Fetal and Mother Numerical Models (FEMONUM)

  1. Outline of the project
  2. Fetal Organs Modeling
    1. 3D Ultrasound Segmentation
    2. MRI Segmentation
  3. Pregnant Woman body
    1. Generic woman body envelope
  4. Scientific Partners &  Roles
  5. Publications    
  6. Documents

Outline of the project

Several international scientific organizations and institutes, such as the World Health Organization (WHO) and the COST 281, have expressed the need for developping numerical models of the human body to enable precise studies of the interactions between radio-frequency electromagnetic waves and biological tissues. Several previous works have focused on developing models of adults and children heads, based on magnetic resonance imaging (MRI) data. New usage habits of mobile phones (hand free kits, ...) and the introduction of new technologies based on electromagnetic fields (Wifi, ...) have raised the need to develop whole body numerical models, based on MRI and computerized tomography (CT) data. With the advent of obstetrical imaging, models of the fetus and the pregnant woman are now considered.



This project, entitled FEMONUM (FEtus and MOther NUmerical Models), aims at providing numerical models generated from a large database of imaging exams. The proposed models, evaluated by an obstericians and a group of pediatric radiologists, will enable precise dosimetry studies on the complex, highly variable and evolving anatomy of the pregnant woman.



This project was funded by the Fondation Santé et Radiofréquence. The numerical models will be made available to the scientific community, filling an important gap in the list of existing human body models. 

The modeling approach is based on medical image segmentation and smooth surface mesh generation. Two types of medical imaging modalities are used to screen and model the fetus:

  1. 3D Ultrasound data for fetuses during the 1st and 2nd trimesters.
  2. MRI data for fetuses during the 3rd trimester.

Utero-Fetal Unit (UFU) Modeling

3D Ultrasound Segmentation

A database of 19 volumes of 3D ultrasound acquired between 8WA et 22 WA was provided by collaborating obstetricians (Hospital Beaujon, Paris, FRANCE) and by Philips Healthcare. Segmentation of the fetus body is performed in two steps:

  1. As a first step, we distinguish the amniotic fluid from maternal and fetal tissues. An automated segmentation framework based on statistical homogeneity measures was designed to perform this task.  This work was publised in a conference paper.





2. As second step, segmentation errors and connections between the fetus and the uterine wall were manually corrected to generate a final mesh model.

(1) field of view, (2) uterine internal wall.

(3) uterine wall and fetus body, (4) fetus body.


MRI Segmentation

MRI image data provide a clear visualization of several maternal abdominal tissues (skin, bone, fat, muscles,...) et of fetal organs (body envelope, brain, lungs, heart,...).

In collaboration with the pediatric radiology department of the hospital Saint Vincent de Paul (Paris, France), we first performed a study to define the best suited MRI protocol for the segmentation and modeling of the fetus. The Steady State Free Precession (SSFP) sequence was selected, offering the best image quality for the segmentation task. Results of this study were published in a paper.

Examples of SSFP & FFSE MRI data of fetuses.

A methodological framework was developed for automated segmentation of several fetal structures, including the eyes, the brain-skull content, the spinal cord, the urinary bladder, and the lungs. The segmentation process is based on the extraction of landmark points based on appearance models, iterative orientation of the fetus based on the landmark points and fine segmentation of tissue interfaces with a contrast-sensitive graph-cut partitioning of image data, within narrow bands of image data. 

.

Extraction of Landmark Points on the Foetal SkeletonRegistration of a Generic Foetal Articulated Model on the Skeleton

Final Segmentation of the Foetal Body Envelope with Graph-Cuts

Refinment of the 3D Fetus Segmentation with Internal Organs


Body envelope models with very high-resolutions were extracted from 3D SSFP MRI data sets
 

Pregnant Woman body

Generic woman body envelope

Victoria is a non-gravid woman body envelope developed by Daz Studio, which can be manipulated and animated with the software tool Blender. 

We used Victoria's envelope to generate hybrid models of pregrant women: 

  1. For models based on 3D ultrasound data, the abdominal surface of Victoria was not modified.
  2. For models based on MRI data, the abdominal surface of Victoria was deformed interactively to fit the contours of the maternal envelope partially visible on the MRI images.
To handle the effect of the gravity and the differences between standing and laying down positions in terms of organ shapes and orientations, we manually ajusted the uterus position and orientation within Victoria's  abdomen, under the control of an obstetrician.

Mesh-based pregnant woman model (fetal model based on MRI data)

A pixel-based model is finaly generated, at a desired spatial resolution, with different labels assigned to the different tissues. 



Pixel-based pregnant woman model

Scientific Partners &  Roles

Publications    

Documents

Retour | Back