Current and emerging artificial intelligence applications for pediatric musculoskeletal radiology

Diagnostic performance of an AI algorithm for the detection of appendicular bone fractures in pediatric patients

PURPOSE

To evaluate the diagnostic performance of an Artificial Intelligence (AI) algorithm, previously trained using both adult and pediatric patients, for the detection of acute appendicular fractures in the pediatric population on conventional X-ray radiography (CXR).

MATERIALS AND METHODS

In this retrospective study, anonymized extremities CXRs of pediatric patients (age <17 years), with or without fractures, were included. Six hundred CXRs (maintaining the positive-for-fracture and negative-for-fracture balance) were included, grouping them per body part (shoulder/clavicle, elbow/upper arm, hand/wrist, leg/knee, foot/ankle). Follow-up CXRs and/or second-level imaging were considered as reference standard. A deep learning algorithm interpreted CXRs for fracture detection on a per-patient, per-radiograph, and per-location level, and its diagnostic performance values were compared with the reference standard. AI diagnostic performance was computed by using cross-tables, and 95 % confidence intervals [CIs] were obtained by bootstrapping.

RESULTS

The final cohort included 312 male and 288 female with a mean age of 8.9±4.5 years. Three undred CXRs (50 %) were positive for fractures, according to the reference standard. For all fractures, the AI tool showed a per-patient 91.3% (95%CIs = 87.6-94.3) sensitivity, 76.7% (71.5-81.3) specificity, and 84% (82.1-86.0) accuracy. In the per-radiograph analysis the AI tool showed 85% (81.9-87.8) sensitivity, 88.5% (86.3-90.4) specificity, and 87.2% (85.7-89.6) accuracy. In the per-location analysis, the AI tool identified 606 bounding boxes: 472 (77.9 %) were correct, 110 (18.1 %) incorrect, and 24 (4.0 %) were not-overlapping.

CONCLUSION

The AI algorithm provides good overall diagnostic performance for detecting appendicular fractures in pediatric patients.

Emerging technologies in pediatric radiology: current developments and future prospects

Radiological imaging is a crucial diagnostic tool for the pediatric population. However, it is associated with several unique challenges in this age group compared to adults. These challenges mainly come from the fact that children are not small-sized adults and differ in development, anatomy, physiology, and pathology compared to adults. This paper reviews relevant articles published between January 2015 and October 2023 to analyze challenges associated with imaging technologies currently used in pediatric radiology, emerging technologies, and their role in resolving the challenges and future prospects of pediatric radiology. In recent decades, imaging technologies have advanced rapidly, developing advanced ultrasound, computed tomography, magnetic resonance, nuclear imaging, teleradiology, artificial intelligence, machine learning, three-dimensional printing, radiomics, and radiogenomics, among many others. By prioritizing the unique needs of pediatric patients while developing such technologies, we can significantly alleviate the challenges faced in pediatric radiology.

Applicability and robustness of an artificial intelligence-based assessment for Greulich and Pyle bone age in a German cohort

PURPOSE

 The determination of bone age (BA) based on the hand and wrist, using the 70-year-old Greulich and Pyle (G&P) atlas, remains a widely employed practice in various institutions today. However, a more recent approach utilizing artificial intelligence (AI) enables automated BA estimation based on the G&P atlas. Nevertheless, AI-based methods encounter limitations when dealing with images that deviate from the standard hand and wrist projections. Generally, the extent to which BA, as determined by the G&P atlas, corresponds to the chronological age (CA) of a contemporary German population remains a subject of continued discourse. This study aims to address two main objectives. Firstly, it seeks to investigate whether the G&P atlas, as applied by the AI software, is still relevant for healthy children in Germany today. Secondly, the study aims to assess the performance of the AI software in handling non-strict posterior-anterior (p. a.) projections of the hand and wrist.

MATERIALS AND METHODS

 The AI software retrospectively estimated the BA in children who had undergone radiographs of a single hand using posterior-anterior and oblique planes. The primary purpose was to rule out any osseous injuries. The prediction error of BA in relation to CA was calculated for each plane and between the two planes.

RESULTS

 A total of 1253 patients (aged 3 to 16 years, median age 10.8 years, 55.7 % male) were included in the study. The average error of BA in posterior-anterior projections compared to CA was 3.0 (± 13.7) months for boys and 1.7 (± 13.7) months for girls. Interestingly, the deviation from CA tended to be even slightly lower in oblique projections than in posterior-anterior projections. The mean error in the posterior-anterior projection plane was 2.5 (± 13.7) months, while in the oblique plane it was 1.8 (± 13.9) months (p = 0.01).

CONCLUSION

 The AI software for BA generally corresponds to the age of the contemporary German population under study, although there is a noticeable prediction error, particularly in younger children. Notably, the software demonstrates robust performance in oblique projections.

KEY POINTS

  · Bone age, as determined by artificial intelligence, aligns with the chronological age of the contemporary German cohort under study.. · As determined by artificial intelligence, bone age is remarkably robust, even when utilizing oblique X-ray projections..

CITATION FORMAT

· Pape J, Hirsch F, Deffaa O et al. Applicability and robustness of an artificial intelligence-based assessment for Greulich and Pyle bone age in a German cohort. Fortschr Röntgenstr 2024; 196: 600 - 606.

Artificial intelligence model system for bone age assessment of preschool children

BACKGROUD

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.

METHODS

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.

RESULTS

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.

CONCLUSION

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.

IMPACT

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.

Adaptation and validation of an artificial intelligence based digital radiogrammetry tool for assessing bone health of indian children and youth with type-1 diabetes

BACKGROUND AND OBJECTIVES

BoneXpert (BX) is an artificial intelligence software used primarily for bone age assessment. Besides, it can also be used to screen for bone health using the digital radiogrammetry tool called bone health index (BHI) for which normative reference values available are calculated from healthy European children. Due to ethnic difference in bone geometry, in a previous study, we generated reference curves based on healthy Indian children. The objectives of this study were: 1) To assess and compare bone health of Indian children with Type 1 diabetes (T1D) using both European and Indian BHI SDS reference data and 2) To identify determinants of poor bone health in Indian children and youth with T1D by using BHI tool (based on BHI-SDS Indian reference data) of BX.

METHOD

The BHI was assessed retrospectively in 1159 subjects with T1D using digitalised left-hand x-rays and SDS were computed using European and Indian data. The demographic, anthropometric, clinical, biochemistry, dual x-ray absorptiometry (DXA) data and peripheral quantitative computed tomography (pQCT) data collection were performed using standard protocols and were extracted from hospital records.

RESULTS

The BHI correlated well with DXA and pQCT parameters in subjects with T1D. BHI-SDS calculated using Indian reference data had better correlation with height and DXA parameters. 8.6% study participants had low (less than -2) BHI-SDS (Indian), with height SDS having significant effect. Subjects with low BHI-SDS were older, shorter and had higher duration of diabetes. They also had lower IGF1 and vitamin D concentrations, bone mineral density, and trabecular density. Female gender, increased duration of illness, poor glycaemic control, and vitamin D deficiency/insufficiency were significant predictors of poor BHI-SDS.

CONCLUSION

Our study highlights the utility of digital radiogrammetry AI tool to screen for bone health of children with T1D and demonstrates and highlights the necessity of interpretation using ethnicity specific normative data.

Does ethnicity influence bone health index in children? A pilot study

BACKGROUND

Several pathological conditions can lead to variations in bone mineral content during growth. When assessing bone age, bone mineral content can be estimated without supplementary cost and irradiation. Manual assessment of bone quality using the Exton-Smith index (ESI) and automated assessment of the bone health index (BHI) provided by the BoneXpert® software are available but still not validated in different ethnic groups.

OBJECTIVE

Our aim is to provide normative values of the ESI and BHI for healthy European Caucasian and first-generation children of North Africans living in Europe.

MATERIALS AND METHODS

A sex- and aged-match population of 214 girls (107 European-Caucasian and 107 North African) and 220 boys (111 European-Caucasian and 109 North African) were retrospectively and consecutively included in the study. Normal radiographs of the left hand and wrist from healthy children were retrieved from those performed in a single institution from 2008 to 2017 to rule out a left-hand fracture. Radiographs were processed by BoneXpert® to obtain the BHI and BHI standard deviation score (SDS). One radiologist, blinded to BHI values, manually calculated ESI for each patient. The variability for both methods was assessed and compared using the standard deviation (SD) of the median (%) for each class of age and sex, and ESI and BHI trends were compared by sex and ethnic group.

RESULTS

The final population comprised 434 children ages 3 to 15 years (214 girls). Overall, BHI was lower in North African children (mean = 4.23 for girls and 4.17 in boys) than in European Caucasians (mean = 4.50 for girls and 4.68 in boys) (P < 0.001). Regardless of ethnicity, 29 girls (13.6%) and 34 boys (15.5%) had BHI more than 2 SD from the mean. While correlated to BHI, ESI has a higher variability than BHI and is more pronounced from 8-12 years for both sexes (mean ESI in European Caucasian girls and boys 17.47 and 20.87, respectively) (P < 0.001). ESI showed more than 15% variability in European girls from 8-12 years and a plateau in North African boys from 12 years to 16 years. However, the BHI has less than 15% variability regardless of age and ethnic group.

CONCLUSION

BHI may be a reliable tool to detect children with abnormal bone mineral content, with lower variability compared to ESI and with specific trends depending on sex and ethnicity.

Artificial intelligence in pediatric surgery

Artificial intelligence (AI) is rapidly changing the landscape of medicine and is already being utilized in conjunction with medical diagnostics and imaging analysis. We hereby explore AI applications in surgery and examine its relevance to pediatric surgery, covering its evolution, current state, and promising future. The various fields of AI are explored including machine learning and applications to predictive analytics and decision support in surgery, computer vision and image analysis in preoperative planning, image segmentation, surgical navigation, and finally, natural language processing assist in expediting clinical documentation, identification of clinical indications, quality improvement, outcome research, and other types of automated data extraction. The purpose of this review is to familiarize the pediatric surgical community with the rise of AI and highlight the ongoing advancements and challenges in its adoption, including data privacy, regulatory considerations, and the imperative for interdisciplinary collaboration. We hope this review serves as a comprehensive guide to AI's transformative influence on surgery, demonstrating its potential to enhance pediatric surgical patient outcomes, improve precision, and usher in a new era of surgical excellence.

Artificial intelligence-based detection of paediatric appendicular skeletal fractures: performance and limitations for common fracture types and locations

BACKGROUND

Research into artificial intelligence (AI)-based fracture detection in children is scarce and has disregarded the detection of indirect fracture signs and dislocations.

OBJECTIVE

To assess the diagnostic accuracy of an existing AI-tool for the detection of fractures, indirect fracture signs, and dislocations.

MATERIALS AND METHODS

An AI software, BoneView (Gleamer, Paris, France), was assessed for diagnostic accuracy of fracture detection using paediatric radiology consensus diagnoses as reference. Radiographs from a single emergency department were enrolled retrospectively going back from December 2021, limited to 1,000 radiographs per body part. Enrolment criteria were as follows: suspected fractures of the forearm, lower leg, or elbow; age 0-18 years; and radiographs in at least two projections.

RESULTS

Lower leg radiographs showed 607 fractures. Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were high (87.5%, 87.5%, 98.3%, 98.3%, respectively). Detection rate was low for toddler's fractures, trampoline fractures, and proximal tibial Salter-Harris-II fractures. Forearm radiographs showed 1,137 fractures. Sensitivity, specificity, PPV, and NPV were high (92.9%, 98.1%, 98.4%, 91.7%, respectively). Radial and ulnar bowing fractures were not reliably detected (one out of 11 radial bowing fractures and zero out of seven ulnar bowing fractures were correctly detected). Detection rate was low for styloid process avulsions, proximal radial buckle, and complete olecranon fractures. Elbow radiographs showed 517 fractures. Sensitivity and NPV were moderate (80.5%, 84.7%, respectively). Specificity and PPV were high (94.9%, 93.3%, respectively). For joint effusion, sensitivity, specificity, PPV, and NPV were moderate (85.1%, 85.7%, 89.5%, 80%, respectively). For elbow dislocations, sensitivity and PPV were low (65.8%, 50%, respectively). Specificity and NPV were high (97.7%, 98.8%, respectively).

CONCLUSIONS

The diagnostic performance of BoneView is promising for forearm and lower leg fractures. However, improvement is mandatory before clinicians can rely solely on AI-based paediatric fracture detection using this software.

BoneXpert-derived bone health index reference curves constructed on healthy Indian children and adolescents

BACKGROUND

Artificial intelligence (AI)-based applications for the assessment of the paediatric musculoskeletal system like BoneXpert are not only useful to assess bone age (BA) but also to provide a bone health index (BHI) and a standard deviation score (SDS) for both. This allows comparison of the BHI with age- and sex-matched healthy Caucasian children.

OBJECTIVE

We conducted this study with the objective of generating BHI curves using BoneXpert in healthy Indian children with BA between 2 and 17 years.

METHOD

We retrospectively reviewed anthropometric parameters, BHI, and BHI SDS data of digitalized left-hand radiographs (joint photographic experts group [jpg] format) of a cohort of 788 paediatric patients from a previous study to which they were recruited to compare various methods of BA assessment. The recruited children represented all age groups for both sexes. The corrected BHI for jpg images was calculated using the formula corrected BHI=BHI*(stature/(avL*50))^0.33333 where stature is height of subject and avL is average length of metacarpal bones. The reference Indian BHI curves and centiles were generated using the Lambda-Mu-Sigma method.

RESULT

The mean BHI and BHI SDS of the study group were 4.02±0.57 and -1.73±1.09, respectively. The average increase in median BHI from each age group was between 2.5% and 3% in both sexes up to age of 14 years after which it increased to 4.5% to 5%. The mean BHI of Indian children was lower than that of Caucasian children with maximum differences noted in boys at 16 years (21.7%) and girls at 14 years (16%). We report 8.4% SD of BHI for our study sample. Reference percentile curves for BHI according to BA were derived separately for boys and girls.

CONCLUSION

Reference data has been provided for the screening of bone health status of Indian children and adolescents.

Deeplasia: deep learning for bone age assessment validated on skeletal dysplasias

BACKGROUND

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.

OBJECTIVE

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

MATERIALS AND METHODS

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.

RESULTS

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.

CONCLUSION

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

Past, present, and future of global research on artificial intelligence applications in dermatology: A bibliometric analysis

In recent decades, artificial intelligence (AI) has played an increasingly important role in medicine, including dermatology. Worldwide, numerous studies have reported on AI applications in dermatology, rapidly increasing interest in this field. However, no bibliometric studies have been conducted to evaluate the past, present, or future of this topic. This study aimed to illustrate past and present research and outline future directions for global research on AI applications in dermatology using bibliometric analysis. We conducted an online search of the Web of Science Core Collection database to identify scientific papers on AI applications in dermatology. The bibliometric metadata of each selected paper were extracted, analyzed, and visualized using VOS viewer and Cite Space. A total of 406 papers, comprising 8 randomized controlled trials and 20 prospective studies, were deemed eligible for inclusion. The United States had the highest number of papers (n = 166). The University of California System (n = 24) and Allan C. Halpern (n = 11) were the institution and author with the highest number of papers, respectively. Based on keyword co-occurrence analysis, the studies were categorized into 9 distinct clusters, with clusters 2, 3, and 7 containing keywords with the latest average publication year. Wound progression prediction using machine learning, the integration of AI into teledermatology, and applications of the algorithms in skin diseases, are the current research priorities and will remain future research aims in this field.

Artificial intelligence to identify fractures on pediatric and young adult upper extremity radiographs

BACKGROUND

Pediatric fractures are challenging to identify given the different response of the pediatric skeleton to injury compared to adults, and most artificial intelligence (AI) fracture detection work has focused on adults.

OBJECTIVE

Develop and transparently share an AI model capable of detecting a range of pediatric upper extremity fractures.

MATERIALS AND METHODS

In total, 58,846 upper extremity radiographs (finger/hand, wrist/forearm, elbow, humerus, shoulder/clavicle) from 14,873 pediatric and young adult patients were divided into train (n = 12,232 patients), tune (n = 1,307), internal test (n = 819), and external test (n = 515) splits. Fracture was determined by manual inspection of all test radiographs and the subset of train/tune radiographs whose reports were classified fracture-positive by a rule-based natural language processing (NLP) algorithm. We trained an object detection model (Faster Region-based Convolutional Neural Network [R-CNN]; "strongly-supervised") and an image classification model (EfficientNetV2-Small; "weakly-supervised") to detect fractures using train/tune data and evaluate on test data. AI fracture detection accuracy was compared with accuracy of on-call residents on cases they preliminarily interpreted overnight.

RESULTS

A strongly-supervised fracture detection AI model achieved overall test area under the receiver operating characteristic curve (AUC) of 0.96 (95% CI 0.95-0.97), accuracy 89.7% (95% CI 88.0-91.3%), sensitivity 90.8% (95% CI 88.5-93.1%), and specificity 88.7% (95% CI 86.4-91.0%), and outperformed a weakly-supervised model (AUC 0.93, 95% CI 0.92-0.94, P < 0.0001). AI accuracy on cases preliminary interpreted overnight was higher than resident accuracy (AI 89.4% vs. 85.1%, 95% CI 87.3-91.5% vs. 82.7-87.5%, P = 0.01).

CONCLUSION

An object detection AI model identified pediatric upper extremity fractures with high accuracy.

Developing an artificial intelligence diagnostic tool for paediatric distal radius fractures, a proof of concept study

INTRODUCTION

In the UK, 1 in 50 children sustain a fractured bone yearly, yet studies have shown that 34% of children sustaining an injury do not have a visible fracture on initial radiographs. Wrist fractures are particularly difficult to identify because the growth plate poses diagnostic challenges when interpreting radiographs.

METHODS

We developed Convolutional Neural Network (CNN) image recognition software to detect fractures in radiographs from children. A consecutive data set of 5,000 radiographs of the distal radius in children aged less than 19 years from 2014 to 2019 was used to train the CNN. In addition, transfer learning from a VGG16 CNN pretrained on non-radiological images was applied to improve generalisation of the network and the classification of radiographs. Hyperparameter tuning techniques were used to compare the model with the radiology reports that accompanied the original images to determine diagnostic test accuracy.

RESULTS

The training set consisted of 2,881 radiographs with a fracture and 1,571 without; 548 radiographs were outliers. With additional augmentation, the final data set consisted of 15,498 images. The data set was randomly split into three subsets: training (70%), validation (10%) and test (20%). After training for 20 epochs, the diagnostic test accuracy was 85%.

DISCUSSION

A CNN model is feasible in diagnosing paediatric wrist fractures. We demonstrated that this application could be utilised as a tool for improving diagnostic accuracy. Future work would involve developing automated treatment pathways for diagnosis, reducing unnecessary hospital visits and allowing staff redeployment to other areas.

Comparison of diagnostic performance of a deep learning algorithm, emergency physicians, junior radiologists and senior radiologists in the detection of appendicular fractures in children

BACKGROUND

Advances have been made in the use of artificial intelligence (AI) in the field of diagnostic imaging, particularly in the detection of fractures on conventional radiographs. Studies looking at the detection of fractures in the pediatric population are few. The anatomical variations and evolution according to the child's age require specific studies of this population. Failure to diagnose fractures early in children may lead to serious consequences for growth.

OBJECTIVE

To evaluate the performance of an AI algorithm based on deep neural networks toward detecting traumatic appendicular fractures in a pediatric population. To compare sensitivity, specificity, positive predictive value and negative predictive value of different readers and the AI algorithm.

MATERIALS AND METHODS

This retrospective study conducted on 878 patients younger than 18 years of age evaluated conventional radiographs obtained after recent non-life-threatening trauma. All radiographs of the shoulder, arm, elbow, forearm, wrist, hand, leg, knee, ankle and foot were evaluated. The diagnostic performance of a consensus of radiology experts in pediatric imaging (reference standard) was compared with those of pediatric radiologists, emergency physicians, senior residents and junior residents. The predictions made by the AI algorithm and the annotations made by the different physicians were compared.

RESULTS

The algorithm predicted 174 fractures out of 182, corresponding to a sensitivity of 95.6%, a specificity of 91.64% and a negative predictive value of 98.76%. The AI predictions were close to that of pediatric radiologists (sensitivity 98.35%) and that of senior residents (95.05%) and were above those of emergency physicians (81.87%) and junior residents (90.1%). The algorithm identified 3 (1.6%) fractures not initially seen by pediatric radiologists.

CONCLUSION

This study suggests that deep learning algorithms can be useful in improving the detection of fractures in children.

Biochemical identification of prepubertal boys with Klinefelter syndrome by combined reproductive hormone profiling using machine learning

Objective

Klinefelter syndrome (KS) is the most common sex chromosome disorder and genetic cause of infertility in males. A highly variable phenotype contributes to the fact that a large proportion of cases are never diagnosed. Typical hallmarks in adults include small testes and azoospermia which may prompt biochemical evaluation that typically shows extremely high follicle-stimulating hormone and low/undetectable inhibin B serum concentrations. However, in prepubertal KS individuals, biochemical parameters are largely overlapping those of prepubertal controls. We aimed to characterize clinical profiles of prepubertal boys with KS in relation to controls and to develop a novel biochemical classification model to identify KS before puberty.

Methods

Retrospective, longitudinal data from 15 prepubertal boys with KS and data from 1475 controls were used to calculate age- and sex-adjusted standard deviation scores (SDS) for height and serum concentrations of reproductive hormones and used to infer a decision tree classification model for KS.

Results

Individual reproductive hormones were low but within reference ranges and did not discriminate KS from controls. Clinical and biochemical profiles including age- and sex-adjusted SDS from multiple reference curves provided input data to train a 'random forest' machine learning (ML) model for the detection of KS. Applied to unseen data, the ML model achieved a classification accuracy of 78% (95% CI, 61-94%).

Conclusions

Supervised ML applied to clinically relevant variables enabled computational classification of control and KS profiles. The application of age- and sex-adjusted SDS provided robust predictions irrespective of age. Specialized ML models applied to combined reproductive hormone concentrations may be useful diagnostic tools to improve the identification of prepubertal boys with KS.

European Society of Paediatric Radiology Artificial Intelligence taskforce: a new taskforce for the digital age

A new task force dedicated to artificial intelligence (AI) with respect to paediatric radiology was created in 2021 at the International Paediatric Radiology (IPR) meeting in Rome, Italy (a joint society meeting by the European Society of Pediatric Radiology [ESPR] and the Society for Pediatric Radiology [SPR]). The concept of a separate task force dedicated to AI was borne from an ESPR-led international survey of health care professionals' opinions, expectations and concerns regarding AI integration within children's imaging departments. In this survey, the majority (> 80%) of ESPR respondents supported the creation of a task force and helped define our key objectives. These include providing educational content about AI relevant for paediatric radiologists, brainstorming ideas for future projects and collaborating on AI-related studies with respect to collating data sets, de-identifying images and engaging in multi-case, multi-reader studies. This manuscript outlines the starting point of the ESPR AI task force and where we wish to go.

Automated detection of acute appendicular skeletal fractures in pediatric patients using deep learning

OBJECTIVE

We aimed to perform an external validation of an existing commercial AI software program (BoneView™) for the detection of acute appendicular fractures in pediatric patients.

MATERIALS AND METHODS

In our retrospective study, anonymized radiographic exams of extremities, with or without fractures, from pediatric patients (aged 2-21) were included. Three hundred exams (150 with fractures and 150 without fractures) were included, comprising 60 exams per body part (hand/wrist, elbow/upper arm, shoulder/clavicle, foot/ankle, leg/knee). The Ground Truth was defined by experienced radiologists. A deep learning algorithm interpreted the radiographs for fracture detection, and its diagnostic performance was compared against the Ground Truth, and receiver operating characteristic analysis was done. Statistical analyses included sensitivity per patient (the proportion of patients for whom all fractures were identified) and sensitivity per fracture (the proportion of fractures identified by the AI among all fractures), specificity per patient, and false-positive rate per patient.

RESULTS

There were 167 boys and 133 girls with a mean age of 10.8 years. For all fractures, sensitivity per patient (average [95% confidence interval]) was 91.3% [85.6, 95.3], specificity per patient was 90.0% [84.0,94.3], sensitivity per fracture was 92.5% [87.0, 96.2], and false-positive rate per patient in patients who had no fracture was 0.11. The patient-wise area under the curve was 0.93 for all fractures. AI diagnostic performance was consistently high across all anatomical locations and different types of fractures except for avulsion fractures (sensitivity per fracture 72.7% [39.0, 94.0]).

CONCLUSION

The BoneView™ deep learning algorithm provides high overall diagnostic performance for appendicular fracture detection in pediatric patients.

Assessment of an artificial intelligence aid for the detection of appendicular skeletal fractures in children and young adults by senior and junior radiologists

BACKGROUND

As the number of conventional radiographic examinations in pediatric emergency departments increases, so, too, does the number of reading errors by radiologists.

OBJECTIVE

The aim of this study is to investigate the ability of artificial intelligence (AI) to improve the detection of fractures by radiologists in children and young adults.

MATERIALS AND METHODS

A cohort of 300 anonymized radiographs performed for the detection of appendicular fractures in patients ages 2 to 21 years was collected retrospectively. The ground truth for each examination was established after an independent review by two radiologists with expertise in musculoskeletal imaging. Discrepancies were resolved by consensus with a third radiologist. Half of the 300 examinations showed at least 1 fracture. Radiographs were read by three senior pediatric radiologists and five radiology residents in the usual manner and then read again immediately after with the help of AI.

RESULTS

The mean sensitivity for all groups was 73.3% (110/150) without AI; it increased significantly by almost 10% (P<0.001) to 82.8% (125/150) with AI. For junior radiologists, it increased by 10.3% (P<0.001) and for senior radiologists by 8.2% (P=0.08). On average, there was no significant change in specificity (from 89.6% to 90.3% [+0.7%, P=0.28]); for junior radiologists, specificity increased from 86.2% to 87.6% (+1.4%, P=0.42) and for senior radiologists, it decreased from 95.1% to 94.9% (-0.2%, P=0.23). The stand-alone sensitivity and specificity of the AI were, respectively, 91% and 90%.

CONCLUSION

With the help of AI, sensitivity increased by an average of 10% without significantly decreasing specificity in fracture detection in a predominantly pediatric population.

Reference Curves for Pediatric Endocrinology: Leveraging Biomarker Z-Scores for Clinical Classifications

CONTEXT

Hormone reference intervals in pediatric endocrinology are traditionally partitioned by age and lack the framework for benchmarking individual blood test results as normalized z-scores and plotting sequential measurements onto a chart. Reference curve modeling is applicable to endocrine variables and represents a standardized method to account for variation with gender and age.

OBJECTIVE

We aimed to establish gender-specific biomarker reference curves for clinical use and benchmark associations between hormones, pubertal phenotype, and body mass index (BMI).

METHODS

Using cross-sectional population sample data from 2139 healthy Norwegian children and adolescents, we analyzed the pubertal status, ultrasound measures of glandular breast tissue (girls) and testicular volume (boys), BMI, and laboratory measurements of 17 clinical biomarkers modeled using the established "LMS" growth chart algorithm in R.

RESULTS

Reference curves for puberty hormones and pertinent biomarkers were modeled to adjust for age and gender. Z-score equivalents of biomarker levels and anthropometric measurements were compiled in a comprehensive beta coefficient matrix for each gender. Excerpted from this analysis and independently of age, BMI was positively associated with female glandular breast volume (β = 0.5, P < 0.001) and leptin (β = 0.6, P < 0.001), and inversely correlated with serum levels of sex hormone-binding globulin (SHBG) (β = -0.4, P < 0.001). Biomarker z-score profiles differed significantly between cohort subgroups stratified by puberty phenotype and BMI weight class.

CONCLUSION

Biomarker reference curves and corresponding z-scores provide an intuitive framework for clinical implementation in pediatric endocrinology and facilitate the application of machine learning classification and covariate precision medicine for pediatric patients.

Application of deep-learning-based artificial intelligence in acetabular index measurement

OBJECTIVE

To construct an artificial intelligence system to measure acetabular index and evaluate its accuracy in clinical application.

METHODS

A total of 10,219 standard anteroposterior pelvic radiographs were collected retrospectively from April 2014 to December 2018 in our hospital. Of these, 9,219 radiographs were randomly selected to train and verify the system. The remaining 1,000 radiographs were used to compare the system's and the clinicians' measurement results. All plain pelvic films were labeled by an expert committee through PACS system based on a uniform standard to measure acetabular index. Subsequently, eight other clinicians independently measured the acetabular index from 200 randomly selected radiographs from the test radiographs. Bland-Altman test was used for consistency analysis between the system and clinician measurements.

RESULTS

The test set included 1,000 cases (2,000 hips). Compared with the expert committee measurement, the 95% limits of agreement (95% LOA) of the system was -4.02° to 3.45° (bias = -0.27°, P < 0.05). The acetabular index measured by the system within all age groups, including normal and abnormal groups, also showed good credibility according to the Bland-Altman principle. Comparison of the measurement evaluations by the system and eight clinicians vs. that of, the expert committee, the 95% LOA of the clinician with the smallest measurement error was -2.76° to 2.56° (bias = -0.10°, P = 0.126). The 95% LOA of the system was -0.93° to 2.86° (bias = -0.03°, P = 0.647). The 95% LOA of the clinician with the largest measurement error was -3.41° to 4.25° (bias = 0.42°, P < 0.05). The measurement error of the system was only greater than that of a senior clinician.

CONCLUSION

The newly constructed artificial intelligence system could quickly and accurately measure the acetabular index of standard anteroposterior pelvic radiographs. There is good data consistency between the system in measuring standard anteroposterior pelvic radiographs. The accuracy of the system is closer to that of senior clinicians.