Related Experiment Video
Updated: Jul 9, 2026
06:56
Human Fetal Blood Flow Quantification with Magnetic Resonance Imaging and Motion Compensation
Published on: January 7, 2021
Automatic fetal measurements in ultrasound using constrained probabilistic boosting tree
Gustavo Carneiro1, Bogdan Georgescu, Sara Good
1Siemens Corporate Research, Integrated Data Systems Dept., Princeton, NJ, USA.
Medical Image Computing and Computer-Assisted Intervention : MICCAI ... International Conference on Medical Image Computing and Computer-Assisted Intervention
|November 30, 2007
Summary
This study introduces a novel system for fast, automatic obstetric measurements using ultrasound images. The AI-powered approach achieves expert-level accuracy in segmenting and measuring fetal anatomy, improving upon previous methods.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Computational Biology
Background:
- Accurate fetal anatomical measurements in 2D ultrasound are crucial for prenatal care but challenging due to image quality issues.
- Existing methods often rely on rigid assumptions and struggle with complex anatomical variations and image artifacts.
- Previous approaches are limited in handling diverse fetal anatomical appearances and segmentation tasks like fetal abdomen delineation.
Purpose of the Study:
- To develop a novel, fast, and automatic system for obstetric measurements from 2D ultrasound images.
- To overcome limitations of prior methods in segmenting complex fetal structures and handling image noise and shadows.
- To achieve expert-level accuracy in automatic fetal measurements, including head circumference, biparietal diameter, abdominal circumference, and femur length.
Main Methods:
- A novel system leveraging a large database of expert-annotated fetal anatomical structures in ultrasound images.
- Training a discriminative constrained probabilistic boosting tree classifier to differentiate between target structures and background.
- Directly learning from data to handle complex appearances and previously unsolved segmentation problems, such as fetal abdomens.
Main Results:
- The system demonstrates effective automatic segmentation of fetal abdomens, a previously challenging task.
- Achieved fully automatic measurements for key fetal parameters: head circumference, biparietal diameter, abdominal circumference, and femur length.
- Extensive experiments show the system's accuracy is comparable to human experts in segmentation and measurement.
Conclusions:
- The proposed system offers a significant advancement in automatic obstetric measurements using ultrasound.
- The AI-driven approach provides fast and accurate fetal anatomical analysis, comparable to expert performance.
- This technology has the potential to enhance efficiency and consistency in prenatal diagnostics.
