MENU

Click here to

×

Are you sure ?

Yes, do it No, cancel

Elemental-Resolved Multi-Energy CBCT Reconstruction Using a Linear Forward CT Model

M Tsai12*, C Shen1 , B Li1 , X Jia1 , (1) UT Southwestern Medical Center, Dallas, TX, (2) National Taiwan University, Taipei, Taiwan

Presentations

(Thursday, 8/2/2018) 1:00 PM - 3:00 PM

Room: Room 202

Purpose: Cone-beam CT (CBCT) is a commonly used image-guidance tool in radiotherapy, yet its usage is limited to patient positioning. Motivated by successful realization of multi-energy CBCT (ME-CBCT) on a CBCT platform, we propose a multi-energy elemental-resolved (MEER) reconstruction framework with a linearized forward CT model (L-MEER). It allows to directly reconstructing images of elemental composition (EC) and relative electron density to water (rED). L-MEER CBCT may permit advanced applications such SPR estimation and low-concentration contrast agent identification.

Methods: L-MEER is formulated as an optimization problem to simultaneously solve for x-ray images at different channels, EC and rED images. Data consistency between x-ray image and EC and rED is realized by a forward CT number model. We further introduced a variable transform to convert the nonlinear forward model into a linear form, simplifying the optimization problem. Moreover, a dictionary containing ECs of commonly encountered human tissues is built to incorporate prior information. We assume that the solution EC can be approximated by a sparse combination of ECs of dictionary materials. We developed an alternating minimization algorithm to solve the L-MEER optimization problem. Simulation and experimental studies were conducted to demonstrate its effectiveness.

Results: We obtained average errors of 1.4% and 0.4% in EC and rED respectively in simulation study using a digital NCAT phantom. In experiments, Gammax phantom was scanned on a Varian TrueBeam CBCT platform with 80, 100 and 120kVp under a kVp switching scheme . The resulting EC and rED errors were 2.2% and 1.9%. Constrast-to-noise ratios (CNR) is more than two times higher than those of images reconstructed by conjugate gradient least square method.

Conclusion: We have developed L-MEER framework for ME-CBCT. The high quality image and EC and rED information potentially permit advanced applications on conventional CBCT platform.

Keywords

Not Applicable / None Entered.

Taxonomy

Not Applicable / None Entered.

Contact Email