone age DL model generalized well to curated external images, it had inconsistent predictions on images that had undergone simple transformations reflective of several real-world variations in image appearance. Keywords: Pediatrics, Hand, Convolutional Neural Network, Radiography Supplemental material is available for this article. © RSNA, 2024 See also commentary by Faghani and Erickson in this issue.

Bone age assessment (BAA) is a crucial task in clinical, forensic, and athletic fields. Since traditional age estimation methods are suffered from potential radiation damage, this study aimed to develop and evaluate a deep learning radiomics method based on multiparametric knee MRI for noninvasive and automatic BAA. This retrospective study enrolled 598 patients (age range,10.00-29.99 years) who underwent MR examinations of the knee joint (T1/T2*/PD-weighted imaging). Three-dimensional convolutional neural networks (3D CNNs) were trained to extract and fuse multimodal and multiscale MRI radiomic features for age estimation and compared to traditional machine learning models based on hand-crafted features. The age estimation error was greater in individuals aged 25-30 years; thus, this method may not be suitable for individuals over 25 years old. In the test set aged 10-25 years (n = 95), the 3D CNN (a fusion of T1WI, T2*WI, and PDWI) demonstrated the lowest mean absolute error of 1.32 ± 1.01 years, which is higher than that of other MRI modalities and the hand-crafted models. In the classification for 12-, 14-, 16-, and 18- year thresholds, accuracies and the areas under the ROC curves were all over 0.91 and 0.96, which is similar to the manual methods. Visualization of important features showed that 3D CNN estimated age by focusing on the epiphyseal plates. The deep learning radiomics method enables non-invasive and automated BAA from multimodal knee MR images. The use of 3D CNN and MRI-based radiomics has the potential to assist radiologists or medicolegists in age estimation.

Magnetic resonance imaging (MRI) is an indispensable medical imaging examination technique in musculoskeletal medicine. Modern MRI techniques achieve superior high-quality multiplanar imaging of soft tissue and skeletal pathologies without the harmful effects of ionizing radiation. Some current limitations of MRI include long acquisition times, artifacts, and noise. In addition, it is often challenging to distinguish abutting or closely applied soft tissue structures with similar signal characteristics. In the past decade, Artificial Intelligence (AI) has been widely employed in musculoskeletal MRI to help reduce the image acquisition time and improve image quality. Apart from being able to reduce medical costs, AI can assist clinicians in diagnosing diseases more accurately. This will effectively help formulate appropriate treatment plans and ultimately improve patient care. This review article intends to summarize AI's current research and application in musculoskeletal MRI, particularly the advancement of DL in identifying the structure and lesions of upper extremity joints in MRI images.

Bone age assessment, as an indicator of biological age, is widely used in orthodontics and pediatric endocrinology. Owing to significant subject variations in the manual method of assessment, artificial intelligence (AI), machine learning (ML), and deep learning (DL) play a significant role in this aspect. A scoping review was conducted to search the existing literature on the role of AI, ML, and DL in skeletal age or bone age assessment in healthy individuals.

A literature search was conducted in PubMed, Scopus, and Web of Science from January 2012 to December 2022 using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses-Extension for Scoping Reviews (PRISMA-ScR) and Joanna Briggs Institute guidelines. Grey literature was searched using Google Scholar and OpenGrey. Hand-searching of the articles in all the reputed orthodontic journals and the references of the included articles were also searched for relevant articles for the present scoping review.

Nineteen articles that fulfilled the inclusion criteria were included. Ten studies used skeletal maturity indicators based on hand and wrist radiographs, two studies used magnetic resonance imaging and seven studies used cervical vertebrae maturity indicators based on lateral cephalograms to assess the skeletal age of the individuals. Most of these studies were published in non-orthodontic medical journals. BoneXpert automated software was the most commonly used software, followed by DL models, and ML models in the studies for assessment of bone age. The automated method was found to be as reliable as the manual method for assessment.

This scoping review validated the use of AI, ML, or DL in bone age assessment of individuals. A more uniform distribution of sufficient samples in different stages of maturation, use of three-dimensional inputs such as magnetic resonance imaging, and cone beam computed tomography is required for better training of the models to generalize the outputs for use in the target population.

Bone age (BA) is needed to assess developmental status and growth disorders. We evaluated the clinical performance of a deep-learning-based BA software to estimate the chronological age (CA) of healthy Korean children.

This retrospective study included 371 healthy children (217 boys, 154 girls), aged between 4 and 17 years, who visited the Department of Pediatrics for health check-ups between January 2017 and December 2018. A total of 553 left-hand radiographs from 371 healthy Korean children were evaluated using a commercial deep-learning-based BA software (BoneAge, Vuno, Seoul, Korea). The clinical performance of the deep learning (DL) software was determined using the concordance rate and Bland-Altman analysis via comparison with the CA.

A 2-sample t-test (P<0.001) and Fisher exact test (P=0.011) showed a significant difference between the normal CA and the BA estimated by the DL software. There was good correlation between the 2 variables (r=0.96, P<0.001); however, the root mean square error was 15.4 months. With a 12-month cutoff, the concordance rate was 58.8%. The Bland-Altman plot showed that the DL software tended to underestimate the BA compared with the CA, especially in children under the age of 8.3 years.

The DL-based BA software showed a low concordance rate and a tendency to underestimate the BA in healthy Korean children.

Artificial intelligence (AI) in medicine is transforming healthcare by automating system tasks, assisting in diagnostics, predicting patient outcomes and personalising patient care, founded on the ability to analyse vast datasets. In paediatric endocrinology, AI has been developed for diabetes, for insulin dose adjustment, detection of hypoglycaemia and retinopathy screening; bone age assessment and thyroid nodule screening; the identification of growth disorders; the diagnosis of precocious puberty; and the use of facial recognition algorithms in conditions such as Cushing syndrome, acromegaly, congenital adrenal hyperplasia and Turner syndrome. AI can also predict those most at risk from childhood obesity by stratifying future interventions to modify lifestyle. AI will facilitate personalised healthcare by integrating data from 'omics' analysis, lifestyle tracking, medical history, laboratory and imaging, therapy response and treatment adherence from multiple sources. As data acquisition and processing becomes fundamental, data privacy and protecting children's health data is crucial. Minimising algorithmic bias generated by AI analysis for rare conditions seen in paediatric endocrinology is an important determinant of AI validity in clinical practice. AI cannot create the patient-doctor relationship or assess the wider holistic determinants of care. Children have individual needs and vulnerabilities and are considered in the context of family relationships and dynamics. Importantly, whilst AI provides value through augmenting efficiency and accuracy, it must not be used to replace clinical skills.

The well-known Greulich and Pyle (GP) method of bone age assessment (BAA) relies on comparing a hand X-ray against templates of discrete maturity classes collected in an atlas. Automated methods have recently shown great success with BAA, especially using deep learning. In this perspective, we first review the success and limitations of various automated BAA methods. We then offer a novel hypothesis: When networks predict bone age that is not aligned with a GP reference class, it is not simply statistical error (although there is that as well); they are picking up nuances in the hand X-ray that lie "outside that class." In other words, trained networks predict distributions around classes. This raises a natural question: How can we further understand the reasons for a prediction to deviate from the nominal class age? We claim that segmental aging, that is, ratings based on characteristic bone groups can be used to qualify predictions. This so-called segmental GP method has excellent properties: It can not only help identify differential maturity in the hand but also provide a systematic way to extend the use of the current GP atlas to various other populations.

The medial clavicle epiphysis is a crucial indicator for bone age estimation (BAE) after hand maturation. This study aimed to develop machine learning (ML) and deep learning (DL) models for BAE based on medial clavicle CT images and evaluate the performance on normal and variant clavicles. This study retrospectively collected 1049 patients (mean± SD: 22.50±4.34 years) and split them into normal training and test sets, and variant training and test sets. An additional 53 variant clavicles were incorporated into the variant test set. The development stages of normal MCE were used to build a linear model and support vector machine (SVM) for BAE. The CT slices of MCE were automatically segmented and used to train DL models for automated BAE. Comparisons were performed by linear versus ML versus DL, and normal versus variant clavicles. Mean absolute error (MAE) and classification accuracy was the primary parameter of comparison. For BAE, the SVM had the best MAE of 1.73 years, followed by the commonly-used CNNs (1.77-1.93 years), the linear model (1.94 years), and the hybrid neural network CoAt Net (2.01 years). In DL models, SE Net 18 was the best-performing DL model with similar results to SVM in the normal test set and achieved an MAE of 2.08 years in the external variant test. For age classification, all the models exhibit superior performance in the classification of 18-, 20-, 21-, and 22-year thresholds with limited value in the 16-year threshold. Both ML and DL models produce desirable performance in BAE based on medial clavicle CT.

In the past, dentistry heavily relied on manual image analysis and diagnostic procedures, which could be time-consuming and prone to human error. The advent of artificial intelligence (AI) has brought transformative potential to the field, promising enhanced accuracy and efficiency in various dental imaging tasks. This systematic review and meta-analysis aimed to comprehensively evaluate the applications of AI in dental imaging modalities, focusing on in-vitro studies.

A systematic literature search was conducted, in accordance with the PRISMA guidelines. The following databases were systematically searched: PubMed/MEDLINE, Embase, Web of Science, Scopus, IEEE Xplore, Cochrane Library, CINAHL (Cumulative Index to Nursing and Allied Health Literature), and Google Scholar. The meta-analysis employed fixed-effects models to assess AI accuracy, calculating odds ratios (OR) for true positive rate (TPR), true negative rate (TNR), positive predictive value (PPV), and negative predictive value (NPV) with 95 % confidence intervals (CI). Heterogeneity and overall effect tests were applied to ensure the reliability of the findings.

9 studies were selected that encompassed various objectives, such as tooth segmentation and classification, caries detection, maxillofacial bone segmentation, and 3D surface model creation. AI techniques included convolutional neural networks (CNNs), deep learning algorithms, and AI-driven tools. Imaging parameters assessed in these studies were specific to the respective dental tasks. The analysis of combined ORs indicated higher odds of accurate dental image assessments, highlighting the potential for AI to improve TPR, TNR, PPV, and NPV. The studies collectively revealed a statistically significant overall effect in favor of AI in dental imaging applications.

In summary, this systematic review and meta-analysis underscore the transformative impact of AI on dental imaging. AI has the potential to revolutionize the field by enhancing accuracy, efficiency, and time savings in various dental tasks. While further research in clinical settings is needed to validate these findings and address study limitations, the future implications of integrating AI into dental practice hold great promise for advancing patient care and the field of dentistry.

Background: Manual bone age assessment (BAA) is associated with longer interpretation time and higher cost and variability, thus posing challenges in areas with restricted medical facilities, such as the high-altitude Tibetan Plateau. The application of artificial intelligence (AI) for automating BAA could facilitate resolving this issue. This study aimed to develop an AI-based BAA model for Han and Tibetan children. Methods: A model named "EVG-BANet" was trained using three datasets, including the Radiology Society of North America (RSNA) dataset (training set n = 12611, validation set n = 1425, and test set n = 200), the Radiological Hand Pose Estimation (RHPE) dataset (training set n = 5491, validation set n = 713, and test set n = 79), and a self-established local dataset [training set n = 825 and test set n = 351 (Han n = 216 and Tibetan n = 135)]. An open-access state-of-the-art model BoNet was used for comparison. The accuracy and generalizability of the two models were evaluated using the abovementioned three test sets and an external test set (n = 256, all were Tibetan). Mean absolute difference (MAD) and accuracy within 1 year were used as indicators. Bias was evaluated by comparing the MAD between the demographic groups. Results: EVG-BANet outperformed BoNet in the MAD on the RHPE test set (0.52 vs. 0.63 years, p < 0.001), the local test set (0.47 vs. 0.62 years, p < 0.001), and the external test set (0.53 vs. 0.66 years, p < 0.001) and exhibited a comparable MAD on the RSNA test set (0.34 vs. 0.35 years, p = 0.934). EVG-BANet achieved accuracy within 1 year of 97.7% on the local test set (BoNet 90%, p < 0.001) and 89.5% on the external test set (BoNet 85.5%, p = 0.066). EVG-BANet showed no bias in the local test set but exhibited a bias related to chronological age in the external test set. Conclusion: EVG-BANet can accurately predict the bone age (BA) for both Han children and Tibetan children living in the Tibetan Plateau with limited healthcare facilities.

The growing prevalence of musculoskeletal diseases increases radiologic workload, highlighting the need for optimized workflow management and automated metadata classification systems. We developed a large-scale, well-characterized dataset of musculoskeletal radiographs and trained deep learning neural networks to classify radiographic projection and body side.

In this IRB-approved retrospective single-center study, a dataset of musculoskeletal radiographs from 2011 to 2019 was retrieved and manually labeled for one of 45 possible radiographic projections and the depicted body side. Two classification networks were trained for the respective tasks using the Xception architecture with a custom network top and pretrained weights. Performance was evaluated on a hold-out test sample, and gradient-weighted class activation mapping (Grad-CAM) heatmaps were computed to visualize the influential image regions for network predictions.

A total of 13,098 studies comprising 23,663 radiographs were included with a patient-level dataset split, resulting in 19,183 training, 2,145 validation, and 2,335 test images. Focusing on paired body regions, training for side detection included 16,319 radiographs (13,284 training, 1,443 validation, and 1,592 test images). The models achieved an overall accuracy of 0.975 for projection and 0.976 for body-side classification on the respective hold-out test sample. Errors were primarily observed in projections with seamless anatomical transitions or non-orthograde adjustment techniques.

The deep learning neural networks demonstrated excellent performance in classifying radiographic projection and body side across a wide range of musculoskeletal radiographs. These networks have the potential to serve as presorting algorithms, optimizing radiologic workflow and enhancing patient care.

The developed networks excel at classifying musculoskeletal radiographs, providing valuable tools for research data extraction, standardized image sorting, and minimizing misclassifications in artificial intelligence systems, ultimately enhancing radiology workflow efficiency and patient care.

• A large-scale, well-characterized dataset was developed, covering a broad spectrum of musculoskeletal radiographs. • Deep learning neural networks achieved high accuracy in classifying radiographic projection and body side. • Grad-CAM heatmaps provided insight into network decisions, contributing to their interpretability and trustworthiness. • The trained models can help optimize radiologic workflow and manage large amounts of data.

To develop a deep learning model to distinguish rheumatoid arthritis (RA) from osteoarthritis (OA) using hand radiographs and to evaluate the effects of changing pretraining and training parameters on model performance.

A convolutional neural network was retrospectively trained on 9714 hand radiograph exams from 8387 patients obtained from 2017 to 2021 at seven hospitals within an integrated healthcare network. Performance was assessed using an independent test set of 250 exams from 146 patients. Binary discriminatory capacity (no arthritis versus arthritis; RA versus not RA) and three-way classification (no arthritis versus OA versus RA) were evaluated. The effects of additional pretraining using musculoskeletal radiographs, using all views as opposed to only the posteroanterior view, and varying image resolution on model performance were also investigated. Area under the receiver operating characteristic curve (AUC) and Cohen's kappa coefficient were used to evaluate diagnostic performance.

For no arthritis versus arthritis, the model achieved an AUC of 0.975 (95% CI: 0.957, 0.989). For RA versus not RA, the model achieved an AUC of 0.955 (95% CI: 0.919, 0.983). For three-way classification, the model achieved a kappa of 0.806 (95% CI: 0.742, 0.866) and accuracy of 87.2% (95% CI: 83.2%, 91.2%) on the test set. Increasing image resolution increased performance up to 1024 × 1024 pixels. Additional pretraining on musculoskeletal radiographs and using all views did not significantly affect performance.

A deep learning model can be used to distinguish no arthritis, OA, and RA on hand radiographs with high performance.

With potential of deep learning in musculoskeletal image interpretation being explored, this paper focuses on the common site of rotator cuff tears, the supraspinatus. It aims to propose and validate a deep learning model to automatically extract the supraspinatus, verifying its superiority through comparison with several classical image segmentation models.

Imaging data were retrospectively collected from 60 patients who underwent inpatient treatment for rotator cuff tears at a hospital between March 2021 and May 2023. A dataset of the supraspinatus from MRI was constructed after collecting, filtering, and manually annotating at the pixel level. This paper proposes a novel A-DAsppUnet network that can automatically extract the supraspinatus after training and optimization. The analysis of model performance is based on three evaluation metrics: precision, intersection over union, and Dice coefficient.

The experimental results demonstrate that the precision, intersection over union, and Dice coefficients of the proposed model are 99.20%, 83.38%, and 90.94%, respectively. Furthermore, the proposed model exhibited significant advantages over the compared models.

The designed model in this paper accurately extracts the supraspinatus from MRI, and the extraction results are complete and continuous with clear boundaries. The feasibility of using deep learning methods for musculoskeletal extraction and assisting in clinical decision-making was verified. This research holds practical significance and application value in the field of utilizing artificial intelligence for assisting medical decision-making.

Double-balloon enteroscopy (DBE) is a standard method for diagnosing and treating small bowel disease. However, DBE may yield false-negative results due to oversight or inexperience. We aim to develop a computer-aided diagnostic (CAD) system for the automatic detection and classification of small bowel abnormalities in DBE.

A total of 5201 images were collected from Renmin Hospital of Wuhan University to construct a detection model for localizing lesions during DBE, and 3021 images were collected to construct a classification model for classifying lesions into four classes, protruding lesion, diverticulum, erosion & ulcer and angioectasia. The performance of the two models was evaluated using 1318 normal images and 915 abnormal images and 65 videos from independent patients and then compared with that of 8 endoscopists. The standard answer was the expert consensus.

For the image test set, the detection model achieved a sensitivity of 92% (843/915) and an area under the curve (AUC) of 0.947, and the classification model achieved an accuracy of 86%. For the video test set, the accuracy of the system was significantly better than that of the endoscopists (85% vs. 77 ± 6%, p < 0.01). For the video test set, the proposed system was superior to novices and comparable to experts.

We established a real-time CAD system for detecting and classifying small bowel lesions in DBE with favourable performance. ENDOANGEL-DBE has the potential to help endoscopists, especially novices, in clinical practice and may reduce the miss rate of small bowel lesions.

Skeletal dysplasias collectively affect a large number of patients worldwide. Most of these disorders cause growth anomalies. Hence, evaluating skeletal maturity via the determination of bone age (BA) is a useful tool. Moreover, consecutive BA measurements are crucial for monitoring the growth of patients with such disorders, especially for timing hormonal treatment or orthopedic interventions. However, manual BA assessment is time-consuming and suffers from high intra- and inter-rater variability. This is further exacerbated by genetic disorders causing severe skeletal malformations. While numerous approaches to automate BA assessment have been proposed, few are validated for BA assessment on children with skeletal dysplasias.

We present Deeplasia, an open-source prior-free deep-learning approach designed for BA assessment specifically validated on patients with skeletal dysplasias.

We trained multiple convolutional neural network models under various conditions and selected three to build a precise model ensemble. We utilized the public BA dataset from the Radiological Society of North America (RSNA) consisting of training, validation, and test subsets containing 12,611, 1,425, and 200 hand and wrist radiographs, respectively. For testing the performance of our model ensemble on dysplastic hands, we retrospectively collected 568 radiographs from 189 patients with molecularly confirmed diagnoses of seven different genetic bone disorders including achondroplasia and hypochondroplasia. A subset of the dysplastic cohort (149 images) was used to estimate the test-retest precision of our model ensemble on longitudinal data.

The mean absolute difference of Deeplasia for the RSNA test set (based on the average of six different reference ratings) and dysplastic set (based on the average of two different reference ratings) were 3.87 and 5.84 months, respectively. The test-retest precision of Deeplasia on longitudinal data (2.74 months) is estimated to be similar to a human expert.

We demonstrated that Deeplasia is competent in assessing the age and monitoring the development of both normal and dysplastic bones.

While musculoskeletal imaging volumes are increasing, there is a relative shortage of subspecialized musculoskeletal radiologists to interpret the studies. Will artificial intelligence (AI) be the solution? For AI to be the solution, the wide implementation of AI-supported data acquisition methods in clinical practice requires establishing trusted and reliable results. This implementation will demand close collaboration between core AI researchers and clinical radiologists. Upon successful clinical implementation, a wide variety of AI-based tools can improve the musculoskeletal radiologist's workflow by triaging imaging examinations, helping with image interpretation, and decreasing the reporting time. Additional AI applications may also be helpful for business, education, and research purposes if successfully integrated into the daily practice of musculoskeletal radiology. The question is not whether AI will replace radiologists, but rather how musculoskeletal radiologists can take advantage of AI to enhance their expert capabilities.

The purpose of this study is to analyze the texture characteristics of chronic non-bacterial osteomyelitis (CNO) bone lesions, identified as areas of altered signal intensity on short tau inversion recovery (STIR) sequences, and to distinguish them from bone marrow growth-related changes through Machine Learning (ML) and Deep Learning (DL) analysis.

We included a group of 66 patients with confirmed diagnosis of CNO and a group of 28 patients with suspected extra-skeletal systemic disease. All examinations were performed on a 1.5 T MRI scanner. Using the opensource 3D Slicer software version 4.10.2, the ROIs on CNO lesions and on the red bone marrow were sampled. Texture analysis (TA) was carried out using Pyradiomics. We applied an optimization search grid algorithm on nine classic ML classifiers and a Deep Learning (DL) Neural Network (NN). The model's performance was evaluated using Accuracy (ACC), AUC-ROC curves, F1-score, Positive Predictive Value (PPV), Mean Absolute Error (MAE) and Root-Mean-Square Error (RMSE). Furthermore, we used Shapley additive explanations to gain insight into the behavior of the prediction model.

Most predictive characteristics were selected by Boruta algorithm for each combination of ROI sequences for the characterization and classification of the two types of signal hyperintensity. The overall best classification result was obtained by the NN with ACC = 0.91, AUC = 0.93 with 95% CI 0.91-0.94, F1-score = 0.94 and PPV = 93.8%. Between classic ML methods, ensemble learners showed high model performance; specifically, the best-performing classifier was the Stack (ST) with ACC = 0.85, AUC = 0.81 with 95% CI 0.8-0.84, F1-score = 0.9, PPV = 90%.

Our results show the potential of ML methods in discerning edema-like lesions, in particular by distinguishing CNO lesions from hematopoietic bone marrow changes in a pediatric population. The Neural Network showed the overall best results, while a Stacking classifier, based on Gradient Boosting and Random Forest as principal estimators and Logistic Regressor as final estimator, achieved the best results between the other ML methods.

Intracranial aneurysms have a high prevalence in human population. It also has a heavy burden of disease and high mortality rate in the case of rupture. Convolutional neural network(CNN) is a type of deep learning architecture which has been proven powerful to detect intracranial aneurysms.

Four databases were searched using artificial intelligence, intracranial aneurysms, and synonyms to find eligible studies. Articles which had applied CNN for detection of intracranial aneurisms were included in this review. Sensitivity and specificity of the models and human readers regarding modality, size, and location of aneurysms were sought to be extracted. Random model was the preferred model for analyses using CMA 2 to determine pooled sensitivity and specificity.

Overall, 20 studies were used in this review. Deep learning models could detect intracranial aneurysms with a sensitivity of 90/6% (CI: 87/2-93/2%) and specificity of 94/6% (CI: 0/914-0/966). CTA was the most sensitive modality (92.0%(CI:85/2-95/8%)). Overall sensitivity of the models for aneurysms more than 3 mm was above 98% (98%-100%) and 74.6 for aneurysms less than 3 mm. With the aid of AI, the clinicians' sensitivity increased to 12/8% and interrater agreement to 0/193.

CNN models had an acceptable sensitivity for detection of intracranial aneurysms, surpassing human readers in some fields. The logical approach for application of deep learning models would be its use as a highly capable assistant. In essence, deep learning models are a groundbreaking technology that can assist clinicians and allow them to diagnose intracranial aneurysms more accurately.

Determine the effect of a multiphase fusion deep-learning model with automatic phase selection in detection of intracranial aneurysm (IA) from computed tomography angiography (CTA) images.

CTA images of intracranial arteries from patients at Ningbo First Hospital were retrospectively analyzed. Images were randomly classified as training data, internal validation data, or test data. CTA images from cases examined by digital subtraction angiography (DSA) were examined for independent validation. A deep-learning model was constructed by automatic phase selection of multiphase fusion, and compared to the single-phase algorithm to evaluate algorithm sensitivity.

We analyzed 1110 patients (1493 aneurysms) as training data, 139 patients (174 aneurysms) as internal validation data, and 134 patients (175 aneurysms) as test data. The sensitivity of the multiphase analysis of the internal validation data, test data, and independent validation data were greater than from the single-phase analysis. The recall of the multiphase selection was greater or equal to that of single-phase selection in the aneurysm position, shape, size, and rupture status. Use of the test data to determine the presence and absence of aneurysm rupture led to a recall from multiphase selection of 94.8% and 87.6% respectively; both of these values were greater than those from single-phase selection (89.6% and 79.4%).

A multiphase fusion deep learning model with automatic phase selection provided automated detection of IAs with high sensitivity.

To develop a deep-learning-based bone age prediction model optimized for Korean children and adolescents and evaluate its feasibility by comparing it with a Greulich-Pyle-based deep-learning model.

A convolutional neural network was trained to predict age according to the bone development shown on a hand radiograph (bone age) using 21036 hand radiographs of Korean children and adolescents without known bone development-affecting diseases/conditions obtained between 1998 and 2019 (median age [interquartile range {IQR}], 9 [7-12] years; male:female, 11794:9242) and their chronological ages as labels (Korean model). We constructed 2 separate external datasets consisting of Korean children and adolescents with healthy bone development (Institution 1: n = 343; median age [IQR], 10 [4-15] years; male: female, 183:160; Institution 2: n = 321; median age [IQR], 9 [5-14] years; male: female, 164:157) to test the model performance. The mean absolute error (MAE), root mean square error (RMSE), and proportions of bone age predictions within 6, 12, 18, and 24 months of the reference age (chronological age) were compared between the Korean model and a commercial model (VUNO Med-BoneAge version 1.1; VUNO) trained with Greulich-Pyle-based age as the label (GP-based model).

Compared with the GP-based model, the Korean model showed a lower RMSE (11.2 vs. 13.8 months; P = 0.004) and MAE (8.2 vs. 10.5 months; P = 0.002), a higher proportion of bone age predictions within 18 months of chronological age (88.3% vs. 82.2%; P = 0.031) for Institution 1, and a lower MAE (9.5 vs. 11.0 months; P = 0.022) and higher proportion of bone age predictions within 6 months (44.5% vs. 36.4%; P = 0.044) for Institution 2.

The Korean model trained using the chronological ages of Korean children and adolescents without known bone development-affecting diseases/conditions as labels performed better in bone age assessment than the GP-based model in the Korean pediatric population. Further validation is required to confirm its accuracy.

Artificial Intelligence (AI) is integrating itself throughout the medical community. AI's ability to analyze complex patterns and interpret large amounts of data will have considerable impact on all areas of medicine, including pediatric endocrinology. In this paper, we review and update the current studies of AI in pediatric endocrinology. Specific topics that are addressed include: diabetes management, bone growth, metabolism, obesity, and puberty. Becoming knowledgeable and comfortable with AI will assist pediatric endocrinologists, the goal of the paper.