Multi-Parametric Fusion of 3D Power Doppler Ultrasound for Fetal Kidney Segmentation Using Fully Convolutional Neural Networks
- 29 September 2020
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Biomedical and Health Informatics
- Vol. 25 (6) , 2050-2057
Abstract
Kidney development is key to the long-term health of the fetus. Renal volume and vascularity assessed by 3D ultrasound (3D-US) are known markers of wellbeing, however, a lack of real-time image segmentation solutions preclude these measures being used in a busy clinical environment. In this work, we aimed to automate kidney segmentation using fully convolutional neural networks (fCNN). We used multi-parametric input fusion incorporating 3D B-Mode and power Doppler (PD) volumes, aiming to improve segmentation accuracy. Three different fusion strategies and their performance were assessed versus a single input (B-Mode) network. Early input-level fusion provided the best segmentation accuracy with an average Dice similarity coefficient (DSC) of 0.81 and Hausdorff distance (HD) of 8.96 mm, an improvement of 0.06 DSC and reduction of 1.43mm HD compared to our baseline network. Compared to manual segmentation for all models, repeatability was assessed by intra-class correlation coefficients (ICC) indicating good to excellent reproducability (ICC >= 0.93). The framework was extended to support multiple graphics processing units (GPUs) to better handle volumetric data, dense fCNN models, batch normalisation and complex fusion networks. This work and available source code provides a framework to increase the parameter space of encoder-decoder style fCNNs across multiple GPUs and shows that application of multi-parametric 3D-US in fCNN training improves segmentation accuracy.Keywords
Funding Information
- Leslie Stevens Fund for Newborn Care, Sydney, Australia
This publication has 25 references indexed in Scilit:
- A Guideline of Selecting and Reporting Intraclass Correlation Coefficients for Reliability ResearchJournal of Chiropractic Medicine, 2016
- Renal Segmentation From 3D Ultrasound via Fuzzy Appearance Models and Patient-Specific Alpha ShapesIEEE Transactions on Medical Imaging, 2016
- Birth weight, malnutrition and kidney-associated outcomes—a global concernNature Reviews Nephrology, 2015
- Ultrasound Shear Wave Elastography Helps Discriminate Low‐grade From High‐grade Bowel Wall Fibrosis in Ex Vivo Human Intestinal SpecimensJournal of Ultrasound in Medicine, 2014
- The Design of SimpleITKFrontiers in Neuroinformatics, 2013
- Quantitative analysis of renal vascularization in fetuses with urinary tract obstruction by three-dimensional power-DopplerAmerican Journal of Obstetrics and Gynecology, 2011
- User-guided 3D active contour segmentation of anatomical structures: Significantly improved efficiency and reliabilityNeuroImage, 2006
- Segmentation of kidney from ultrasound images based on texture and shape priorsIEEE Transactions on Medical Imaging, 2005
- Validation of Image Segmentation and Expert Quality with an Expectation-Maximization Algorithm2002
- The effect of intrauterine growth retardation on the development of renal nephronsBJOG: An International Journal of Obstetrics & Gynaecology, 1992