Using deep-learning algorithms to classify fetal brain ultrasound images as normal or abnormal.

OBJECTIVES: To evaluate the feasibility of using deep-learning algorithms to classify as normal or abnormal sonographic images of the fetal brain obtained in standard axial planes. METHODS: We included in the study images retrieved from a large hospital database from 10 251 normal and 2529 abnormal pregnancies. Abnormal cases were confirmed by neonatal ultrasound, follow-up examination or autopsy. After a series of pretraining data processing steps, 15 372 normal and 14 047 abnormal fetal brain images in standard axial planes were obtained. These were divided into training and test datasets (at case level rather than image level), at a ratio of approximately 8:2. The training data were used to train the algorithms for three purposes: performance of image segmentation along the fetal skull, classification of the image as normal or abnormal and localization of the lesion. The accuracy was then tested on the test datasets, with performance of segmentation being assessed using precision, recall and Dice's coefficient (DICE), calculated to measure the extent of overlap between human-labeled and machine-segmented regions. We assessed classification accuracy by calculating the sensitivity and specificity for abnormal images. Additionally, for 2491 abnormal images, we determined how well each lesion had been localized by overlaying heat maps created by an algorithm on the segmented ultrasound images; an expert judged these in terms of how satisfactory was the lesion localization by the algorithm, classifying this as having been done precisely, closely or irrelevantly. RESULTS: Segmentation precision, recall and DICE were 97.9%, 90.9% and 94.1%, respectively. For classification, the overall accuracy was 96.3%. The sensitivity and specificity for identification of abnormal images were 96.9% and 95.9%, respectively, and the area under the receiver-operating-characteristics curve was 0.989 (95% CI, 0.986-0.991). The algorithms located lesions precisely in 61.6% (1535/2491) of the abnormal images, closely in 24.6% (614/2491) and irrelevantly in 13.7% (342/2491). CONCLUSIONS: Deep-learning algorithms can be trained for segmentation and classification of normal and abnormal fetal brain ultrasound images in standard axial planes and can provide heat maps for lesion localization. This study lays the foundation for further research on the differential diagnosis of fetal intracranial abnormalities. Copyright © 2020 ISUOG. Published by John Wiley & Sons Ltd.

[Analysis of copy number variation by CMA in fetus with increased nuchal translucency].

Objective: To investigated the clinical value of chromosomal microarray analysis (CMA) in fetuses with increased nuchal translucency (NT) . Methods: Totally 101 cases out of 19 261 singleton fetuses who underwent the first trimester (11-13+6 weeks) ultrasound examination from January 2015 to June 2017 at First Affiliated Hospital of Sun Yat-sen University were diagnosed with NT ≥2.5 mm and underwent invasive prenatal test for fetal karyotype and CMA. According to the combination of other ultrasound abnormalities, the cases were divided into isolated group (67.3%, 68/101) and complicated group (32.7%, 33/101) . In addition, the cases were divided into 5 groups according to the thickness of NT, 2.5-2.9 mm (borderline thickening; 16.8%, 17/101) , 3.0-3.4 mm (33.7%, 34/101) , 3.5-4.4 mm (16.8%, 17/101) , 4.5-5.4 mm (15.8%, 16/101) , and ≥5.5 mm (16.8%, 17/101) . Chi square test was used to detect the different rates of other combined ultrasound abnormalities and abnormal chromosome between 5 groups. Results: The median thickness of NT was 3.4 mm (2.5-8.5 mm) . And 32 cases (31.7%, 32/101) had abnormal karyotype. There was a significant difference in the frequency of abnormal karyotype between the isolated and the complicated group (20.6% vs 54.5%, P<0.01) . Among 69 cases (68.3%, 69/101) of normal karyotype, 3 cases (4.3%, 3/69) were detected with pathogenic copy number variation (CNV) by CMA. Thirty-five cases with chromosomal abnormalities (include abnormal karyotype and pathogenic CNV) , there was a significant difference in the frequency of chromosomal abnormalities between the isolated and the complicated group (23.5% vs 57.6%, P=0.001) . The median age of pregnant women in 5 groups was 35 years (24-39 years) , 33 years (23-46 years) , 31 years (21-46 years) , 33 years (21-41 years) and 35 years (21-43 years) . The rates of chromosomal abnormalities increased with the increase of NT thickness. There was significant difference in the incidence of associated chromosomal abnormalities among 5 groups (P<0.05) . Comparative analysis within the 5 groups, the incidence of associated chromosomal abnormalities between NT 2.5-2.9 mm and ≥5.5 mm was significantly different (P=0.005) , while the differences between the other groups were not significant (P>0.05) . Conclusions: There is a high risk of fetal chromosomal abnormalities in borderline NT thickening (2.5-2.9 mm) at advanced maternal age, but the pathogenic CNV is not detected. Chromosomal microdeletion or microduplication could be further detected in the NT thickening (≥3.0 mm) fetuses with normal karyotype by chromosome microarray analysis, while the positive rate is relatively low, and the variants of unknown significance might be detected.

Prenatal diagnosis of total and partial anomalous pulmonary venous connection: multicenter cohort study and meta-analysis.

OBJECTIVES: The aims of this study were to review systematically literature on and describe the sonographic features and associated anomalies of total (TAPVC) and partial (PAPVC) anomalous pulmonary venous connection and scimitar syndrome (SS). METHODS: A retrospective cohort study was carried out of cases of TAPVC, PAPVC and SS that underwent comprehensive ultrasound examination, seen over a 20-year period at two tertiary referral centers. Assessed variables included TAPVC subtype, gestational age at diagnosis, area behind the left atrium, ventricular disproportion, vertical vein, pulmonary venous obstruction, mode of diagnosis, association with cardiac and extracardiac conditions, and pregnancy and fetoneonatal outcomes. The outcome was considered favorable if the individual was alive and well (no functional impairment from surgery or cardiac or extracardiac conditions). Cases associated with right isomerism were excluded from the analysis, as TAPVC in these cases was only one of several major cardiac anomalies affecting sonographic signs. A systematic review was performed in order to obtain a synthesis of characteristics associated with TAPVC, PAPVC and SS. The literature search of PubMed and EMBASE (1970-2016) included reviews, case series and case reports. A meta-analysis was conducted only for TAPVC. Random-effects models were used to obtain pooled estimates of the frequencies of clinical characteristics and sonographic features. RESULTS: For TAPVC, a total of 15 studies involving 71 patients (including 13 from the current cohort study) were included in the systematic review and meta-analysis. The pooled estimate for the association of TAPVC with congenital heart disease was 28.3% (95% CI, 18.1-41.3%) and with extracardiac anomalies it was 18.5% (95% CI, 10.5-30.6%). Of TAPVC cases, obstructed venous return was observed in 34.1% (95% CI, 22.7-47.7%), a favorable outcome in 43.8% (95% CI, 24.0-65.8%), ventricular disproportion in 59.2% (95% CI, 45.1-72.0%), increased area behind the left atrium in 58.1% (95% CI, 41.1-73.5%) and a vertical vein in 59.3% (95% CI, 41.1-75.3%). Diagnosis was established by using color or power Doppler in 84.9% (95% CI, 67.3-93.9%) of cases. For SS, there were only three studies describing eight cases, to which the current study added another five. Ventricular disproportion was present in three out of nine SS cases for which data were available, but for two of these, there was a concurrent heart anomaly. Color Doppler was used for all SS diagnoses, and four-dimensional echocardiography was useful in two out of six cases in which it was used. Outcome for SS cases was generally good. For PAPVC, there were only five studies describing five cases, to which the current study added another two. Major cardiac anomalies were associated in four out of seven of these cases, and extracardiac anomalies in three out of six cases for which data were available. CONCLUSIONS: TAPVC can be associated with other cardiac and extracardiac anomalies in a significant percentage of cases. Leading sonographic signs are ventricular disproportion, increased area behind the left atrium and the finding of a vertical vein. Color/power Doppler is the key mode for diagnosis of TAPVC. Obstructed venous return can be expected in roughly one-third of cases of TAPVC and outcome is favorable in less than half of cases. Data for SS and PAPVC are too few to synthesize. Copyright © 2017 ISUOG. Published by John Wiley & Sons Ltd.

Congenital hydrocephalus and hemivertebrae associated with de novo partial monosomy 6q (6q25.3→qter).

This study was conducted to describe a prenatal case of congenital hydrocephalus and hemivertebrae with a 6q terminal deletion and to investigate the possible correlation between the genotype and phenotype of the proband. We performed an array-based comparative genomic hybridization (aCGH) analysis on a fetus diagnosed with congenital hydrocephalus and hemivertebrae. The deletion, spanning 10.06 Mb from 6q25.3 to 6qter, was detected in this fetus. The results of aCGH, karyotype and fluorescent in situ hybridization (FISH) analyses in the healthy parents were normal, which confirmed that the proband's copy-number variant (CNV) was de novo. This deleted region encompassed 97 genes, including 28 OMIM genes. We discussed four genes (TBP, PSMB1, QKI and Pacrg) that may be responsible for hydrocephalus while the T gene may have a role in hemivertebra. We speculate that five genes in the 6q terminal deletion region were potentially associated with hemivertebrae and hydrocephalus in the proband.

Expression and significance of plasma 3-NT and ox-LDL in patients with Alzheimer's disease.

To examine the expression and clinical significance of plasma 3-nitrotyrosine (3-NT) and oxidized low-density lipoprotein (ox-LDL) levels in patients with Alzheimer's disease (AD), we examined 48 AD patients and 37 healthy control subjects. The Mini-Mental State Examination, Activities of Daily Living Scale, and Hachinski Ischemic Scale were examined in all subjects. AD patients were classified using the Global Deterioration Scale. The concentrations of plasma 3-NT and ox-LDL were detected using an enzyme-linked immunosorbent assay. We found that the plasma 3-NT concentration in the AD group (119.46 ± 21.82 nM) was significantly higher than that in the control group (55.09 ± 9.63 nM) (P < 0.05). Spearman analysis showed that plasma 3-NT level was negatively associated with the Mini-Mental State Examination results of AD patients. Plasma ox-LDL level in the AD group (112.25 ± 17.81 mg/L) was significantly higher than that in the control group (47.46 ± 10.04 mg/L) (P < 0.05). Spearman analysis showed that plasma ox-LDL level was positively correlated with AD severity in AD patients. However, plasma 3-NT level in the AD group was not associated with plasma ox-LDL level. Therefore, plasma 3-NT and ox-LDL levels in AD patients were significantly increased, which may be related to the degree of AD severity in AD patients.

Cardiovascular Z-scores in fetuses with tetralogy of Fallot.

OBJECTIVE: To establish formulae for the calculation of fetal cardiovascular Z-scores based on femur length (FL), and to compare cardiovascular parameters between fetuses with tetralogy of Fallot (TOF) and normal fetuses in order to assess their value for the prenatal diagnosis of TOF. METHODS: A total of 329 normal fetuses and 43 fetuses with TOF were included in this study. Eleven cardiovascular dimensions were measured offline after cardiac spatiotemporal image correlation volume acquisition. Normal cardiovascular Z-score formulae were constructed for these measurements based on FL by performing a standard regression analysis followed by weighted regression of absolute residual values. The following ratios were calculated: right ventricular internal diameter (RVID) to left ventricular internal diameter (LVID) and pulmonary artery diameter (PA) to aorta diameter (Ao). Subsequently, all parameters were compared between the normal and TOF groups. RESULTS: Formulae for calculating Z-scores for the 11 cardiovascular dimensions were constructed. FL was significantly correlated with all cardiovascular dimensions assessed (r = 0.816-0.944, P < 0.001 for all). RVID, Ao, PA, aortic arch isthmus, and ductus arteriosus mean Z-scores and PA/Ao ratio were statistically significantly different between normal fetuses and those with TOF. In the TOF group, all Ao Z-scores (43/43) were > + 2 and all PA/Ao ratios (43/43) were below the normal 95% range. However, only 48.8% (21/43) of the PA Z-scores were < -2. CONCLUSIONS: The cardiovascular Z-score formulae developed can provide a quantitative basis for the prenatal diagnosis of TOF. Aortic dilatation and abnormal PA/Ao ratio may be markers for the antenatal diagnosis of TOF.