OBJECTIVE: To retrospectively evaluate the reliability of offline manipulation of archived three-dimensional (3D) ultrasound volumes in the assessment of the normal fetal brain and the diagnosis of fetal brain abnormalities. METHODS: Seventy-nine 3D volume datasets, archived at the time of scanning, from women attending a tertiary center, were analyzed. They comprised 52 cases with brain abnormalities and 27 normal controls. Postnatal magnetic resonance imaging or postmortem examination confirmed the final diagnosis in all cases with brain anomalies. Offline analysis of the 79 anonymized 3D volume datasets was carried out by three fetal medicine experts, examiner one (E1), examiner two (E2) and examiner three (E3), using 4D View software. The experts were blinded to any prior diagnosis or history. Data were collected on a specially designed data sheet and entered into a specialized database for analysis. Results were compared between examiners, with the initial two-dimensional (2D) diagnosis and with the final definitive diagnosis by calculation of percentage agreement and kappa coefficients. RESULTS: Of the 52 cases with brain anomalies, the correct diagnosis was reached in 88.5% (46/52), 98.1% (51/52) and 92.3% (48/52) of cases on analysis of the 3D volumes by E1, E2 and E3, respectively, whereas only 82.7% (43/52) of cases were correctly diagnosed on the initial 2D examination when compared with the definitive diagnosis. Fetuses without brain anomalies were identified as such by the three experts with 100% agreement. There was good agreement between the initial 2D examination and the analysis of 3D volumes by each of the experts: 86.1% for E1 (κ = 0.7), 89.9% for E2 (κ = 0.79) and 88.6% for E3 (κ = 0.76). CONCLUSIONS: 3D volume datasets are an appropriate and reliable format for storing data from examination of the fetal brain. Offline analysis of 3D datasets is a reliable method that can be used to help in the assessment of brain anomalies and could be a useful adjunct to realtime 2D ultrasonography.
OBJECTIVE: To retrospectively evaluate the reliability of offline manipulation of archived three-dimensional (3D) ultrasound volumes in the assessment of the normal fetal brain and the diagnosis of fetal brain abnormalities. METHODS: Seventy-nine 3D volume datasets, archived at the time of scanning, from women attending a tertiary center, were analyzed. They comprised 52 cases with brain abnormalities and 27 normal controls. Postnatal magnetic resonance imaging or postmortem examination confirmed the final diagnosis in all cases with brain anomalies. Offline analysis of the 79 anonymized 3D volume datasets was carried out by three fetal medicine experts, examiner one (E1), examiner two (E2) and examiner three (E3), using 4D View software. The experts were blinded to any prior diagnosis or history. Data were collected on a specially designed data sheet and entered into a specialized database for analysis. Results were compared between examiners, with the initial two-dimensional (2D) diagnosis and with the final definitive diagnosis by calculation of percentage agreement and kappa coefficients. RESULTS: Of the 52 cases with brain anomalies, the correct diagnosis was reached in 88.5% (46/52), 98.1% (51/52) and 92.3% (48/52) of cases on analysis of the 3D volumes by E1, E2 and E3, respectively, whereas only 82.7% (43/52) of cases were correctly diagnosed on the initial 2D examination when compared with the definitive diagnosis. Fetuses without brain anomalies were identified as such by the three experts with 100% agreement. There was good agreement between the initial 2D examination and the analysis of 3D volumes by each of the experts: 86.1% for E1 (κ = 0.7), 89.9% for E2 (κ = 0.79) and 88.6% for E3 (κ = 0.76). CONCLUSIONS: 3D volume datasets are an appropriate and reliable format for storing data from examination of the fetal brain. Offline analysis of 3D datasets is a reliable method that can be used to help in the assessment of brain anomalies and could be a useful adjunct to realtime 2D ultrasonography.
Authors: Maria E W A Albers; Erato T I A Buisman; René S Kahn; Arie Franx; N Charlotte Onland-Moret; Roel de Heus Journal: Hum Brain Mapp Date: 2018-04-10 Impact factor: 5.038