Comparison of 3D bone models of the knee joint derived from CT and 3T MR imaging. Issue 93 (August 2017)

• Record Type: Journal Article
• Authors: Neubert, Aleš; Wilson, Katharine J.; Engstrom, Craig; Surowiec, Rachel K.; Paproki, Anthony; Johnson, Nicholas; Crozier, Stuart; Fripp, Jurgen; Ho, Charles P.
• Is Part Of: European journal of radiology. Issue 93(2017)
• Page Start: 178
• Page End: 184
• Publication Date: 2017-08
• Subjects: 3D Three-dimensional; Magnetic resonance imaging -- Computed tomography -- Knee -- 3D bone modeling -- Patient specific templatesCT − computed tomography -- FOV − field of view -- FROI − focused region of interest -- MR − magnetic resonance; Medical radiology -- Periodicals; Radiology -- Periodicals; Radiologie médicale -- Périodiques; Medical radiology; Periodicals; 616.075705
• DOI: 10.1016/j.ejrad.2017.05.042
• Languages: English
• ISSNs: 0720-048X

Abstract

Highlights: MR based knee bone models have sub-millimeter differences compared to the reference CT model. VIBE based knee bone models have, compared to SPACE and TrueFISP, the best agreement with reference CT model. Novel approach using global and local similarity metrics for analysis of bone model similarities is proposed. Abstract: Purpose: To examine whether magnetic resonance (MR) imaging can offer a viable alternative to computed tomography (CT) based 3D bone modeling. Methods: CT and MR (SPACE, TrueFISP, VIBE) images were acquired from the left knee joint of a fresh-frozen cadaver. The distal femur, proximal tibia, proximal fibula and patella were manually segmented from the MR and CT examinations. The MR bone models obtained from manual segmentations of all three sequences were compared to CT models using a similarity measure based on absolute mesh differences. Results: The average absolute distance between the CT and the various MR-based bone models were all below 1 mm across all bones. The VIBE sequence provided the best agreement with the CT model, followed by the SPACE, then the TrueFISP data. The most notable difference was for the proximal tibia (VIBE 0.45 mm, SPACE 0.82 mm, TrueFISP 0.83 mm). Conclusions: The study indicates that 3D MR bone models may offer a feasible alternative to traditional CT-based modeling. A single radiological examination using the MR imaging would allow simultaneous assessment of both bones and soft-tissues, providing anatomically comprehensive joint models for clinical evaluation, without the ionizing radiation of CT imaging.