Related Experiment Video
Updated: Dec 27, 2025
10:25
Technical Approach for Infrared Tracking for Soft Tissue Navigation with a Holographic Head-Mounted Display and Preclinical Validation
Published on: September 2, 2025
420
Probabilistic Learning Coherent Point Drift for 3D Ultrasound Fetal Head Registration.
Jorge Perez-Gonzalez1,2, Fernando Arámbula Cosío1, Joel C Huegel2,3
1Instituto de Investigaciones en Matemáticas Aplicadas y en Sistemas, Universidad Nacional Autónoma de México, Mérida, Yucatán, Mexico.
Computational and Mathematical Methods in Medicine
|February 25, 2020
Summary
This study introduces a new method for automatically registering 3D fetal brain ultrasound volumes, even with significant image artifacts. The approach improves accuracy in fetal growth assessment and brain structure analysis.
Area of Science:
- Medical Imaging
- Biomedical Engineering
- Computational Anatomy
Background:
- Accurate fetal brain growth quantification is vital for assessing fetal well-being using ultrasound (US).
- Fetal US images often suffer from artifacts like acoustic occlusion and calcification, complicating analysis, especially after 18 weeks gestation.
- Fetal US volume registration aids in monitoring fetometry, segmenting brain structures, and aligning multiple acquisitions.
Purpose of the Study:
- To present a novel approach for automatic registration of 3D fetal brain ultrasound volumes.
- To address challenges posed by occlusion artifacts, noise, and missing data in fetal US brain imaging.
- To enable accurate 3D registration without requiring an initial point cloud.
Main Methods:
- A novel variant of the coherent point drift method is proposed.
- Supervised learning is employed for automatic point cloud segmentation and weight factor estimation.
- Random forest classification is used to assign non-uniform Gaussian mixture model membership probabilities.
Main Results:
- The proposed method achieves automatic registration of 3D US fetal brain volumes with occlusions and multiplicative noise.
- Error reduction ranges from 7.4% to 60.7% compared to existing algorithms.
- A target registration error of 6.38 ± 3.24 mm was achieved.
Conclusions:
- The developed approach is highly suitable for automatic 3D registration of fetal head US volumes.
- This technique can enhance fetal growth monitoring, brain structure segmentation, and the compounding of multiple ultrasound acquisitions.
