Determining Cardiac Fiber Orientation Using FSL and Registered Ultrasound/DTI volumes

• Type of Material: Conference Paper
• Authors: James Dormer, Georgia Institute of Technology; Xulei Qin, Emory University; Ming Shen, Emory University; Silun Wang, Emory University; Xiaodong Zhang, Emory University; Rong Jiang, Emory University; Mary Wagner, Emory University; Baowei Fei, Emory University
• Language: English
• Date: 2016-01-01
• Publisher: Emory University Libraries
• Publication Version: Author Accepted Manuscript (After Peer Review)
• Copyright Statement: © 2016 Society of Photo Optical Instrumentation Engineers. One print or electronic copy may be made for personal use only. Systematic reproduction and distribution, duplication of any material in this paper for a fee or for commercial purposes, or modification of the content of the paper are prohibited.
• Final Published Version (URL): http://dx.doi.org/10.1117/12.2217296
• Title of Journal or Parent Work: Proceedings of SPIE
• Conference or Event Name: Conference on Medical Imaging - Ultrasonic Imaging and Tomography
• Volume: 9790
• Grant/Funding Information: This work was partially supported by NIH grants (CA176684 and CA156775).
• Keywords: Radiology, Nuclear Medicine & Medical Imaging; Science & Technology; VALIDATION; FIELD; Physical Sciences; ARCHITECTURE; DIFFUSION; Life Sciences & Biomedicine; Optics; MODEL; IMAGES
• Research Categories: Engineering, Biomedical; Health Sciences, Radiology

Abstract

Accurate extraction of cardiac fiber orientation from diffusion tensor imaging is important for determining heart structure and function. However, the acquisition of magnetic resonance (MR) diffusion tensor images is costly and time consuming. By comparison, cardiac ultrasound imaging is rapid and relatively inexpensive, but it lacks the capability to directly measure fiber orientations. In order to create a detailed heart model from ultrasound data, a three-dimensional (3D) diffusion tensor imaging (DTI) with known fiber orientations can be registered to an ultrasound volume through a geometric mask. After registration, the cardiac orientations from the template DTI can be mapped to the heart using a deformable transformation field. This process depends heavily on accurate fiber orientation extraction from the DTI. In this study, we use the FMRIB Software Library (FSL) to determine cardiac fiber orientations in diffusion weighted images. For the registration between ultrasound and MRI volumes, we achieved an average Dice similarity coefficient (DSC) of 81.6±2.1%. For the estimation of fiber orientations from the proposed method, we achieved an acute angle error (AAE) of 22.7±3.1° as compared to the direct measurements from DTI. This work provides a new approach to generate cardiac fiber orientation that may be used for many cardiac applications.