Correction of respiratory and cardiac motion in cardiac PET/MR using MR-based motion modeling

Phys Med Biol. 2018 Nov 14;63(22):225011. doi: 10.1088/1361-6560/aaea97.

Abstract

Cardiac positron emission tomography (PET) imaging suffers from image blurring due to the constant motion of the heart that can impact interpretation. Hybrid PET/magnetic resonance (MR) has the potential to use radiation-free MR imaging to correct for the effects of cardio-respiratory motion in the PET data, improving qualitative and quantitative PET imaging in the heart. The purpose of this study was (i) to implement a MR image-based motion-corrected PET/MR method and (ii) to perform a proof-of-concept study of quantitative myocardial PET data in patients. The proposed method takes reconstructions of respiratory and cardiac gated PET data and applies spatial transformations to a single reference frame before averaging to form a single motion-corrected PET (MC-PET) image. Motion vector fields (MVFs) describing the transformations were derived from affine or non-rigid registration of respiratory and cardiac gated MR data. Eight patients with suspected cardiac sarcoidosis underwent cardiac PET/MR imaging after injection of 5 MBq kg-1 of 18F-fluorodeoxyglucose (18F-FDG). Myocardial regions affected by motion were identified by expert readers within which target-to-background ratios (TBR) and contrast-to-noise ratios (CNR) were measured on non-MC-non-gated, MC-PET, and double respiratory and cardiac gated PET images. Paired t-tests were used to determine statistical differences in quantitative uptake-measures between the different types of PET images. MC-PET images showed less blurring compared to non-MC-non-gated PET and tracer activity qualitatively aligned better with the underlying myocardial anatomy when fused with MR. TBR and CNR were significantly greater for MC-PET (2.8 ± 0.9; 21 ± 22) compared to non-MC-non-gated PET (2.4 ± 0.9, p = 0.0001; 15 ± 13, p = 0.02), while TBR was lower and CNR greater compared to double-gated PET (3.2 ± 0.9, p = 0.04; 6 ± 3, p = 0.004). This study demonstrated in a patient cohort that motion-corrected (MC) cardiac PET/MR is feasible using a retrospective MR image-based method and that improvement in TBR and CNR are achievable. MC PET/MR holds promise for improving interpretation and quantification in cardiac PET imaging.

Publication types

  • Research Support, N.I.H., Extramural

MeSH terms

  • Aged
  • Cardiac-Gated Imaging Techniques / methods*
  • Cardiac-Gated Imaging Techniques / standards
  • Fluorodeoxyglucose F18
  • Humans
  • Magnetic Resonance Imaging / methods*
  • Magnetic Resonance Imaging / standards
  • Middle Aged
  • Motion
  • Multimodal Imaging / methods
  • Multimodal Imaging / standards
  • Positron-Emission Tomography / methods*
  • Positron-Emission Tomography / standards
  • Radiopharmaceuticals

Substances

  • Radiopharmaceuticals
  • Fluorodeoxyglucose F18