Fat-Water Swaps in Iterative Decomposition of Water and Fat With Echo Asymmetry and Least-Squares Estimation Magnetic Resonance Imaging for Postinstrumentation Spine

J Comput Assist Tomogr. 2020 Nov/Dec;44(6):977-983. doi: 10.1097/RCT.0000000000001092.


Objective: To determine the frequencies of fat-water swaps in iterative decomposition of water and fat with echo asymmetry and least-squares estimation (IDEAL) water-only images of the postinstrumentation spine and discuss the efficiency of in-phase imaging in improving visibility of the thecal sac.

Materials and methods: A total of 276 patients (167 women; mean age, 62.3 years; range, 23-89 years) with metallic devices on the lumbar spine who received complete routine 1.5 T MR imaging, including axial and sagittal images of T1-weighted, T2-weighted, and T2-weighted IDEAL sequences, were included. The exclusion criteria were significant motion artifacts and severe metallic artifacts in any one of the sequences. The images were reviewed by two radiologists to identify fat-water swaps that were divided into 3 groups: extraspinal swaps, intraspinal swaps in sagittal images, and intraspinal swaps in axial images. The qualitative evaluations for the spinal canal in axial images were performed by rating on a five-point scale. Side-by-side comparisons of T2-weighted images and IDEAL in-phase images were also performed.

Results: In patient-based data of 276 patients, extraspinal fat-water swaps were noted in 10 patients (3.6%) and intraspinal swaps were noted in 160 patients (58.0%). The intraspinal swaps had a higher incidence in the patients with more levels of metallic devices with screws and the trend was not noted in the extraspinal swaps. A total of 928 axial levels were evaluated in the level-based data of axial images. T1-weighted, T2-weighted, and IDEAL in-phase images had significantly better imaging quality than the IDEAL water-only images (3.9 ± 0.4, 3.9 ± 0.3, 3.8 ± 0.4 vs 3.0 ± 1.3, all P < 0.001). Compared with T2-weighted images, most of the IDEAL in-phase images (reader 1, 90.9%; reader 2, 86.7%) present similar quality.

Conclusion: Iterative decomposition of water and fat with echo asymmetry and least-squares estimation sequence can provide good fat suppression in most spine MRI with metallic devices but the loss of cerebrospinal fluid signal intensities due to fat-water swaps are noted in more than half of postinstrumentation spine. Routine reconstruction of in-phase images is recommended to improve evaluation of the thecal sac by avoiding pitfall caused by fat-water swaps.

MeSH terms

  • Adipose Tissue / diagnostic imaging*
  • Adult
  • Aged
  • Aged, 80 and over
  • Artifacts*
  • Female
  • Humans
  • Image Processing, Computer-Assisted / methods*
  • Internal Fixators*
  • Least-Squares Analysis
  • Lumbar Vertebrae / diagnostic imaging
  • Magnetic Resonance Imaging / methods*
  • Male
  • Metals
  • Middle Aged
  • Retrospective Studies
  • Signal-To-Noise Ratio
  • Spine / diagnostic imaging
  • Water
  • Young Adult


  • Metals
  • Water