Bone age assessment: automated techniques coming of age?

An artificial intelligence-based bone age assessment model for Han and Tibetan children

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.

Applicability of Shoulder, Olecranon, and Wrist-based Skeletal Maturity Estimation Systems to the Modern Pediatric Population

BACKGROUND

The proximal humerus ossification system (PHOS), olecranon apophyseal ossification system (OAOS), and modified Fels wrist skeletal maturity system (mFWS) were recently developed or updated using a historical, mostly White, pediatric population. These upper extremity skeletal maturity systems have demonstrated skeletal age estimation performance superior or equivalent to Greulich and Pyle in historical patients. Their applicability to modern pediatric populations has not yet been evaluated.

METHODS

We reviewed anteroposterior shoulder, lateral elbow, and anteroposterior hand and wrist x-rays of 4 pediatric cohorts: White males, Black males, White females, and Black females. Peripubertal x-rays were evaluated: males 9 to17 years and females 7 to 15 years. Five nonpathologic radiographs for each age and joint were randomly selected from each group. Skeletal age estimates made by each of the 3 skeletal maturity systems were plotted against the chronological age associated with each radiograph and compared between cohorts, and with the historical patients.

RESULTS

Five hundred forty modern radiographs were evaluated (180 shoulders, 180 elbows, and 180 wrists). All radiographic parameters had inter- and intra-rater reliability coefficients at or above 0.79, indicating very good reliability. For PHOS, White males had delayed skeletal age compared with Black males (Δ-0.12 y, P=0.02) and historical males (Δ-0.17 y, P<0.001). Black females were skeletally advanced compared with historical females (Δ0.11 y, P=0.01). For OAOS, White males (Δ-0.31 y, P<0.001) and Black males (Δ-0.24 y, P<0.001) had delayed skeletal age compared with historical males. For mFWS, White males (Δ0.29 y, P=0.024), Black males (Δ0.58 y, P<0.001), and Black females (Δ0.44 y, P<0.001) had advanced skeletal age compared with historical counterparts of the same sex. All other comparisons were not significant (P>0.05).

CONCLUSIONS

The PHOS, OAOS, and mFWS have mild discrepancies in skeletal age estimates when applied to modern pediatric populations depending on the race and sex of the patient.

LEVEL OF EVIDENCE

Level III - retrospective chart review.

The uncovered biases and errors in clinical determination of bone age by using deep learning models

OBJECTIVES

To evaluate AI biases and errors in estimating bone age (BA) by comparing AI and radiologists' clinical determinations of BA.

METHODS

We established three deep learning models from a Chinese private dataset (CHNm), an American public dataset (USAm), and a joint dataset combining the above two datasets (JOIm). The test data CHNt (n = 1246) were labeled by ten senior pediatric radiologists. The effects of data site differences, interpretation bias, and interobserver variability on BA assessment were evaluated. The differences between the AI models' and radiologists' clinical determinations of BA (normal, advanced, and delayed BA groups by using the Brush data) were evaluated by the chi-square test and Kappa values. The heatmaps of CHNm-CHNt were generated by using Grad-CAM.

RESULTS

We obtained an MAD value of 0.42 years on CHNm-CHNt; this result indicated an appropriate accuracy for the whole group but did not indicate an accurate estimation of individual BA because with a kappa value of 0.714, the agreement between AI and human clinical determinations of BA was significantly different. The features of the heatmaps were not fully consistent with the human vision on the X-ray films. Variable performance in BA estimation by different AI models and the disagreement between AI and radiologists' clinical determinations of BA may be caused by data biases, including patients' sex and age, institutions, and radiologists.

CONCLUSIONS

The deep learning models outperform external validation in predicting BA on both internal and joint datasets. However, the biases and errors in the models' clinical determinations of child development should be carefully considered.

KEY POINTS

• With a kappa value of 0.714, clinical determinations of bone age by using AI did not accord well with clinical determinations by radiologists. • Several biases, including patients' sex and age, institutions, and radiologists, may cause variable performance by AI bone age models and disagreement between AI and radiologists' clinical determinations of bone age. • AI heatmaps of bone age were not fully consistent with human vision on X-ray films.

Prediction of Fishman's skeletal maturity indicators using artificial intelligence

The present study aimed to evaluate the performance of automated skeletal maturation assessment system for Fishman's skeletal maturity indicators (SMI) for the use in dental fields. Skeletal maturity is particularly important in orthodontics for the determination of treatment timing and method. SMI is widely used for this purpose, as it is less time-consuming and practical in clinical use compared to other methods. Thus, the existing automated skeletal age assessment system based on Greulich and Pyle and Tanner-Whitehouse3 methods was further developed to include SMI using artificial intelligence. This hybrid SMI-modified system consists of three major steps: (1) automated detection of region of interest; (2) automated evaluation of skeletal maturity of each region; and (3) SMI stage mapping. The primary validation was carried out using a dataset of 2593 hand-wrist radiographs, and the SMI mapping algorithm was adjusted accordingly. The performance of the final system was evaluated on a test dataset of 711 hand-wrist radiographs from a different institution. The system achieved a prediction accuracy of 0.772 and mean absolute error and root mean square error of 0.27 and 0.604, respectively, indicating a clinically reliable performance. Thus, it can be used to improve clinical efficiency and reproducibility of SMI prediction.

A real-time automated bone age assessment system based on the RUS-CHN method

Background

Bone age is the age of skeletal development and is a direct indicator of physical growth and development in children. Most bone age assessment (BAA) systems use direct regression with the entire hand bone map or first segmenting the region of interest (ROI) using the clinical a priori method and then deriving the bone age based on the characteristics of the ROI, which takes more time and requires more computation.

Materials and methods

Key bone grades and locations were determined using three real-time target detection models and Key Bone Search (KBS) post-processing using the RUS-CHN approach, and then the age of the bones was predicted using a Lightgbm regression model. Intersection over Union (IOU) was used to evaluate the precision of the key bone locations, while the mean absolute error (MAE), the root mean square error (RMSE), and the root mean squared percentage error (RMSPE) were used to evaluate the discrepancy between predicted and true bone age. The model was finally transformed into an Open Neural Network Exchange (ONNX) model and tested for inference speed on the GPU (RTX 3060).

Results

The three real-time models achieved good results with an average (IOU) of no less than 0.9 in all key bones. The most accurate outcomes for the inference results utilizing KBS were a MAE of 0.35 years, a RMSE of 0.46 years, and a RMSPE of 0.11. Using the GPU RTX3060 for inference, the critical bone level and position inference time was 26 ms. The bone age inference time was 2 ms.

Conclusions

We developed an automated end-to-end BAA system that is based on real-time target detection, obtaining key bone developmental grade and location in a single pass with the aid of KBS, and using Lightgbm to obtain bone age, capable of outputting results in real-time with good accuracy and stability, and able to be used without hand-shaped segmentation. The BAA system automatically implements the entire process of the RUS-CHN method and outputs information on the location and developmental grade of the 13 key bones of the RUS-CHN method along with the bone age to assist the physician in making judgments, making full use of clinical a priori knowledge.

Bone age assessment: Large-scale comparison of Greulich-Pyle method and Tanner-Whitehouse 3 method for Taiwanese children

BACKGROUND

The Greulich-Pyle (GP) and Tanner-Whitehouse 3 (TW3) methods are two common methods for assessing bone age (BA). The applicability of these methods for populations other than those in the United States and Europe has been questioned. Thus, this study tested the applicability of these methods for Taiwanese children.

METHODS

In total, 1476 radiographs (654 boys, 822 girls) were analyzed. A subset of 200 radiographs was evaluated to determine intrarater and interrater reliability and the time required to yield a BA assessment. BA was determined by two reviewers using the GP method and two of the TW3 methods (the Radial-Ulnar-Short bones [RUS] method and the carpals method [Carpal]). The GP and TW3 methods were directly compared using statistical techniques. A subgroup analysis by age was performed to compare BA and chronological age using a paired t test for each age group.

RESULTS

The average times required to yield an assessment using the GP and TW3-RUS methods were 0.79 ± 0.14 and 3.01 ± 0.84 min (p < 0.001), respectively. Both the intrarater and interrater correlation coefficients were higher for the GP method (0.993, 0.992) than the TW3-RUS (0.985, 0.984) and TW3-Carpal (0.981, 0.973) methods. The correlation coefficient for the GP and TW3-RUS methods was highest in the pubertal stage (0.898 for boys and 0.909 for girls). The mean absolute deviations for the GP and TW3-RUS methods in the pubertal stage were 0.468 years (boys) and 0.496 years (girls). Both the GP and TW3-Carpal methods underestimated BA for boys in the prepubertal stage. Both the GP and TW3-RUS methods overestimated BA for girls in the pubertal and postpubertal stages.

CONCLUSION

The GP and TW3-RUS methods exhibit strong agreement in the pubertal and postpubertal stages for both sexes. With appropriate adjustments based on Taiwanese data, both methods are applicable to our children.

A comprehensive validation study of the latest version of BoneXpert on a large cohort of Caucasian children and adolescents

INTRODUCTION

Automated bone age assessment has recently become increasingly popular. The aim of this study was to assess the agreement between automated and manual evaluation of bone age using the method according to Tanner-Whitehouse (TW3) and Greulich-Pyle (GP).

METHODS

We evaluated 1285 bone age scans from 1202 children (657 scans from 612 boys) by using both manual and automated (TW3 as well as GP) bone age assessment. BoneXpert software versions 2.4.5.1. (BX2) and 3.2.1. (BX3) (Visiana, Holte, Denmark) were compared with manual evaluation using root mean squared error (RMSE) analysis.

RESULTS

RMSE for BX2 was 0.57 and 0.55 years in boys and 0.72 and 0.59 years in girls, respectively for TW3 and GP. For BX3, RMSE was 0.51 and 0.68 years in boys and 0.49 and 0.52 years in girls, respectively for TW3 and GP. Sex- and age-specific analysis for BX2 identified the largest differences between manual and automated TW3 evaluation in girls between 6-7, 12-13, 13-14 and 14-15 years, with RMSE 0.88, 0.81, 0.92 and 0.84 years, respectively. The BX3 version showed better agreement with manual TW3 evaluation (RMSE 0.64, 0.45, 0.46 and 0.57).

CONCLUSION

The latest version of the BoneXpert software provides improved and clinically sufficient agreement with manual bone age evaluation in children of both sexes compared to the previous version and may be used for routine bone age evaluation in non-selected cases in pediatric endocrinology care.

Current and emerging artificial intelligence applications for pediatric musculoskeletal radiology

Artificial intelligence (AI) is playing an ever-increasing role in radiology (more so in the adult world than in pediatrics), to the extent that there are unfounded fears it will completely take over the role of the radiologist. In relation to musculoskeletal applications of AI in pediatric radiology, we are far from the time when AI will replace radiologists; even for the commonest application (bone age assessment), AI is more often employed in an AI-assist mode rather than an AI-replace or AI-extend mode. AI for bone age assessment has been in clinical use for more than a decade and is the area in which most research has been conducted. Most other potential indications in children (such as appendicular and vertebral fracture detection) remain largely in the research domain. This article reviews the areas in which AI is most prominent in relation to the pediatric musculoskeletal system, briefly summarizing the current literature and highlighting areas for future research. Pediatric radiologists are encouraged to participate as members of the research teams conducting pediatric radiology artificial intelligence research.

Bone Age Reading by DXA Images should not Replace Bone Age Reading by X-ray Images

X-ray image of the hand is the most used technique to estimate bone age in children. For the analysis of bone mineral density using DXA in children, bone age may help to adjust such measurement in some cases. During image acquisition in DXA, an anteroposterior image of the hand may be acquired and used to evaluate bone age but few studies have evaluated the agreement between conventional X-ray and DXA images. The aim of the study was to determine bone age estimation agreement between conventional X-ray images and DXA in children and adolescents aged 5 to 16 years of age. We performed an analytical cross-sectional study of 711 healthy subjects. Subject´s bone age, both in conventional X-ray, and DXA images were read independently by two expert evaluators blinded for chronological age. Intraobserver and inter-observer reproducibility were evaluated using Intraclass Correlation Coefficient (ICC), and the agreement between bone age estimations made by both evaluators was analyzed using ICC and Bland-Altman analysis. General agreement between techniques measured through ICC was 0.99 with a mean difference of 6 months between techniques being older the ages obtained by DXA. The agreement limits were around ±2 years, which means that 95% of all differences between techniques were covered within this range. We found a high level of ICC agreement in bone age readings from X-ray and DXA images although we observed overestimation of bone age measurements in DXA. Differences between techniques were greater in women than in men, especially at the ages corresponding to puberty. Bone age measurement in DXA images appears not to be reliable; hence it should be suggested to perform conventional radiography of the hand to assess bone age taking into account that X-ray images have better resolution.

Validation of automated bone age analysis from hand radiographs in a North American pediatric population

BACKGROUND

Radiographic bone age assessment by automated software is precise and instantaneous.

OBJECTIVE

The aim of this study was to evaluate the accuracy of an automated tool for bone age assessment.

MATERIALS AND METHODS

We compared a total of 586 bone age radiographs from 451 patients, which had been assessed by three radiologists from 2013 to 2018, with bone age analysis by BoneXpert, using the Greulich and Pyle method. We made bone age comparisons in different patient groups based on gender, diagnosis and race, and in a subset with repeated bone age studies. We calculated Spearman correlation (r) and accuracy (root mean square error, or R²).

RESULTS

Bone age analyses by automated and manual assessments showed a strong correlation (r=0.98; R²=0.96; P<0.0001), with the mean bone age difference of 0.12±0.76 years. Bone age comparisons by the two methods remained strongly correlated (P<0.0001) when stratified by gender, common endocrine conditions including growth disorders and early/precocious puberty, and race. In the longitudinal analysis, we also found a strong correlation between the automated software and manual bone age over time (r=0.7852; R²=0.63; P<0.01).

CONCLUSION

Automated bone age assessment was found to be reliable and accurate in a large cohort of pediatric patients in a clinical practice setting in North America.

Alternative methods for skeletal maturity estimation with the EOS scanner—Experience from 934 patients

Background

Hand-wrist bone age assessment methods are not possible on typical EOS 2D/3D images without body position modifications that may affect spinal position. We aimed to identify and assess lesser known bone age assessment alternatives that may be applied retrospectively and without the need for extra imaging.

Materials and methods

After review of 2857 articles, nine bone age methods were selected and applied retrospectively in pilot study (thirteen individuals), followed by evaluation of EOS images of 934 4-24-year-olds. Difficulty of assessment and time taken were recorded, and reliability calculated.

Results

Five methods proved promising after pilot study. Risser ‘plus’ could be applied with no difficulty in 89.5% of scans (836/934) followed by the Oxford hip method (78.6%, 734/934), cervical (79.0%, 738/934), calcaneus (70.8%, 669/934) and the knee (68.2%, 667/934). Calcaneus and cervical methods proved to be fastest at 17.7s (95% confidence interval, 16.0s to 19.38s & 26.5s (95% CI, 22.16s to 30.75s), respectively, with Oxford hip the slowest at 82.0 s (95% CI, 76.12 to 87.88s). Difficulties included: regions lying outside of the image—assessment was difficult or impossible in upper cervical vertebrae (46/934 images 4.9%) and calcaneus methods (144/934 images, 15.4%); position: lower step length was associated with difficult lateral knee assessment & head/hand position with cervical evaluation; and resolution: in the higher stages of the hip, calcaneal and knee methods.

Conclusions

Hip, iliac crest and cervical regions can be assessed on the majority of EOS scans and may be useful for retrospective application. Calcaneus evaluation is a simple and rapidly applicable method that may be appropriate if consideration is given to include full imaging of the foot.

Clinical application of artificial intelligence in longitudinal image analysis of bone age among GHD patients

OBJECTIVE

This study aims to explore the clinical value of artificial intelligence (AI)-assisted bone age assessment (BAA) among children with growth hormone deficiency (GHD).

METHODS

A total of 290 bone age (BA) radiographs were collected from 52 children who participated in the study at Sun Yat-sen Memorial Hospital between January 2016 and August 2017. Senior pediatric endocrinologists independently evaluated BA according to the China 05 (CH05) method, and their consistent results were regarded as the gold standard (GS). Meanwhile, two junior pediatric endocrinologists were asked to assessed BA both with and without assistance from the AI-based BA evaluation system. Six months later, around 20% of the images assessed by the junior pediatric endocrinologists were randomly selected to be re-evaluated with the same procedure half a year later. Root mean square error (RMSE), mean absolute error (MAE), accuracy, and Bland-Altman plots were used to compare differences in BA. The intra-class correlation coefficient (ICC) and one-way repeated ANOVA were used to assess inter- and intra-observer variabilities in BAA. A boxplot of BA evaluated by different raters during the course of treatment and a mixed linear model were used to illustrate inter-rater effect over time.

RESULTS

A total of 52 children with GHD were included, with mean chronological age and BA by GS of 6.64 ± 2.49 and 5.85 ± 2.30 years at baseline, respectively. After incorporating AI assistance, the performance of the junior pediatric endocrinologists improved (P < 0.001), with MAE and RMSE both decreased by more than 1.65 years (Rater 1: ΔMAE = 1.780, ΔRMSE = 1.655; Rater 2: ΔMAE = 1.794, ΔRMSE = 1.719), and accuracy increasing from approximately 10% to over 91%. The ICC also increased from 0.951 to 0.990. During GHD treatment (at baseline, 6-, 12-, 18-, and 24-months), the difference decreased sharply when AI was applied. Furthermore, a significant inter-rater effect (P = 0.002) also vanished upon AI involvement.

CONCLUSION

AI-assisted interpretation of BA can improve accuracy and decrease variability in results among junior pediatric endocrinologists in longitudinal cohort studies, which shows potential for further clinical application.

Autonomous artificial intelligence in pediatric radiology: the use and perception of BoneXpert for bone age assessment

BACKGROUND

The autonomous artificial intelligence (AI) system for bone age rating (BoneXpert) was designed to be used in clinical radiology practice as an AI-replace tool, replacing the radiologist completely.

OBJECTIVE

The aim of this study was to investigate how the tool is used in clinical practice. Are radiologists more inclined to use BoneXpert to assist rather than replace themselves, and how much time is saved?

MATERIALS AND METHODS

We sent a survey consisting of eight multiple-choice questions to 282 radiologists in departments in Europe already using the software.

RESULTS

The 97 (34%) respondents came from 18 countries. Their answers revealed that before installing the automated method, 83 (86%) of the respondents took more than 2 min per bone age rating; this fell to 20 (21%) respondents after installation. Only 17/97 (18%) respondents used BoneXpert to completely replace the radiologist; the rest used it to assist radiologists to varying degrees. For instance, 39/97 (40%) never overruled the automated reading, while 9/97 (9%) overruled more than 5% of the automated ratings. The majority 58/97 (60%) of respondents checked the radiographs themselves to exclude features of underlying disease.

CONCLUSION

BoneXpert significantly reduces reporting times for bone age determination. However, radiographic analysis involves more than just determining bone age. It also involves identification of abnormalities, and for this reason, radiologists cannot be completely replaced. AI systems originally developed to replace the radiologist might be more suitable as AI assist tools, particularly if they have not been validated to work autonomously, including the ability to omit ratings when the image is outside the range of validity.

Clinical Validation of a Deep Learning-Based Hybrid (Greulich-Pyle and Modified Tanner-Whitehouse) Method for Bone Age Assessment

OBJECTIVE

To evaluate the accuracy and clinical efficacy of a hybrid Greulich-Pyle (GP) and modified Tanner-Whitehouse (TW) artificial intelligence (AI) model for bone age assessment.

MATERIALS AND METHODS

A deep learning-based model was trained on an open dataset of multiple ethnicities. A total of 102 hand radiographs (51 male and 51 female; mean age ± standard deviation = 10.95 ± 2.37 years) from a single institution were selected for external validation. Three human experts performed bone age assessments based on the GP atlas to develop a reference standard. Two study radiologists performed bone age assessments with and without AI model assistance in two separate sessions, for which the reading time was recorded. The performance of the AI software was assessed by comparing the mean absolute difference between the AI-calculated bone age and the reference standard. The reading time was compared between reading with and without AI using a paired t test. Furthermore, the reliability between the two study radiologists' bone age assessments was assessed using intraclass correlation coefficients (ICCs), and the results were compared between reading with and without AI.

RESULTS

The bone ages assessed by the experts and the AI model were not significantly different (11.39 ± 2.74 years and 11.35 ± 2.76 years, respectively, p = 0.31). The mean absolute difference was 0.39 years (95% confidence interval, 0.33-0.45 years) between the automated AI assessment and the reference standard. The mean reading time of the two study radiologists was reduced from 54.29 to 35.37 seconds with AI model assistance (p < 0.001). The ICC of the two study radiologists slightly increased with AI model assistance (from 0.945 to 0.990).

CONCLUSION

The proposed AI model was accurate for assessing bone age. Furthermore, this model appeared to enhance the clinical efficacy by reducing the reading time and improving the inter-observer reliability.

Assessment of rapidly advancing bone age during puberty on elbow radiographs using a deep neural network model

OBJECTIVES

Bone age is considered an indicator for the diagnosis of precocious or delayed puberty and a predictor of adult height. We aimed to evaluate the performance of a deep neural network model in assessing rapidly advancing bone age during puberty using elbow radiographs.

METHODS

In all, 4437 anteroposterior and lateral pairs of elbow radiographs were obtained from pubertal individuals from two institutions to implement and validate a deep neural network model. The reference standard bone age was established by five trained researchers using the Sauvegrain method, a scoring system based on the shapes of the lateral condyle, trochlea, olecranon apophysis, and proximal radial epiphysis. A test set (n = 141) was obtained from an external institution. The differences between the assessment of the model and that of reviewers were compared.

RESULTS

The mean absolute difference (MAD) in bone age estimation between the model and reviewers was 0.15 years on internal validation. In the test set, the MAD between the model and the five experts ranged from 0.19 to 0.30 years. Compared with the reference standard, the MAD was 0.22 years. Interobserver agreement was excellent among reviewers (ICC: 0.99) and between the model and the reviewers (ICC: 0.98). In the subpart analysis, the olecranon apophysis exhibited the highest accuracy (74.5%), followed by the trochlea (73.7%), lateral condyle (73.7%), and radial epiphysis (63.1%).

CONCLUSIONS

Assessment of rapidly advancing bone age during puberty on elbow radiographs using our deep neural network model was similar to that of experts.

KEY POINTS

• Bone age during puberty is particularly important for patients with scoliosis or limb-length discrepancy to determine the phase of the disease, which influences the timing and method of surgery. • The commonly used hand radiographs-based methods have limitations in assessing bone age during puberty due to the less prominent morphological changes of the hand and wrist bones in this period. • A deep neural network model trained with elbow radiographs exhibited similar performance to human experts on estimating rapidly advancing bone age during puberty.

Automated Bone Age Assessment Using Artificial Intelligence: The Future of Bone Age Assessment

Bone age assessments are a complicated and lengthy process, which are prone to inter- and intra-observer variabilities. Despite the great demand for fully automated systems, developing an accurate and robust bone age assessment solution has remained challenging. The rapidly evolving deep learning technology has shown promising results in automated bone age assessment. In this review article, we will provide information regarding the history of automated bone age assessments, discuss the current status, and present a literature review, as well as the future directions of artificial intelligence-based bone age assessments.

Imaging of pediatric bone and growth disorders: Of diagnostic workhorses and new horizons

Children and adolescents with bone and growth disorders require interdisciplinary care from various specialists including pediatric radiologists with a focus on musculoskeletal disorders. This article covers routine topics, differential diagnoses, and selected research imaging in children with osteogenesis imperfecta (OI), X‑linked hypophosphatemic rickets (XLH), achondroplasia, and other bone and growth disorders from the standpoint of a tertiary referral center.

Performance of an artificial intelligence system for bone age assessment in Tibet

OBJECTIVE

To investigate whether bone age (BA) of children living in Tibet Highland could be accurately assessed using a fully automated artificial intelligence (AI) system.

METHODS：

Left hand radiographs of 385 children (300 Tibetan and 85 immigrant Han) aged 4-18 years who presented to the largest medical center of Tibet between September 2013 and November 2019 were consecutively collected. From these radiographs, BA was determined using the Greulich and Pyle (GP) method by experts in a consensus manner; furthermore, BA was estimated by a previously reported artificial intelligence (AI) BA system based on Han children from southern China. The performance of the AI system was compared with that of experts by using statistical analysis.

RESULTS

Compared with the experts' results, the accuracy of the AI system for Tibetan and Han children within 1 year was 84.67 and 89.41%, respectively, and its mean absolute difference (MAD) was 0.65 and 0.56 years, respectively. The discrepancy in hand-wrist bone maturation was the main cause of low accuracy of the system in the 4- to 6-year-old group.

CONCLUSION

The AI BA system developed for Han Chinese children living in flat regions could enable to assess BA accurately in Tibet where medical resources are limited.

ADVANCES IN KNOWLEDGE

AI-based BA system may serve as an effective and efficient solution to assess BA in Tibet.

Evaluation of Bone Age in Children: A Mini-Review

Bone age represents a common index utilized in pediatric radiology and endocrinology departments worldwide for the definition of skeletal maturity for medical and non-medical purpose. It is defined by the age expressed in years that corresponds to the level of maturation of bones. Although several bones have been studied to better define bone age, the hand and wrist X-rays are the most used images. In fact, the images obtained by hand and wrist X-ray reflect the maturity of different types of bones of the skeletal segment evaluated. This information, associated to the characterization of the shape and changes of bone components configuration, represent an important factor of the biological maturation process of a subject. Bone age may be affected by several factors, including gender, nutrition, as well as metabolic, genetic, and social factors and either acute and chronic pathologies especially hormone alteration. As well several differences can be characterized according to the numerous standardized methods developed over the past decades. Therefore, the complete characterization of the main methods and procedure available and particularly of all their advantages and disadvantages need to be known in order to properly utilized this information for all its medical and non-medical main fields of application.

Artificial intelligence system can achieve comparable results to experts for bone age assessment of Chinese children with abnormal growth and development

Objective

Bone age (BA) is a crucial indicator for revealing the growth and development of children. This study tested the performance of a fully automated artificial intelligence (AI) system for BA assessment of Chinese children with abnormal growth and development.

Materials and Methods

A fully automated AI system based on the Greulich and Pyle (GP) method was developed for Chinese children by using 8,000 BA radiographs from five medical centers nationwide in China. Then, a total of 745 cases (360 boys and 385 girls) with abnormal growth and development from another tertiary medical center of north China were consecutively collected between January and October 2018 to test the system. The reference standard was defined as the result interpreted by two experienced reviewers (a radiologist with 10 years and an endocrinologist with 15 years of experience in BA reading) through consensus using the GP atlas. BA accuracy within 1 year, root mean square error (RMSE), mean absolute difference (MAD), and 95% limits of agreement according to the Bland-Altman plot were statistically calculated.

Results

For Chinese pediatric patients with abnormal growth and development, the accuracy of this new automated AI system within 1 year was 84.60% as compared to the reference standard, with the highest percentage of 89.45% in the 12- to 18-year group. The RMSE, MAD, and 95% limits of agreement of the AI system were 0.76 years, 0.58 years, and -1.547 to 1.428, respectively, according to the Bland-Altman plot. The largest difference between the AI and experts' BA result was noted for patients of short stature with bone deformities, severe osteomalacia, or different rates of maturation of the carpals and phalanges.

Conclusions

The developed automated AI system could achieve comparable BA results to experienced reviewers for Chinese children with abnormal growth and development.

Diagnostic performance of convolutional neural network-based Tanner-Whitehouse 3 bone age assessment system

Background

Bone age can reflect the true growth and development status of a child; thus, it plays a critical role in evaluating growth and endocrine disorders. This study established and validated an optimized Tanner-Whitehouse 3 artificial intelligence (TW3-AI) bone age assessment (BAA) system based on a convolutional neural network (CNN).

Methods

A data set of 9,059 clinical radiographs of the left hand was obtained from the picture archives and communication systems (PACS) between January 2012 and December 2016. Among these, 8,005/9,059 (88%) samples were treated as the training set for model implementation, 804/9,059 (9%) samples as the validation set for parameters optimization, and the remaining 250/9,059 (3%) samples were used to verify the accuracy and reliability of the model compared to that of 4 experienced endocrinologists and 2 experienced radiologists. The overall variation of TW3-metacarpophalangeal, radius, ulna and short bones (RUS) and TW3-Carpal bone score, as well as each bone (13 RUS + 7 Carpal) between reviewers and the AI, were compared by Bland-Altman (BA) chart and Kappa test, respectively. Furthermore, the time consumption between the model and reviewers was also compared.

Results

The performance of TW3-AI model was highly consistent with the reviewers' overall estimation, and the root mean square (RMS) was 0.50 years. The accuracy of the BAA of the TW3-AI model was better than the estimate of the reviewers. Further analysis revealed that human interpretations of the male capitate, hamate, the first distal and fifth middle phalanx and female capitate, the trapezoid, and the third and fifth middle phalanx, were most inconsistent. The average image processing time was 1.5±0.2 s in the TW3-AI model, which was significantly shorter than manual interpretation.

Conclusions

The diagnostic performance of CNN-based TW3 BAA was accurate and timesaving, and possesses better stability compared to diagnostics made by experienced experts.

DIFFERENCES IN BONE AGE READINGS BETWEEN PEDIATRIC ENDOCRINOLOGISTS AND RADIOLOGISTS

Objective: Pediatric endocrinologists (PEs) have historically read their own bone age (BA) X-rays based on the belief that radiologists do not accurately interpret these tests. Whether there are significant differences in BA interpretations between these two groups has not been systematically explored. The objectives of the study were to compare BA readings performed by PEs and radiologists and determine whether clinical variables were associated with discrepancies in readings. Methods: A retrospective chart review of children presenting for initial evaluation of short stature (SS) or precocious puberty (PP) who had a BA X-ray completed was performed. Clinical variables analyzed included age, gender, ethnicity, Tanner stage, body mass index, reason for referral, radiologist location (Children's vs. outside hospital), and PE and radiologist BA readings using the Greulich and Pyle method. Results: Of 103 patients aged 9 ± 3.66 years, there was a discrepancy between the PE and radiologist readings on 70 images (68%). Discrepancy ranged from -1.5 to 3.5 years, with a mean of 4 ± 12 months. Patients referred for PP were more likely to have discrepant interpretations than those referred for SS (8.4 months vs. 0.8 months; P = .007). No differences were seen in interpretations between in-house radiologists and those at outside hospitals. Conclusion: Radiologists interpreted BAs differently than PEs in the majority of images. In patients referred for PP, BAs were interpreted as being older by radiologists than by PEs, perhaps due to bias from the reason for referral. Our results provide support for continued independent BA interpretations by PEs. Abbreviations: BA = bone age; GP = Greulich and Pyle; PE = pediatric endocrinologist; PP = precocious puberty; SS = short stature.

The importance of anthropological methods in the diagnosis of rare diseases

Most of inborn errors of metabolism (IEMs) and rare endocrine-metabolic diseases (REMD) are rare diseases. According to the European Commission on Public Health, a rare disease is defined, based on its prevalence, as one affecting one in 2000 people. Many IEMs affect body stature, cause craniofacial abnormalities, and disturb the developmental process. Therefore, body proportion, dysmorphic characteristics, and morphological parameters must be assessed and closely monitored. This can be achieved only with the help of an anthropologist who has adequate tools. This is why the role of an anthropologist in collaboration with the physician in the diagnostic process is not to be underestimated. Clinical anthropologists contribute to assessing physical development and improve our understanding of the natural history of rare metabolic diseases. This paper presents anthropometric techniques and methods, such as analysis of demographic data, anthropometric parameters at birth, percentile charts, growth patterns, bioimpedance, somatometric profiles, craniofacial profiles, body proportion indices, and mathematical models of growth curves used in certain rare diseases. Contemporary anthropological methods play an important role in the diagnostic process of rare genetic diseases.

Forensic age estimation for pelvic X-ray images using deep learning

PURPOSE

To develop a deep learning bone age assessment model based on pelvic radiographs for forensic age estimation and compare its performance to that of the existing cubic regression model.

MATERIALS AND METHOD

A retrospective collection data of 1875 clinical pelvic radiographs between 10 and 25 years of age was obtained to develop the model. Model performance was assessed by comparing the testing results to estimated ages calculated directly using the existing cubic regression model based on ossification staging methods. The mean absolute error (MAE) and root-mean-squared error (RMSE) between the estimated ages and chronological age were calculated for both models.

RESULTS

For all test samples (between 10 and 25 years old), the mean MAE and RMSE between the automatic estimates using the proposed deep learning model and the reference standard were 0.94 and 1.30 years, respectively. For the test samples comparable to those of the existing cubic regression model (between 14 and 22 years old), the mean MAE and RMSE for the deep learning model were 0.89 and 1.21 years, respectively. For the existing cubic regression model, the mean MAE and RMSE were 1.05 and 1.61 years, respectively.

CONCLUSION

The deep learning convolutional neural network model achieves performance on par with the existing cubic regression model, demonstrating predictive ability capable of automated skeletal bone assessment based on pelvic radiographic images.

KEY POINTS

• The pelvis has considerable value in determining the bone age. • Deep learning can be used to create an automated bone age assessment model based on pelvic radiographs. • The deep learning convolutional neural network model achieves performance on par with the existing cubic regression model.

Reducing acquisition time for MRI-based forensic age estimation

Radiology-based estimation of a living person’s unknown age has recently attracted increasing attention due to large numbers of undocumented immigrants entering Europe. To avoid the application of X-ray-based imaging techniques, magnetic resonance imaging (MRI) has been suggested as an alternative imaging modality. Unfortunately, MRI requires prolonged acquisition times, which potentially represents an additional stressor for young refugees. To eliminate this shortcoming, we investigated the degree of reduction in acquisition time that still led to reliable age estimates. Two radiologists randomly assessed original images and two sets of retrospectively undersampled data of 15 volunteers (N = 45 data sets) applying an established radiological age estimation method to images of the hand and wrist. Additionally, a neural network-based age estimation method analyzed four sets of further undersampled images from the 15 volunteers (N = 105 data sets). Furthermore, we compared retrospectively undersampled and acquired undersampled data for three volunteers. To assess reliability with increasing degree of undersampling, intra-rater and inter-rater agreement were analyzed computing signed differences and intra-class correlation. While our findings have to be confirmed by a larger prospective study, the results from both radiological and automatic age estimation showed that reliable age estimation was still possible for acquisition times of 15 seconds.

A simple method for bone age assessment: the capitohamate planimetry

Objectives

To determine if the capitohamate (CH) planimetry could be a reliable indicator of bone age, and to compare it with Greulich-Pyle (GP) method.

Methods

This retrospective study included 391 children (age, 1–180 months). Two reviewers manually measured the areas of the capitate and hamate on plain radiographs. CH planimetry was defined as the measurement of the sum of areas of the capitate and hamate. Two reviewers independently applied the CH planimetry and GP methods in 109 children whose heights were at the 50th percentile of the growth chart.

Results

There was a strong positive correlation between chronological age and CH planimetry measurement (right, r = 0.9702; left, r = 0.9709). There was no significant difference in accuracy between CH planimetry (84.39–84.46 %) and the GP method (85.15–87.66 %) (p ≥ 0.0867). The interobserver reproducibility of CH planimetry (precision, 4.42 %; 95 % limits of agreement [LOA], −10.5 to 13.4 months) was greater than that of the GP method (precision, 8.45 %; LOA, −29.5 to 21.1 months).

Conclusions

CH planimetry may be a reliable method for bone age assessment.

Key Points

• Bone age assessment is important in the work-up of paediatric endocrine disorders.

• Radiography of the left hand is widely used to estimate bone age.

• Capitatohamate planimetry is a reliable and reproducible method for assessing bone age.

A Critical Appraisal of the Effect of Gonadotropin-Releasing Hormon Analog Treatment on Adult Height of Girls with Central Precocious Puberty

Central precocious puberty (CPP) is a diagnosis that pediatric endocrinologists worldwide increasingly make in girls of age 6-8 years and is mostly idiopathic. Part of the reason for increasing referral and diagnosis is the perception among the doctors as well as the patients that treatment of CPP with long-acting gonadotropin-releasing hormon analogues (GnRHa) promote height of the child. Although, the timing and the tempo of puberty does influence statural growth and achieved adult height, the extent of this effect is variable depending on several factors and is modest in most cases. Studies investigating GnRHa treatment in girls with idiopathic CPP demonstrate that treatment is able to restore adult height compromised by precocious puberty. However, reports on untreated girls with precocious puberty demonstrate that some of these girls achieve their target height without treatment as well, thus, blurring the net effect of GnRHa treatment on height in girls with CPP. Clinical studies on treatment of girls with idiopathic CPP on adult stature suffers from the solid evidence-base due mainly to the lack of well-designed randomized controlled studies and our insufficiencies of predicting adult height of a child with narrow precision. This is particularly true for girls in whom age of pubertal onset is close to physiological age of puberty, which are the majority of cases treated with GnRHa nowadays. Heterogeneous nature of pubertal tempo (progressive vs. nonprogressive) leading to different height outcomes also complicates the interpretation of the results in both treated and untreated cases. This review will attemp to summarize and critically appraise available data in the field.

Bone Age: A Handy Tool for Pediatric Providers

Pediatricians have relied on methods for determining skeletal maturation for >75 years. Bone age continues to be a valuable tool in assessing children's health. New technology for bone age determination includes computer-automated readings and assessments obtained from alternative imaging modalities. In addition, new nonclinical bone age applications are evolving, particularly pertaining to immigration and children's rights to asylum. Given the significant implications when bone ages are used in high-stake decisions, it is necessary to recognize recently described limitations in predicting accurate age in various ethnicities and diseases. Current methods of assessing skeletal maturation are derived from primarily white populations. In modern studies, researchers have explored the accuracy of bone age across various ethnicities in the United States. Researchers suggest there is evidence that indicates the bone ages obtained from current methods are less generalizable to children of other ethnicities, particularly children with African and certain Asian backgrounds. Many of the contemporary methods of bone age determination may be calibrated to individual populations and hold promise to perform better in a wider range of ethnicities, but more data are needed.

Performance of a Deep-Learning Neural Network Model in Assessing Skeletal Maturity on Pediatric Hand Radiographs

Purpose To compare the performance of a deep-learning bone age assessment model based on hand radiographs with that of expert radiologists and that of existing automated models. Materials and Methods The institutional review board approved the study. A total of 14 036 clinical hand radiographs and corresponding reports were obtained from two children's hospitals to train and validate the model. For the first test set, composed of 200 examinations, the mean of bone age estimates from the clinical report and three additional human reviewers was used as the reference standard. Overall model performance was assessed by comparing the root mean square (RMS) and mean absolute difference (MAD) between the model estimates and the reference standard bone ages. Ninety-five percent limits of agreement were calculated in a pairwise fashion for all reviewers and the model. The RMS of a second test set composed of 913 examinations from the publicly available Digital Hand Atlas was compared with published reports of an existing automated model. Results The mean difference between bone age estimates of the model and of the reviewers was 0 years, with a mean RMS and MAD of 0.63 and 0.50 years, respectively. The estimates of the model, the clinical report, and the three reviewers were within the 95% limits of agreement. RMS for the Digital Hand Atlas data set was 0.73 years, compared with 0.61 years of a previously reported model. Conclusion A deep-learning convolutional neural network model can estimate skeletal maturity with accuracy similar to that of an expert radiologist and to that of existing automated models. ^© RSNA, 2017 An earlier incorrect version of this article appeared online. This article was corrected on January 19, 2018.

The follicle-stimulating hormone (FSH) and luteinizing hormone (LH) response to a gonadotropin-releasing hormone analogue test in healthy prepubertal girls aged 10 months to 6 years

OBJECTIVE

Premature thelarche and precocious puberty are frequently diagnosed in girls even below 6 years of age and may be difficult to differentiate in the early stages. A GnRH test is often included in the diagnostic work-up, although interpretation of the GnRH test in girls below 6 years of age is challenging, as no reference interval exists for this age group. The objective is to determine the normal FSH and LH response to a GnRH test in healthy prepubertal girls below 6 years of age.

DESIGN AND METHODS

A standardized GnRH test, baseline reproductive hormones, clinical evaluation and bone age were determined in all participants. Forty-eight healthy normal-weight girls aged 3.5 ± 0.2 years (range: 0.8-5.9 years) were included. Serum concentrations of LH and FSH were measured before and 30 min after the gonadorelin injection.

RESULTS

The 30-min LH responses (mean ± 2 s.d.) were 5.2 ± 4.0 and 2.9 ± 2.5 IU/L and the FSH responses were 23.3 ± 16.2 and 14.5 ± 10.3 IU/L in girls aged 0.8-3.0 years and 3.0-5.9 years respectively. This corresponds to upper cut-off limits for LH of 9.2 IU/L (<3 years) and 5.3 IU/L (3-6 years). The stimulated LH/FSH ratio was 0.23 ± 0.19 (range 0.06-0.43) and did not correlate with age.

CONCLUSIONS

We found that LH increases up to 9.2 IU/L during GnRH test in healthy normal-weight girls below 3 years of age and that the stimulated LH/FSH ratio did not exceed 0.43. Our findings have important implications for appropriate diagnosis of central precocious puberty in girls below 6 years of age.

Maturation Disparity between Hand-Wrist Bones in a Chinese Sample of Normal Children: An Analysis Based on Automatic BoneXpert and Manual Greulich and Pyle Atlas Assessment

OBJECTIVE

To assess the maturation disparity of hand-wrist bones using the BoneXpert system and Greulich and Pyle (GP) atlas in a sample of normal children from China.

MATERIALS AND METHODS

Our study included 229 boys and 168 girls aged 2-14 years. The bones in the hand and wrist were divided into five groups: distal radius and ulna, metacarpals, proximal phalanges, middle phalanges and distal phalanges. Bone age (BA) was assessed separately using the automatic BoneXpert and GP atlas by two raters. Differences in the BA between the most advanced and retarded individual bones and bone groups were analyzed.

RESULTS

In 75.8% of children assessed with the BoneXpert and 59.4% of children assessed with the GP atlas, the BA difference between the most advanced and most retarded individual bones exceeded 2.0 years. The BA mean differences between the most advanced and most retarded individual bones were 2.58 and 2.25 years for the BoneXpert and GP atlas methods, respectively. Furthermore, for both methods, the middle phalanges were the most advanced group. The most retarded group was metacarpals for BoneXpert, while metacarpals and the distal radius and ulna were the most retarded groups according to the GP atlas. Overall, the BAs of the proximal and distal phalanges were closer to the chronological ages than those of the other bone groups.

CONCLUSION

Obvious and regular maturation disparities are common in normal children. Overall, the BAs of the proximal and distal phalanges are more useful for BA estimation than those of the other bone groups.