NA-MIC Project Weeks

Back to Projects List

Using statistical shape modeling for improving 3D reconstruction of fetal ultrasound images

Key Investigators

• Chi Zhang (Texas A&M University School of Dentistry)
• Emet Schneiderman (Texas A&M University School of Dentistry)
• Preetam Schramm (Texas A&M University School of Dentistry)
• Zohre German (Texas A&M University School of Dentistry)
• Ju-Ying Lin (Texas A&M University School of Dentistry)

Presenter location: In-person

Project Description

We are initiating a multi-center (led by Dr. Emet Schneiderman) study to understand the impact of maternal obstructive sleep apnea (OSA) and different treatment plans (i.e., mouth appliances) on both pregnant women and fetuses, including fetal craniofacial dysmorphology and growth issues. A large sample size of ultrasound images will be collected. However, the fetuses may have different poses and partial face covered, and the ultrasonographic images have lots of noise. We are interested in using statistical modeling to refine 3D segmentation to diagnose fetal orofacial dysmorphology, reconstruct 3D growth trajectories, explore epigenetic effects of facial growth problems and dysmorphology, predict newborn facial shapes, etc.

Objective

1. Use statistical shape modeling tools or other tools from Slicer and Kitware to refine 3D reconstruct of fetal faces based on ultrasonography. Aesthetics is also important because part of the goal is to make parents aware of the sleep treatment.

Approach and Plan

1. Segmentation of noisy US images
2. Learn and explore techniques in statistical shape modeling and ultrasound image processing using 3D slicer and other tools developed by Kitware; set up a plan.
3. Locate sample data for experimenting.
4. Develop potential collaborations

Progress and Next Steps

1. Preparing grant application and data collection plans
2. See above

PW Progress

1. Potential collaborators: Haichong Zhang and Xihan Ma from WPI
2. Find fetal facial features related to maternal sleep apnea that can only be quantified by 3D rather than standard 2D US imaging.

Illustrations

No response

Background and References

The most recent study is [Alomar et al. 2022]https://www.sciencedirect.com/science/article/pii/S0169260722002759 using new born babies to build a statistical model for US refinement.