Repetitive MR measurements of lung volume in fetuses with congenital diaphragmatic hernia: individual development of pulmonary hypoplasia during pregnancy and calculation of weekly lung growth rates.

OBJECTIVE: To investigate individual changes in fetal lung volume (FLV) in fetuses with isolated congenital diaphragmatic hernia (CDH) and to calculate weekly growth rates of the FLV using serial MR examinations during pregnancy. METHODS: MR-FLV was measured in 89 fetuses with CDH. All fetuses received two MRIs. A mean weekly growth rate of the FLV was determined for each fetus and compared with the growth rate of healthy fetuses. RESULTS: Mean observed-to-expected MR-FLV (o/e MR-FLV) measured at the first MRI was 33.3 ± 12.2% and 29.5 ± 10.9% at the second MRI. In 61% of all fetuses (54/89) the o/e MR-FLV decreased during pregnancy, 26% (23/89) showed an increase in the o/e MR-FLV and 13 % (12/89) had stable values. First and last o/e MR-FLV values were significantly associated with mortality and neonatal extracorporeal membrane oxygenation (ECMO) requirement with a higher prognostic accuracy of MR-FLV measurements near delivery. Patients with CDH had lower weekly lung growth rates than healthy fetuses. There was a significant difference in the mean weekly growth rate between survivors and non-survivors and patients with and without ECMO requirement. CONCLUSION: Individual development of FLV in patients with CDH during pregnancy is extremely variable. Follow-up MR-FLV measurements are advisable before deciding upon pre- and postnatal therapeutic options. KEY POINTS: • Lung development in congenital diaphragmatic hernia (CDH) during pregnancy is extremely variable. • MRI demonstrates that lung growth rate is reduced in fetuses with CDH. • The final observed-to-expected fetal lung volume provides the best prognostic information. • Follow-up measurements are advisable before deciding upon therapeutic options.

MRI-based ratio of fetal lung volume to fetal body volume as a new prognostic marker in congenital diaphragmatic hernia.

OBJECTIVE: The objective of our study was to evaluate the prognostic accuracy of the MRI-based ratio of fetal lung volume (FLV) to fetal body volume (FBV) for predicting survival and the need for extracorporeal membrane oxygenation (ECMO) therapy in fetuses with congenital diaphragmatic hernia (CDH). MATERIALS AND METHODS: MRI was performed in 96 fetuses (34 females and 62 males; mean gestational age, 33 ± 5 [SD] weeks; range, 23-39 weeks) with CDH and 44 healthy control subjects. FLV and FBV were measured for each fetus by one of three investigators, and we calculated the FLV/FBV ratio. Additionally, a subgroup analysis of growth-restricted fetuses was performed. Logistic regression analysis was used to model the dependence of survival and ECMO requirement on the FLV/FBV ratio. Prognostic accuracy was evaluated by applying the area under the curve (AUC). RESULTS: Seventy-eight of the 96 (81%) fetuses survived and 35 (36%) neonates needed ECMO therapy. The FLV/FBV ratio was significantly higher in fetuses who survived (p < 0.0001; AUC = 0.829). Neonates who received ECMO therapy showed a significantly lower prenatal FLV/FBV ratio (p = 0.0001; AUC = 0.811) than neonates who did not need ECMO therapy. A subgroup analysis (n = 13) showed a trend toward a higher FLV/FBV ratio of survivors in comparison with nonsurvivors (p = 0.065; AUC = 0.875). CONCLUSION: In our study, the MRI-based FLV/FBV ratio was able to predict neonatal survival and ECMO requirement in children with CDH with high accuracy. Unlike other prognostic parameters, FLV/FBV ratio is independent of a reference to a control group and may enhance prognostic accuracy particularly in growth-restricted neonatal patients.

Periodic MRI lung volume assessment in fetuses with congenital diaphragmatic hernia: prediction of survival, need for ECMO, and development of chronic lung disease.

OBJECTIVE: The purpose of the study was to investigate the ability to predict survival, need for extracorporeal membrane oxygenation (ECMO), and incidence of chronic lung disease in patients with congenital diaphragmatic hernia in the context of a classification into three different times of gestation (< 28, 28-32, and > 32 weeks) by assessing the ratio between observed and expected MRI fetal lung volume. MATERIALS AND METHODS: The data analysis included 226 fetuses with congenital diaphragmatic hernia. MRI was performed at different times of gestation with a T2-weighted HASTE sequence. Receiver operating characteristic curve analysis was performed to investigate the prognostic value of assessment of the ratio between observed and expected MRI fetal lung volumes at different stages of fetal growth. RESULTS: For all reviewed times of gestation, the ratio between observed and expected MRI fetal lung volumes had almost equivalent statistically significant differences for neonatal survival (p ≤ 0.0029), need for ECMO therapy (p ≤ 0.0195), and development of chronic lung disease (p ≤ 0.0064). Results with high prognostic accuracy for early and for medium and late times in gestation also were obtained. Receiver operating characteristic analysis showed the highest area under the curve (≥ 0.819) for neonatal survival. CONCLUSION: In patients with congenital diaphragmatic hernia, the relation between observed and expected MRI fetal lung volume is a valuable prognostic parameter for predicting neonatal mortality, morbidity represented by the development of chronic lung disease, and the need for ECMO therapy in early gestation (< 28 weeks) as well as later gestation with no statistically significant differences.

Cloud-Based Evaluation of Anatomical Structure Segmentation and Landmark Detection Algorithms: VISCERAL Anatomy Benchmarks.

Variations in the shape and appearance of anatomical structures in medical images are often relevant radiological signs of disease. Automatic tools can help automate parts of this manual process. A cloud-based evaluation framework is presented in this paper including results of benchmarking current state-of-the-art medical imaging algorithms for anatomical structure segmentation and landmark detection: the VISCERAL Anatomy benchmarks. The algorithms are implemented in virtual machines in the cloud where participants can only access the training data and can be run privately by the benchmark administrators to objectively compare their performance in an unseen common test set. Overall, 120 computed tomography and magnetic resonance patient volumes were manually annotated to create a standard Gold Corpus containing a total of 1295 structures and 1760 landmarks. Ten participants contributed with automatic algorithms for the organ segmentation task, and three for the landmark localization task. Different algorithms obtained the best scores in the four available imaging modalities and for subsets of anatomical structures. The annotation framework, resulting data set, evaluation setup, results and performance analysis from the three VISCERAL Anatomy benchmarks are presented in this article. Both the VISCERAL data set and Silver Corpus generated with the fusion of the participant algorithms on a larger set of non-manually-annotated medical images are available to the research community.