Université d'Auvergne Clermont1 | CNRS


Axis 4: 3D regional myocardial perfusion and function


Collaboration with the Nuclear Medicine Department of the Jean Perrin Center has been going on for years. It has been strengthened thanks to the recent need of multimodal registration methods in 3D cardiac imaging. A 2D/3D robust mutual information framework previously used in the orthopaedic field [Aouadi08a] has been successfully extended to 3D/3D SPECT vs. CT registration and adaptation to 3D/3D SPECT vs. MRI is planned.

Two clinical studies about myocardial perfusion are underlying these developments:

  • The assessment of perfusion quantification with the new D-spect camera in comparison with delayed-enhancement MRI
  • The superposition of the coronary tree geometry segmented from CT with SPECT perfusion in order to derive patient-specific sectorization [Chauvet09a]
Coronary tree geometry obtained from CT superimposed to SPECT perfusion after registration  Bull's eye representation and pixel affectation to the main coronary branches ( LAD in blue, LCX in green and RCA in yellow) with the AHA 17-segment model

Deformables surfaces with parameterization adapted to the AHA 17-segment model are being developped in order to segment or track myocardial walls from cine-MRI, 3D ultrasound and gated-SPECT:

B-spline tensor deformable surfaces

One of our major challenge dealing with myocardial function is to be able to couple parietal movement with regional movement estimation. Preliminary results were obtained in 2D cardiac ultrasound with independent segmentation and movement estimation and compared to other contraction estimation methods [Frouin06a,Kachenoura07a]. The coupling will be studied in a variational framework assuming conservation of densitometric 3D image gray levels.