RESUMEN
INTRODUCTION: Automated bone age assessment (BAA) is of growing interest because of its accuracy and time efficiency in daily practice. In this study, we validated the clinical applicability of a commercially available artificial intelligence (AI)-powered X-ray bone age analyzer equipped with a deep learning-based automated BAA system and compared its performance with that of the Tanner-Whitehouse 3 (TW-3) method. METHODS: Radiographs prospectively collected from 30 centers across various regions in China, including 900 Chinese children and adolescents, were assessed independently by six doctors (three experts and three residents) and an AI analyzer for TW3 radius, ulna, and short bones (RUS) and TW3 carpal bone age. The experts' mean estimates were accepted as the gold standard. The performance of the AI analyzer was compared with that of each resident. RESULTS: For the estimation of TW3-RUS, the AI analyzer had a mean absolute error (MAE) of 0.48 ± 0.42. The percentage of patients with an absolute error of < 1.0 years was 86.78%. The MAE was significantly lower than that of rater 1 (0.54 ± 0.49, P = 0.0068); however, it was not significant for rater 2 (0.48 ± 0.48) or rater 3 (0.49 ± 0.46). For TW3 carpal, the AI analyzer had an MAE of 0.48 ± 0.65. The percentage of patients with an absolute error of < 1.0 years was 88.78%. The MAE was significantly lower than that of rater 2 (0.58 ± 0.67, P = 0.0018) and numerically lower for rater 1 (0.54 ± 0.64) and rater 3 (0.50 ± 0.53). These results were consistent for the subgroups according to sex, and differences between the age groups were observed. CONCLUSION: In this comprehensive validation study conducted in China, an AI-powered X-ray bone age analyzer showed accuracies that matched or exceeded those of doctor raters. This method may improve the efficiency of clinical routines by reducing reading time without compromising accuracy.
Assessing bone age, or how developed a child's skeleton is, is important in medical care, but the standard method can be time-consuming. Using AI to automatically assess bone age from X-ray images may improve efficiency without reducing accuracy. In this study, we evaluated how well an AI-powered X-ray bone age analyzer performed compared to the established TannerWhitehouse 3 (TW-3) method. X-ray images from 900 Chinese children and adolescents were collected from 30 centers. Six doctors (three experts, three residents) and the AI system independently assessed the TW-3 radius, ulna, and short bones (RUS) and TW-3 carpal bone age. The experts' assessments were considered the gold standard. The AI analyzer had an average error of 0.48 years for TW3-RUS bone age, with 87% of assessments within 1 year of the experts. For TW3 carpal bone age, the AI had an average error of 0.48 years, with 89% within 1 year. These results were similar to or better than those of the resident raters. These findings show the AI-powered analyzer can assess bone age as accurately as human raters. This technology may improve clinical efficiency by reducing the time required for bone age assessments without compromising accuracy.
RESUMEN
OBJECTIVE: To determine which bones and which grades had the highest inter-rater variability when employing the Tanner-Whitehouse (T-W) method. MATERIALS AND METHODS: Twenty-four radiologists were recruited and trained in the T-W classification of skeletal development. The consistency and skill of the radiologists in determining bone development status were assessed using 20 pediatric hand radiographs of children aged 1 to 18 years old. Four radiologists had a poor concordance rate and were excluded. The remaining 20 radiologists undertook a repeat reading of the radiographs, and their results were analyzed by comparing them with the mean assessment of two senior experts as the reference standard. Concordance rate, scoring, and Kendall's W were calculated to evaluate accuracy and consistency. RESULTS: Both the radius, ulna, and short finger (RUS) system (Kendall's W = 0.833) and the carpal (C) system (Kendall's W = 0.944) had excellent consistency, with the RUS system outperforming the C system in terms of scores. The repeatability analysis showed that the second rating test, performed after 2 months of further bone age assessment (BAA) practice, was more consistent and accurate than the first. The capitate had the lowest average concordance rate and scoring, as well as the lowest overall concordance rate for its D classification. Moreover, the G classifications of the seven carpal bones all had a concordance rate less than 0.6. The bones with lower Kendall's W were likewise those with lower scores and concordance rates. CONCLUSION: The D grade of the capitate showed the highest variation, and the use of the Tanner-Whitehouse 3rd edition (T-W3) to determine bone age (BA) was frequently inconsistent. A more comprehensive description with a focus on inaccuracy bones or ratings and a modification to the T-W3 approach would significantly advance BAA.
RESUMEN
Though the Greulich and Pyle (GP) method is easy, inter-observer variability, differential maturation of hand bones influences ratings. The Tanner-Whitehouse (TW) method is more accurate, but cumbersome. A simpler method combining the above, such that it utilizes fewer bones without affecting accuracy, would be widely used and more applicable in clinical practice. OBJECTIVES: 1. Devising a simplified method utilizing three bones of the hand and wrist for bone age (BA) assessment. 2. Testing whether the 3 bone method gives comparable results to standard methods (GP,TW2,TW3) in Indian children. METHODS: Developmental stages and corresponding BA for radius, hamate, terminal phalanx (left middle finger) epiphyses combining stages from GP,TW3 atlases were described; BA were rated by two blinded observers. 3 bone method ratings were compared with the same dataset analyzed earlier using GP,TW2,TW3 (4 raters). RESULTS: Radiographs analysed:493 (Girls=226). Mean chronological age:9.4 ± 4.6 yrs, mean BA 3 bone:9.8 ± 4.8 yrs, GP:9.6 ± 4.8 yrs, TW3:9.3 ± 4.5 yrs, TW2:9.9 ± 5.0 yrs. The 3 bone method demonstrated no significant inter-observer variability (p = 0.3, mean difference = 0.02 ± 0.6 yrs); a strong positive correlation (p < 0.0001) with GP (r = 0.985), TW3 (r = 0.983) and TW2 (r = 0.982) was noted. Bland-Altman plots demonstrated good agreement; the root mean square errors between 3 bone and GP,TW3,TW2 ratings were 0.6,0.7,0.6 years; mean differences were 0.19,0.49,-0.14 years respectively. Greatest proportion of outliers (beyond ±1.96 SD of mean difference) was between 6 and 8 years age for difference in 3 bone and GP, and between 4-6 years for difference in 3 bone and TW3,TW2. CONCLUSION: The 3 bone method has multiple advantages; it is easier, tackles differential maturation of wrist and hand bones, has good reproducibility, without compromising on accuracy rendering it suitable for office practice.
Asunto(s)
Determinación de la Edad por el Esqueleto , Huesos de la Mano , Determinación de la Edad por el Esqueleto/métodos , Humanos , Femenino , Niño , Masculino , Huesos de la Mano/diagnóstico por imagen , Huesos de la Mano/crecimiento & desarrollo , Huesos de la Mano/anatomía & histología , Adolescente , Preescolar , Radio (Anatomía)/diagnóstico por imagen , Radio (Anatomía)/anatomía & histología , Variaciones Dependientes del Observador , Falanges de los Dedos de la Mano/diagnóstico por imagen , Falanges de los Dedos de la Mano/anatomía & histología , Reproducibilidad de los Resultados , Muñeca/diagnóstico por imagen , Muñeca/anatomía & histología , Desarrollo Óseo/fisiologíaRESUMEN
Background: In 2020, our center established a Tanner-Whitehouse 3 (TW3) artificial intelligence (AI) system using a convolutional neural network (CNN), which was built upon 9059 radiographs. However, the system, upon which our study is based, lacked a gold standard for comparison and had not undergone thorough evaluation in different working environments. Methods: To further verify the applicability of the AI system in clinical bone age assessment (BAA) and to enhance the accuracy and homogeneity of BAA, a prospective multi-center validation was conducted. This study utilized 744 left-hand radiographs of patients, ranging from 1 to 20 years of age, with 378 boys and 366 girls. These radiographs were obtained from nine different children's hospitals between August and December 2020. The BAAs were performed using the TW3 AI system and were also reviewed by experienced reviewers. Bone age accuracy within 1 year, root mean square error (RMSE), and mean absolute error (MAE) were statistically calculated to evaluate the accuracy. Kappa test and Bland-Altman (B-A) plot were conducted to measure the diagnostic consistency. Results: The system exhibited a high level of performance, producing results that closely aligned with those of the reviewers. It achieved a RMSE of 0.52 years and an accuracy of 94.55% for the radius, ulna, and short bones series. When assessing the carpal series of bones, the system achieved a RMSE of 0.85 years and an accuracy of 80.38%. Overall, the system displayed satisfactory accuracy and RMSE, particularly in patients over 7 years old. The system excelled in evaluating the carpal bone age of patients aged 1-6. Both the Kappa test and B-A plot demonstrated substantial consistency between the system and the reviewers, although the model encountered challenges in consistently distinguishing specific bones, such as the capitate. Furthermore, the system's performance proved acceptable across different genders and age groups, as well as radiography instruments. Conclusions: In this multi-center validation, the system showcased its potential to enhance the efficiency and consistency of healthy delivery, ultimately resulting in improved patient outcomes and reduced healthcare costs.
RESUMEN
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.
Asunto(s)
Determinación de la Edad por el Esqueleto , Programas Informáticos , Adolescente , Niño , Femenino , Humanos , Masculino , Determinación de la Edad por el Esqueleto/métodos , Población BlancaRESUMEN
BACKGROUND: The applicability and accuracy of artificial intelligence (AI)-assisted bone age assessment and adult height prediction methods in girls with early puberty are unknown. OBJECTIVE: To analyze the performance of AI-assisted bone age assessment methods by comparing the corresponding methods for predicted adult height with actual adult height. MATERIALS AND METHODS: This retrospective review included 726 girls with early puberty, 87 of whom had reached adult height at last follow-up. Bone age was evaluated using the Greulich-Pyle (GP), Tanner-Whitehouse (TW3-RUS) and China 05 RUS-CHN (RUS-CHN) methods. Predicted adult height was calculated using the China 05 (CH05), TW3 and Bayley-Pinneau (BP) methods. RESULTS: We analyzed 1,663 left-hand radiographs, including 155 from girls who had reached adult height. In the 6-8- and 9-11-years age groups, bone age differences were smaller than those in the 12-14-years group; however, the differences between predicted adult height and actual adult height were larger than those in the 12-14-years group. TW3 overestimated adult height by 0.4±2.8 cm, while CH05 and BP significantly underestimated adult height by 2.9±3.6 cm and 1.3±3.8 cm, respectively. TW3 yielded the highest proportion of predicted adult height within ±5 cm of actual adult height (92.9%), with the highest correlation between predicted and actual adult heights. CONCLUSION: The differences in measured bone ages increased with increasing bone age. However, the corresponding method for predicting adult height was more accurate when the bone age was older. TW3 might be more suitable than CH05 and BP for predicting adult height in girls with early puberty. Methods for predicting adult height should be optimized for populations of the same ethnicity and disease.
Asunto(s)
Determinación de la Edad por el Esqueleto , Inteligencia Artificial , Estatura , Pueblos del Este de Asia , Adolescente , Niño , Femenino , Humanos , Determinación de la Edad por el Esqueleto/métodos , Pubertad , Pubertad Precoz , Estudios RetrospectivosRESUMEN
Objective:To investigate the differences between Tanner-Whitehouse (TW)3-Carpal and TW3-RUS(radius, ulna and short bone)-based artificial intelligence (AI)-assisted bone age assessment system using real world data.Methods:The image data of 262 children who received X-ray examination of left wrist in the Affiliated Children′s Hospital, Capital Institute of Pediatrics from July to September 2021 were retrospectively collected. The AI bone age assistant methods based on TW3-RUS and TW3-Carpal criteria were used to obtain the bone age results, respectively. Two senior pediatric radiologists evaluated the bone age on the basis of TW3-RUS and TW3-Carpal criteria, and the averaged values of two reviewers was calculated and taken as the gold standard reference. The cases were stratified into six age groups at 3-year intervals according to the gold standard reference, including 1-3 ( n=10), 4-6 ( n=35), 7-9 ( n=70), 10-12 ( n=118), 13-15 ( n=27) and 16-18 ( n=2) years old groups. Intraclass correlation coefficient (ICC) was used to evaluate the consistency between AI results and the gold standard bone age results. Pearson correlation method was used to measure the reliability between AI results and the gold standard results. The difference of bone age results between using TW3-RUS and TW3-Carpal criteria in different age groups was compared using paired t-test. Results:As for the whole sample, the results based on TW3-RUS criteria were 8.9±3.1 years old for AI assessment and 8.7±2.9 years old for the golden standard reference, with the ICC of 0.983; and the results based on TW3-Carpal criteria were 8.7±3.0 years old for AI and 8.8±2.8 years old for the golden standard reference, with the ICC of 0.976. Positive correlation were found in both TW3-RUS ( r=0.985, P<0.001) and TW3-Carpal criteria groups ( r=0.978, P<0.001). There were significant differences between TW3-RUS and TW3-Carpal at age groups of 7-9( t=-3.36, P=0.001), 10-12( t=-1.77, P=0.046), and 13-15 years old ( t=1.84, P=0.040). The bone age assessment using TW3-RUS and TW3-Carpal criteria were both in good agreement with the gold standard reference in age group of 4-6 years old (ICC=0.929 and 0.940), as well as in age group of 7-9 years old (ICC=0.882 and 0.927, respectively), with the results using TW3-Carpal criteria were slightly higher. As for the age groups of 10-12 and 13-15 years old, the method using TW3-RUS criteria showed excellent agreement with the gold standard reference (ICC=0.962 and 0.963, respectively), which were better than the performance of method using TW3-Carpal criteria (ICC=0.744 and 0.605, respectively). Conclusions:AI-assisted bone age system based TW3-Carpal and TW3-RUS criteria both show good reliability and accuracy in the bone age measurements. The AI method based TW3-Carpal criteria shows better performance in age group of 4-9 years old, while the method based on TW3-RUS criteria may be better for children of age 10-15 years old.
RESUMEN
Background: Bone age assessment (BAA) is an essential tool utilized in outpatient pediatric clinics. Three major BAA methods, Greulich-Pyle (GP), Tanner-Whitehouse 3 (TW3), and China 05 RUS-CHN (RUS-CHN), were applied to comprehensively compare bone age (BA) and chronological age (CA) in a Chinese sample of preschool children. This study was designed to determine the most reliable method. Methods: The BAA sample consisted of 207 females and 183 males aged 3-6 years from the Zhejiang Province in China. The radiographs were estimated according to the GP, TW3, and RUS-CHN methods by two pediatric radiologists. The data was analyzed statistically using boxplots, the Wilcoxon rank test, and Student's t-test to explore the difference (D) between BA and CA. Results: According to the distributions of D, the boxplots showed that the median D of the TW3 method was close to zero for both male and female subjects. The TW3 and RUS-CHN methods overestimated the age of both genders. The TW3 method had the highest correct classification rate for males but a similar rate for females. The GP method did not show any significant difference between the BA and CA when applied to 3-year-old males and 4-year-old females while the TW3 method showed similar results when applied to 6-year-old females. The RUS-CHN method showed the least consistent results among the three methods. Conclusion: The TW3 method was superior to the GP and RUS-CHN methods but not reliable on its own. It should be noted that a precise age diagnosis for preschool children cannot be easily made if only one of the methods is utilized. Therefore, it is advantageous to combine multiple methods when assessing bone age.
RESUMEN
BACKGROUND: Bone age is useful for pediatric endocrinologists in evaluating various disorders related to growth and puberty. Traditional methods of bone age assessment, namely Greulich and Pyle (GP) and Tanner-Whitehouse (TW), have intra- and interobserver variations. Use of computer-automated methods like BoneXpert might overcome these subjective variations. OBJECTIVE: The aim of our study was to assess the validity of BoneXpert in comparison to manual GP and TW methods for assessing bone age in children of Asian Indian ethnicity. MATERIALS AND METHODS: We extracted from a previous study the deidentified left hand radiographs of 920 healthy children aged 2-19 years. We compared bone age as determined by four well-trained manual raters using GP and TW methods with the BoneXpert ratings. We computed accuracy using root mean square error (RMSE) to assess how close the bone age estimated by BoneXpert was to the reference rating. RESULTS: The standard deviations (SDs) of rating among the four manual raters were 0.52 years, 0.52 years and 0.47 years for GP, TW2 and TW3 methods, respectively. The RMSEs between the automated bone age estimates and the true ratings were 0.39 years, 0.41 years and 0.36 years, respectively, for the same methods. The RMSE values were significantly lower in girls than in boys (0.53, 0.5 and 0.47 vs. 0.39, 0.47 and 0.4) by all the methods; however, no such difference was noted in classification by body mass index. The best agreement between BoneXpert and manual rating was obtained by using 50% weight on carpals (GP50). The carpal bone age was retarded in Indian children, more so in boys. CONCLUSION: BoneXpert was accurate and performed well in estimating bone age by both GP and TW methods in healthy Asian Indian children; the error was larger in boys. The GP50 establishes "backward compatibility" with manual rating.
Asunto(s)
Determinación de la Edad por el Esqueleto , Etnicidad , Determinación de la Edad por el Esqueleto/métodos , Niño , Femenino , Mano/diagnóstico por imagen , Humanos , Masculino , RadiografíaRESUMEN
Background: Bone age assessment (BAA) is a crucial research topic in pediatric radiology. Interest in the development of automated methods for BAA is increasing. The current BAA algorithms based on deep learning have displayed the following deficiencies: (I) most methods involve end-to-end prediction, lacking integration with clinically interpretable methods; (II) BAA methods exhibit racial and geographical differences. Methods: A novel, automatic skeletal maturity assessment (SMA) method with clinically interpretable methods was proposed based on a multi-region ensemble of convolutional neural networks (CNNs). This method predicted skeletal maturity scores and thus assessed bone age by utilizing left-hand radiographs and key regional patches of clinical concern. Results: Experiments included 4,861 left-hand radiographs from the database of Beijing Jishuitan Hospital and revealed that the mean absolute error (MAE) was 31.4±0.19 points (skeletal maturity scores) and 0.45±0.13 years (bone age) for the carpal bones-series and 29.9±0.21 points and 0.43±0.17 years, respectively, for the radius, ulna, and short (RUS) bones series based on the Tanner-Whitehouse 3 (TW3) method. Conclusions: The proposed automatic SMA method, which was without racial and geographical influence, is a novel, automatic method for assessing childhood bone development by utilizing skeletal maturity. Furthermore, it provides a comparable performance to endocrinologists, with greater stability and efficiency.
RESUMEN
INTRODUCTION: Bone age (BA) is a quantitative determination of skeletal maturation. The role of puberty in variations in BA is poorly understood as hypothalamic-pituitary-gonadal (HPG) axis maturation and skeletal maturation are regulated in parallel but independently by multiple different factors. In countries like India where there is rapid nutrition transition and increase in prevalence of obesity, their impact on height and BA is not well understood. OBJECTIVES: To study if in 2-17 year old healthy children, the difference between chronological age (CA), height age (HA) and BA is less than 1 year on either side of the chronological age and to assess relationship of BA with height, weight and BMI with special reference to gender and puberty. METHODS: This cross-sectional study included 804 preschool/school-going Indian children. Anthropometric measurements and pubertal assessments were performed using standard protocols and were converted to age and sex standardized z-scores using Indian references while BA was estimated by Tanner-Whitehouse (TW3) method. p<0.05 was considered statistically significant. RESULTS: The mean age and gender standardized z-scores for height, weight, body mass index (BMI) and BA were -0.3 ± 0.7, -0.7 ± 0.8, -0.1 ± 1.0, and -0.2 ± 0.9 respectively. HA was more delayed in girls while BA was more delayed in boys. The mean BA z-score increased with increasing BMI. After the onset of puberty, there was higher increment in BA in girls and HA in boys (p<0.05). CONCLUSIONS: HA, BA and CA were tightly correlated in healthy Indian children with a significant role of nutritional status and puberty in causing variation in the same.
Asunto(s)
Estatura , Pubertad , Adolescente , Índice de Masa Corporal , Niño , Preescolar , Estudios Transversales , Femenino , Humanos , Masculino , ObesidadRESUMEN
Objective: This study aimed to assess the accuracy of previously developed height prediction models in male Japanese football players and create new height prediction models. Materials: The participants were elite academy male football players. We collected current height, parent's height, calendar age and bone age in 6th grade of primary school and obtained actual final height at 20 to 28 years old. Methods: We compared the accuracy of two conventional models for predicting final height. These used current height, calendar age and either bone age (Model 1) or parental height (Model 2). We then developed a new model to optimize the coefficients of Model 1 (Model 3). The final model added parental height to Model 3 and optimized the coefficients (Model 4). Results: Prediction accuracy was higher for Model 2 (R = 0.52, P < 0.001) than Model 1 (p = 0.33, P < 0.001). The equation of Model 3 was final height = 0.63229313×actual measured height-8.2541327×calendar age-2.3009853×bone age (TW2)+206.627184. The R-square was 0.49 (P < 0.0001). The equation of Model 4 was final height = 0.32156081×actual measured height - 4.6652063×calendar age+0.41903909×father's height+0.34952508×mother's height-0.740469×bone age(TW2)+62.1007751. The R-square was 0.61 (P < 0.0001). Conclusions: In the two previous conventional models, a formula using parental height had better predictive accuracy. We developed a new height prediction model using current height, calendar age, father's and mother's height and bone age.
RESUMEN
BACKGROUND: There are several methods of bone age (BA) assessment, which include Gruelich-Pyle (GP), Gilsanz-Ratib (GR), and Tanner Whitehouse-3 (TW-3) methods. Although GP atlas is the most widely used, there are concerns about its accuracy in children of different ethnicities, making the use of the TW-3 method an attractive option in Indian children. OBJECTIVES: 1) To assess the relationship of BA with chronological age (CA) as assessed by different methods (GP, GR, and TW-3) in healthy Indian children 2) To assess which of the three methods of BA assessment is more suitable in Indian children. METHODOLOGY: X-rays of 851 children (438 boys and 413 girls, aged 2-16.5 years) were analyzed by four independent observers using three different methods of BA estimation (GP, GR, and TW-3). Mean BAs were converted to Z-scores. For purpose of deciding which method of BA was most suitable in our cohort, a test of proportions and root mean square (RMS) deviations were computed. RESULTS: Using the test of proportions, the TW-3 method was most suitable overall (P < 0.05). TW-3 method was again most applicable in prepubertal boys (P < 0.05), in prepubertal girls (although not significant, P > 0.1), and pubertal girls (P < 0.05). However, in pubertal boys, the GR atlas method was most suitable (P < 0.05). The same results were obtained when root mean square (RMS) deviations were computed. Interestingly, BA was underestimated in Indian boys irrespective of the method used. In Indian girls, however, the BA was underestimated till the pubertal growth spurt, after which there was rapid advancement of BA. CONCLUSIONS: Among the three methods (GP, GR, and TW-3), the BAs estimated by the TW-3 method were closest to CAs. Hence, it seems reasonable to recommend the use of the TW-3 method for BA estimation in the Indian population till an Indian standard bone age atlas is developed.
RESUMEN
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.
Asunto(s)
Inteligencia Artificial , Aprendizaje Profundo , Adolescente , Determinación de la Edad por el Esqueleto , Niño , Femenino , Humanos , Masculino , Radiografía , Reproducibilidad de los ResultadosRESUMEN
BACKGROUND AND AIM: Determination of skeletal maturity and bone age from cervical vertebrae has been well documented. Most methods described use subjective evaluation of morphological characteristics of cervical vertebrae and may be prone to variability and error. A few objective methods have also been developed, specific for certain populations and genders. The aim of this study was to establish and validate an objective method to determine cervical vertebral bone age from lateral cephalometric radiographs, for Asian South Indian patients of both genders. METHODS: Ninety boys and 90 girls between 9 and 15 years of age were recruited, and their lateral cephalograms were taken. Using measurements made from the third and fourth cervical vertebrae, a formula to determine cervical vertebral bone age was derived using stepwise regression analysis. To validate the formula, a separate sample of 30 boys and 30 girls was chosen, and hand-wrist radiographs and lateral cephalograms were obtained. Cervical vertebral bone age (CVBA) was determined by applying the formula derived. Bone age was also calculated using the Tanner-Whitehouse 3 method. The bone ages determined by both methods were compared to each other and chronological age, using one-way ANOVA, Tukey's post hoc analysis, and Pearson's correlation coefficient. RESULTS: The formulae derived in the current study to determine CVBA differed for both genders. No statistically significant difference was found between CVBA, bone age derived by the Tanner-Whitehouse 3 method, and chronological age for both boys (p value = 0.425) and girls (p value = 0.995). A moderate to strong positive correlation was found between CVBA, bone age, and chronological age. CONCLUSION: The formulae derived in this study were validated and are reliable for objectively determining cervical vertebral bone age and skeletal maturation from lateral cephalograms for Asian South Indian patients of both genders.
Asunto(s)
Determinación de la Edad por el Esqueleto , Vértebras Cervicales , Adolescente , Cefalometría , Vértebras Cervicales/diagnóstico por imagen , Niño , Femenino , Humanos , Masculino , Análisis de Regresión , Reproducibilidad de los ResultadosRESUMEN
BACKGROUND: The precision of adult height prediction by bone age determination in children with idiopathic growth hormone deficiency (IGHD) is unknown. METHODS: The near adult height (NAH) of patients with IGHD in the KIGS database was compared retrospectively to adult height prediction calculated by the Bayley-Pinneau (BP) prediction based on bone age by Greulich-Pyle (GP) in 315 children and based on the Tanner-Whitehouse 2 (TW2) method in 121 children. Multiple linear regression analyses adjusted for age at GH start, age at puberty, mean dose and years of of GH treatment, and maximum GH peak in stimulation test were calculated. RESULTS: The mean underestimation of adult height based on the BP method was at baseline 4.1 ± 0.7 cm in girls and 6.1 ± 0.6 cm in boys, at 1 year of GH treatment 2.5 ± 0.5 cm in girls and 0.9 ± 0.4 cm in boys, while at last bone age determination adult height was overestimated in mean by 0.4 ± 0.6 cm in girls and 3.8 ± 0.5 cm in boys. The mean underestimation of adult height based on the TW2 method was at baseline 5.3 ± 2.0 cm in girls and 7.9 ± 0.8 cm in boys, at 1 year of GH treatment adult height was overestimated in girls 0.1 ± 0.6 cm in girls and underestimated 4.1 ± 0.4 cm in boys, while at last bone age determination adult height was overestimated in mean by 3.1 ± 1.5 cm in girls and 3.6 ± 0.8 cm in boys. CONCLUSIONS: Height prediction by BP and TW2 at onset of GH treatment underestimates adult height in prepubertal IGHD children, while in mean 6 years after onset of GH treatment these prediction methods overestimated adult height.
RESUMEN
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.
RESUMEN
STUDY DESIGN: This is a retrospective clinical study. PURPOSE: In this study, we aim to evaluate the reliability of the distal radius and ulna assessment (DRU) and simplified Tanner-Whitehouse III classification (sTW3) in Japanese patients with adolescent idiopathic scoliosis (AIS). OVERVIEW OF LITERATURE: The greatest curvature of a scoliotic spine occurs at peak-height velocity (PHV), which is the time during which an individual's height increases at the maximum rate. Diagnosing and appropriately treating AIS before PHV is the most effective way in order to prevent unnecessary deterioration of the scoliosis curve. Although it is difficult to detect scoliosis before PHV, DRU and sTW3, which involve evaluations using a left-hand radiograph, have been reported to be effective. METHODS: We retrospectively evaluated 54 hands of 40 girls with AIS who visited Nara Medical University Hospital from 2000 to 2015 using previously collected radiographs. The examiners included a spine surgeon and a pediatric orthopedic surgeon, each with over 10 years of experience. The reliability of the DRU and sTW3 was evaluated using the kappa coefficient. RESULTS: The left-hand radiographs of 40 female patients with AIS (mean age, 13.9±1.7 years; N=54 hands) were evaluated by two blinded examiners using the sTW3 and DRU methods. The highest inter-observer and intra-observer reliabilities (kappa, 0.64 and 0.62, respectively) for radius evaluation were determined. Radius evaluation by the DRU showed the highest agreement rate and smallest error between the inter- and intra-observer examinations. CONCLUSIONS: The DRU was the most reliable assessment tool, and it has the potential to be useful for precisely determining the stage of skeletal maturity in outpatient clinics.
RESUMEN
PURPOSE: The EOS-imaging system is increasingly adopted for clinical follow-up in scoliosis with the advantages of simultaneous biplanar imaging of the spine in an erect position. Skeletal maturity assessment using a hand radiograph is an essential adjunct to spinal radiography in scoliosis follow-up. This study aims at testing the feasibility and validity of a newly proposed EOS workflow with sequential spine-hand radiography for skeletal maturity assessment and bracing recommendation. METHODS: EOS spine-hand radiographs from patients with diagnosis of idiopathic scoliosis, including both sexes and an age range of ten to 14 years, were scored using the Thumb Ossification Composite Index (TOCI), Sanders and Risser methods. Intraclass correlation coefficients (ICCs) were calculated for inter/intraobserver agreement and were tested with Cronbach's alpha values. RESULTS: In all, 60 EOS-spine hand radiographs selected from subjects with diagnosis of adolescent idiopathic scoliosis (AIS), including 32 male patients (mean age 11.53 years; 10 to 14) and 28 female patients (mean age 11.50 years; 10 to 13) who underwent sequential spine-hand low dose EOS imaging were generated for analysis. The overall interobserver (ICC = 0.997) and intraobserver agreement (α > 0.9) demonstrated excellent agreement for TOCI staging; ICC > 0.994 for both TOCI and Sanders staging comparing traditional digital versus EOS hand radiography; ICC ≥ 0.841 for agreement on bracing recommendation among TOCI versus the Risser and Sanders system. CONCLUSION: With the proposed new EOS workflow it was feasible to produce high image quality for skeletal maturity assessment with excellent reliability and validity to inform consistent bracing recommendation in AIS. The workflow is applicable for busy daily clinic settings in tertiary scoliosis centres with reduced time cost, improved efficiency and throughput of the radiology department. LEVEL OF EVIDENCE: III.
RESUMEN
Bone age assessment plays an important role in the endocrinology and genetic investigation of patients. In this paper, we proposed a deep learning-based approach for bone age assessment by integration of the Tanner-Whitehouse (TW3) methods and deep convolution networks based on extracted regions of interest (ROI)-detection and classification using Faster-RCNN and Inception-v4 networks, respectively. The proposed method allows exploration of expert knowledge from TW3 and features engineering from deep convolution networks to enhance the accuracy of bone age assessment. The experimental results showed a mean absolute error of about 0.59 years between expert radiologists and the proposed method, which is the best performance among state-of-the-art methods.