Title: Visualization of Tidal and Reserve Ventilation from Four Dimensional Computed Tomography

Speaker: Thomas Guerrero and Richard Castillo, University of Texas MD Anderson Cancer Center, Edward Castillo, Mathematics Department, University of California at Irvine

Date/Time: September 4, 2008, 2:00-3:00pm

Location: CSRI Building, Room 148 (Sandia NM)

Brief Abstract: Four dimensional computed tomography (4D CT) developed for radiotherapy treatment planning, is a new imaging technique that allows the acquisition of a time-varying 3D CT image sequence of of the patient’s lungs during the respiratory cycle (Ford, Mageras et al. 2003; Vedam, Keall et al. 2003; Pan, Lee et al. 2004). We have developed a method for extracting ventilation images from 4D CT (Guerrero, Sanders et al. 2006) which represent a new paradigm for the use of CT. Extracting ventilation images from 4D CT (Guerrero, Sanders et al. 2005; Guerrero, Sanders et al. 2006) requires an accurate voxel by voxel mapping by a deformable image registration (DIR) algorithm to link corresponding tissue elements. This dependence on spatial accuracy is even more crucial when ventilation is extracted directly from the deformation alone as other investigators have proposed (Reinhardt, Christensen et al. 2007). Our CT ventilation implementation utilizes the CT values to calculate ventilation. To reduce image artifacts and improve the quality of 4D CT derived ventilation images we need the most spatially accurate DIR algorithm. Recently we demonstrated improved spatial accuracy for the lung using a compressible flow (versus incompressible) optical flow DIR algorithm in a preliminary study (Castillo, Castillo et al. 2008). In this proposal, we will generate a library of manual deformation fields from radiotherapy 4D CT images of patients with esophagus or lung cancer cases for testing of DIR algorithms. We will test DIR algorithms available to us and establish a website where the medical imaging community at large can download (a subset) test images and upload deformations for evaluation. We propose to focus on the development of next generation DIR algorithms which utilize the entire 4D CT image set to increase spatial accuracy and extract 4D ventilation images in collaboration with Sandia National Laboratories. Since 4D CT images are acquired while the patient breaths and patients breathing varies breath-to-breath, there are motion artifacts and discontinuities in these images. We propose that a DIR algorithm can be developed which identifies and skips over artifact regions in the 4D image set. The ideal DIR algorithm will also provide an estimate of the individual result's spatial accuracy.

CSRI POC: James R. Stewart, (505) 844-8630