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

BACKGROUND: Skeletal dysplasias collectively affect a large number of patients worldwide. Most of these disorders cause growth anomalies. Hence, evaluating skeletal maturity via the determination of bone age (BA) is a useful tool. Moreover, consecutive BA measurements are crucial for monitoring the growth of patients with such disorders, especially for timing hormonal treatment or orthopedic interventions. However, manual BA assessment is time-consuming and suffers from high intra- and inter-rater variability. This is further exacerbated by genetic disorders causing severe skeletal malformations. While numerous approaches to automate BA assessment have been proposed, few are validated for BA assessment on children with skeletal dysplasias. OBJECTIVE: We present Deeplasia, an open-source prior-free deep-learning approach designed for BA assessment specifically validated on patients with skeletal dysplasias. MATERIALS AND METHODS: We trained multiple convolutional neural network models under various conditions and selected three to build a precise model ensemble. We utilized the public BA dataset from the Radiological Society of North America (RSNA) consisting of training, validation, and test subsets containing 12,611, 1,425, and 200 hand and wrist radiographs, respectively. For testing the performance of our model ensemble on dysplastic hands, we retrospectively collected 568 radiographs from 189 patients with molecularly confirmed diagnoses of seven different genetic bone disorders including achondroplasia and hypochondroplasia. A subset of the dysplastic cohort (149 images) was used to estimate the test-retest precision of our model ensemble on longitudinal data. RESULTS: The mean absolute difference of Deeplasia for the RSNA test set (based on the average of six different reference ratings) and dysplastic set (based on the average of two different reference ratings) were 3.87 and 5.84 months, respectively. The test-retest precision of Deeplasia on longitudinal data (2.74 months) is estimated to be similar to a human expert. CONCLUSION: We demonstrated that Deeplasia is competent in assessing the age and monitoring the development of both normal and dysplastic bones.

[Neighbourhood deprivation and community nutrition environment: associations between the social and built environment and food availability]. / Stadtteildeprivation und Lebensmittelumwelt: Zusammenhänge von sozialen und baulichen Merkmalen der Wohnumgebung und Lebensmittelverfügbarkeit.

OBJECTIVE: Eating habits are influenced by individual socioeconomic status (SES). As the association between neighbourhood characteristics and food availability is still unclear, we investigated the community nutrition environment in different neighbourhoods. METHODS: Using official data of the city of Leipzig, we selected three neighbourhoods which differed in terms of their social (2 deprived vs. 1 affluent) and built (1 large housing estates vs. 2 historic buildings) environment. Data on food outlets were assessed via direct observation (ground truthing). RESULTS: Social and residential environment characteristics are associated with food availability. The proportion of fast food outlets is higher in deprived neighbourhoods compared to the affluent one (25%, 30.4% vs. 13.5%). Neighbourhoods with historic buildings offer a greater variety of food outlets. CONCLUSION: The greater availability of unhealthy food may contribute to the development of obesity. Preventive measures should consider the physical and social environment.

Prevention of childhood obesity: recruiting strategies via local paediatricians and study protocol for a telephone-based counselling programme.

AIM: Prevention research has to elucidate how families with overweight children can be convinced to participate in obesity intervention trials. Here we describe the detailed recruitment process for a telephone-based obesity prevention programme for families with overweight children and present participation rates, the study design and a socio-demographic description of participating families. METHODS: Overweight (BMI > 90th percentile) children and adolescents 4-17 years of age were screened via the German paediatrician network CrescNet. The prevention programme (multiple computer aided telephone counselling interviews) was suggested to eligible families via local paediatricians. Participating families were compared anthropometrically and socio-demographically to the entire screening population and to micro-census data, a representative national survey. RESULTS: The screening process assessed 4005 candidates for eligibility. Paediatricians reported having suggested programme participation to 3387 candidates (referred to as 100%). 427 candidates (12.6%) returned a written consent for programme participation. 303 candidates (9.0%) started the intervention. The study population (n = 303) included more obese (45.6% vs. 33.2%; p < 0.001) and fewer overweight participants (40.4% vs. 55.2%; p < 0.001) than the entire screening population. Compared to the micro-census, families with adolescents (8.2% vs. 16.9%; p < 0.001) and single parents (16.0% vs. 23.3%; p < 0.005) were underrepresented in the study population, whereas families living in rural areas were overrepresented (58.7% vs. 50.2%; p < 0.005). CONCLUSIONS: Although 84.6% of the paediatricians forwarded the suggestion for programme participation to eligible families, participation rates for this low-threshold; telephone-based obesity intervention remained very low. "Hard-to-engage groups" for effective obesity prevention seem to include single parents, as well as parents of adolescents.