Registration of Total Hip Replacement Implants from Subjects Imaged with 1.5 and 3 Tesla MRI Scanners Using Mavric
Date of Award
Master of Science (MS)
Koch, Kevin M.
Introduction: Total hip arthroplasty (THA) is a common treatment to correct arthritis and necrosis ailing the hip joint. An array of diseases can occur in response to the presence and wear of the implant. MRI is an imaging modality that can assess and diagnosis the diseases ailing the periprosthetic tissues, however images of a metal implants are susceptible to artifacts. Aim 1 of this study was to register the THAs imaged at 3T to 1.5T images; Aim 2 was to validate that metal susceptibility artifacts increase with field strength.Methods: 21 subjects with a total of 29 hip prosthetics that had coronal MAVRIC SL scans at 1.5T of their hips were recruited to be rescanned at 3T. MAVRIC SL is GE’s metal artifact reducing acquisition technology. The prosthetics imaged at 1.5T and 3T were then registered together. The acetabular and femoral component had to be registered independently since they are separate rigid bodies. The acetabular was registered by finding the registration transformation of a bone fixed to the component, then applying this transform to the 3T acetabular volume. The femoral components were registered by taking three anatomical landmarks and registering them together using a least-squares fitting of 3-D points algorithm. Lastly, the two 3T components were combined and then registered to the 1.5T prosthetic. The size of the volume of the of 1.5T and 3T images were compared.Results: Of the 29 hips scanned, 25 were viable for this study. The percent alignment of the 1.5T and 3T acetabular volumes had a mean = 90.03%, SD = 8.64%. For the femoral component the alignment had a mean = 67.89% and SD = 14.74. The total alignment had a mean = 77.64% and SD = 12.64%. The 3T artifact volumes were 7.93% and 5.49% smaller than the 1.5T volumes for the acetabular and femoral components respectively.Conclusion: While the acetabular component registration was successful, work still needs to be done for the femoral registration. The size of the artifact was larger at 1.5T than 3T, this was not expected and is a result of user error when tracing the two components.