Visualization of the Heart Through 3-D Reconstruction
The specific aim of this project was to optimize the acquisition parameters developed at the Ohio State University Magnetic Resonance Facility and the 3-D reconstructive techniques developed at the Ohio Supercomputer Center to aid in visualizing the structural qualities of the human heart. It was intended to serve as a feasibility experiment that would provide the basis for subsequent studies.
MRI can generate images of vascular flow non-invasively, without the need for contrast agent injection. The image contrast results from the fact that blood moves, while the surrounding tissue is stationary. Different acquisition techniques can make use of this basic fact in various ways.
For this study, two white females, one 28 years of age (Subject 1) and one 25 (Subject 2), were used in data acquisition. For each subject, two data sets were acquired. The first data set was obtained using a spin echo protocol by which the flow provides a low signal and appears dark in the image. The second data set was acquired with a gradient echo protocol, by which the flow provides a high signal and appears light in the image.
Spin Echo Imaging
Study 1: In this spin echo (dark flow) acquisition, the ability to segment the volumes of the heart's chambers from the endocardium is greatly facilitated. This is also true for the great vessels such as the ascending aorta. However, the external surface of the myocardium is difficult to discritize. Subsequently, stepping can be seen in the surface of the heart. For TE = 25ms, slow flowing blood is dark. This technique improves the contrast between the inner heart chambers. The TR = RR = 800ms; thus the heart signal is lower, and thus the contrast between the heart and the lung is lower.
Study 1: In this spin echo (dark flow) acquisition, it was easier to segment surface anatomy of the heart as well as the lumens of the great vessels and chambers of the heart. Compared to Subject 1, Study 1, the intensity gradient between the epicardium and the surrounding tissue was greater and facilitated segmentation. This gradient was less as the greater vessels were imaged superiorly. Additional images provided a data set of the entire heart, from apex to base, including some the great vessels. Overall, this acquisition was superior in facilitating segmentation and providing high quality surface anatomy. For TE = 5ms, only very fast moving blood is black. For TR = 2RR = 2000ms, heart muscle provides a higher signal; thus the heart lung contrast is better.
Example of 4 temporal acquisitions using gradient echo technique.
Study 2: In this gradient echo (light flow) acquisition, the ability to segment the chambers and lumens of the great vessels is difficult. Intensity values between the lumens and the myocardia presented some contrast; however, it was neither sufficient nor consistent in all sections. The surface of the myocardium is easily separated from surrounding structures. In addition, the 8 temporal data sets representing different times in the cardiac cycle, when played back in sequence, present a plausible representation of the beating heart.
Study 2: The gradient echo (light flow) presented a high quality representation of the external morphology of the heart. Segmentation of the lumens of the chambers and great vessels was difficult. In order to acquire more slices, the acquisition provided 4 images in time that comprised 15 sections. Surface anatomy was superior to Subject 1: Study 2: However, the appearance of beating was less convincing.
This research was supported by funding from General Electric and Childrens Hospital Research Foundation, Columbus. Additional thanks to Dr. John Wheller, Chief of Cardiac Catherization at Childrens, and Dr. Petra Schmalbrock, Research Scientist at the MRI Facility at The Ohio State University Hospitals. Further Reading
A Distributed Three-Dimensional Brain Visualization System
Shieh WK, Torello MW, Stredney DL, IEEE International Conference on Engineering in Medicine and Biology Society, 1990;12(3) 1175-1177
Volume MR Angiography: Methods to Achieve Very Short Echo Time
Schmalbrock P. et al, Radiology, 1990: 175: 861-865