A hierarchical model for automated standard sagittal-view detection from 3D ultrasound data in 11–14 weeks

• A hierarchical model is proposed for automated standard sagittal-view detection from 3D ultrasound data.
• Hessian-matrix based filtering is applied for obtaining the plate-structure distribution.
• The sphere distribution is calculated by convolving sphere detectors with the ultrasound data.
• The sampling-based Hough transform is performed for the plane detection.

The nuchal translucency thickness is an important parameter for the diagnosis of fetal abnormalities during 11–14 weeks. Currently in clinical practice, it first requires manual scanning operations to determine the fetal standard sagittal-view plane and then the measurements can be performed in the corresponding plane images. Besides the difficulty of such standard plane detection, this also leads to the time-consuming and detection-variability problems. In the paper, a hierarchical model is proposed to automatically detect the standard sagittal-view plane based on 3D ultrasound data. In the model, Hessian-matrix based filtering is first applied for obtaining the plate-structure distribution in the data. Then the sphere distribution is calculated by convolving sphere detectors with the ultrasound data. Based on the two prior distributions, the sampling-based Hough transform is further performed for the plane detection. The performance of the proposed model is verified by the experimental results on 3D synthetic data and 241 clinical 3D ultrasound data in 11–14 weeks.