Projects:ScarIdentification

From NAMIC Wiki
Jump to: navigation, search
Home < Projects:ScarIdentification

Back to Boston University Algorithms


Description

To segment the scaring tissue from DE-MRI images, we proposed an effective method that utilizes the geometric and intensity information of the left atrial wall. The key observation is that the scaring tissue lies around a thin layer outside the LA chamber and has high intensity values. In order to identify the scar in the image, we first extract the wall region of the LA. Then, the extraction of the scar is restricted in the wall region.

  • ScarIntensities.png

Scars in a DE-MRI image. The LA is highlighted with red contour. From left to right: scars in axial, sagittal, and coronal views, respectively.

Results

  • LAWall.png

Extracted LA wall. From left to right: LA wall in axial, sagittal, and coronal views, respectively.

  • LAScar.png

From left to right: scars in axial, sagittal, and coronal views, respectively.

Key Investigators

  • Georgia Tech: Liangjia Zhu and Anthony Yezzi
  • BWH: Yi Gao and Sylvain Bouix
  • Boston University: Allen Tannenbaum

Publications

Y. Gao, L. Zhu, A. Yezzi, S. Bouix , A. Tannenbaum. Scar Segmentation in DE-MRI, IEEE International Symposium on Biomedical Imaging (ISBI) , 2012.