Date of Award

January 2012

Document Type

Open Access Thesis

Degree Name

Medical Doctor (MD)



First Advisor

Jonathan N. Grauer

Subject Area(s)

Surgery, Medicine


Daniel J. Blizzard, Andrew H. Haims, Andrew W. Lischuk, Rattalerk Arunakul, Joshua W. Hustedt, and Jonathan N. Grauer. Department of Orthopaedics and Rehabilitation, Yale University School of Medicine, New Haven, CT.

The purpose of this study was to assess the diagnostic performance of three-dimensional, isotropic fast/turbo spin-echo (3D-TSE) in routine lumbar spine MR imaging.

Conventional 2D-FSE MRI requires independent acquisition of each desired imaging plane. This is time consuming and potentially problematic in spine imaging, as the plane of interest varies along the vertical axis due to lordosis, kyphosis, or possible deformity. 3D-TSE provides the capability to acquire volumetric datasets that can be dynamically reformatted to create images in any desired plane.

Eighty subjects scheduled for routine lumbar MRI were included in a retrospective trial. Each subject underwent both 3D-TSE and conventional 2D-FSE axial and sagittal MRI sequences. For each subject, the 3D-TSE and 2D-FSE sequences were separately evaluated (minimum 4 weeks apart) in a randomized order and read independently by four reviewers. Images were evaluated using specific criteria for stenosis, herniation, and degenerative changes.

The inter-method reliability for the four reviewers was 85.3%. Modified inter-method reliability analysis, disregarding disagreements between the lowest two descriptors for appropriate criteria (equivalent to "none" and "mild"), revealed average overall agreement of 94.6%.

Using the above, modified criteria, inter-observer variability for 3D-TSE was 89.1% and 88.3% for 2D-FSE (p=0.05), and intra-observer variability for 3D-TSE was 87.2% and 82.0% for 2D-FSE (p<0.01). The inter-method agreement between 3D-TSE and 2D-FSE was statistically non-inferior to intra-observer 2D-FSE variability (p<0.01).

This systematic evaluation showed there is a very high degree of agreement between diagnostic findings assessed on 3D-TSE and conventional 2D-FSE sequences. Overall, inter-method agreement was statistically non-inferior to the intra-observer agreement between repeated 2D-FSE evaluations.

Overall, this study shows that 3D-TSE performs equivalently, if not superiorly to 2D-FSE sequences. Reviewers found particular utility for the ability to manipulate image planes with the 3D-TSE if there was greater pathology or anatomic variation.