An Approach to Multimodal Image Segmentation

• Publication date: 2006

Radiotherapy treatment planning requires exact delineation of tumors in acquired images. Further, motion and deformation of a tumor due to respiratory movements must be known in order to better target and dose the treatment radiation. Here we present and implement a pipeline for the registration of respiratory correlated multimodal RC-CT-PET image sequences. Further, a multimodal region-based non-parametric active-mesh segmentation framework is presented, that allows to correctly delineate objects in multimodal volumetric images. Applications to MRI brain image registration and extraction, as well as RC-CT-PET lung tumor delineation are shown. These tools allow for the delineation and the volumetric motion tracking and deformation analysis of lung tumors throughout the respiratory cycle, and therefore represent a major achievement for radiotherapy treatment planning.

Keywords: Image Processing ; Multimodal ; Registration ; Segmentation ; Motion Tracking ; Deformation Analysis ; Respiration Correlation ; CT-PET

Reference

• LTS-STUDENT-2007-001

Record created on 2007-03-07, modified on 2016-08-08