Overview: The parent project for this supplement aims to provide routine MRI of subjects with total joint replacements by reducing the severe image artifacts near metal, while offering highly efficient patient-specific scans that can detect bone loss, infection, and temperature changes near the implant in clinically feasible scan times. The supplement aims to incorporate deep learning techniques to better meet the parent grant goals. Relevance: Total joint replacements are one of the most successful orthopedic procedures, used annually to reduce pain from joint diseases in about one million patients in the United States (a number projected to double by 2030). However, about 10% of joint replacements fail in 5-10 years due to bone loss (osteolysis), infection, or other complications, often leading to revision surgery. Accurate, early, non-invasive assessment of complications remains limited, but would offer earlier and less invasive treatments, reduce unnecessary surgery, or allow better surgical planning. Approach: Prior to, and during the parent grant period, we have developed novel ?multi-spectral imaging? (MSI) MRI techniques that allow visualization of pathology adjacent to metallic implants, and together with other groups have successfully applied them to imaging of patients with devices including joint replacements and spinal fixation hardware. However these methods remain slow, have limited spatial resolution, and are challenging to use routinely. The recent growth of the machine learning field including convolutional neural networks (CNNs), and its application to medical imaging offers unique opportunities to substantially improve MRI near metal, and specifically the goals of the parent grant. We propose 3 small, independent aims in the supplement: (1) to bring fast, isotropic imaging near metal to clinical practice by using CNN-based methods to reduce reconstruction times to under 30 seconds, (2) to improve image quality away from metal by using a new reconstruction and CNN to avoid needing standard imaging in addition to MSI methods and (3) to reduce background-gradient induced artifacts near to metal using a CNN-based approach to enable better diagnosis of abnormalities adjacent to metal. Summary:
We aim to supplement our parent grant with CNN-based approaches to speed up scanning and image reconstruction, and to improve image quality near to and way from metal. These techniques will allow routine, non-invasive evaluation for earlier and more accurate detection and treatment of complications in these patients, as well as numerous other applications of MRI near metal implants.
There is a growing need for accurate diagnosis of complications surrounding joint arthroplasty, where MRI would provide excellent contrast if not for the fact that the presence of metal severely degrades images. Building on recent ideas in MRI and deep learning, we propose to develop practical methods for routine clinical imaging of patients with metal implants by increasing speed as well as offering image contrast that shows infection and other complications near the metal devices. Ultimately these methods will be tested and offered for widespread use to enable earlier and better treatment of complications resulting from arthroplasty, as well as for better understanding of the implications of different devices.
| Weber, Hans; Hargreaves, Brian A; Daniel, Bruce L (2018) Artifact-reduced imaging of biopsy needles with 2D multispectral imaging. Magn Reson Med 80:655-661 |
| Weber, Hans; Ghanouni, Pejman; Pascal-Tenorio, Aurea et al. (2018) MRI monitoring of focused ultrasound sonications near metallic hardware. Magn Reson Med 80:259-271 |
| Levine, Evan; Hargreaves, Brian (2018) On-the-Fly Adaptive ${k}$ -Space Sampling for Linear MRI Reconstruction Using Moment-Based Spectral Analysis. IEEE Trans Med Imaging 37:557-567 |
| Hargreaves, Brian A; Taviani, Valentina; Litwiller, Daniel V et al. (2018) 2D multi-spectral imaging for fast MRI near metal. Magn Reson Med 79:968-973 |
| Levine, Evan; Stevens, Kathryn; Beaulieu, Christopher et al. (2018) Accelerated three-dimensional multispectral MRI with robust principal component analysis for separation of on- and off-resonance signals. Magn Reson Med 79:1495-1505 |
| Levine, Evan; Daniel, Bruce; Vasanawala, Shreyas et al. (2017) 3D Cartesian MRI with compressed sensing and variable view sharing using complementary poisson-disc sampling. Magn Reson Med 77:1774-1785 |
| Quist, Brady; Shi, Xinwei; Weber, Hans et al. (2017) Improved field-mapping and artifact correction in multispectral imaging. Magn Reson Med 78:2022-2034 |
| Weber, Hans; Taviani, Valentina; Yoon, Daehyun et al. (2017) MR thermometry near metallic devices using multispectral imaging. Magn Reson Med 77:1162-1169 |
| Shi, Xinwei; Yoon, Daehyun; Koch, Kevin M et al. (2017) Metallic implant geometry and susceptibility estimation using multispectral B0 field maps. Magn Reson Med 77:2402-2413 |
| Choi, Soo-Jung; Koch, Kevin M; Hargreaves, Brian A et al. (2015) Metal artifact reduction with MAVRIC SL at 3-T MRI in patients with hip arthroplasty. AJR Am J Roentgenol 204:140-7 |
Showing the most recent 10 out of 11 publications
