Purpose: To develop a novel 5D-MRI technique capable of producing new 4D-MRI images with versatile image contrasts and improved image quality for radiation therapy of abdominal cancers.
Methods: The 5D-MRI technique extended 4D-MRI with an additional dimension of â€œimage contrastâ€? and was developed by synergizing three main techniques: 4D-MRI, deformable image registration (DIR), and multi-parametric MRI fusion (MPMF). The MPMF method consisted of five key components: input multiple standard contrast MRIs, image pre-processing, fusion algorithm, fusion adaptation based on clinical needs, and output new contrast 3D-MRI. The 4D displacement vector fields (4D-DVF) were generated by deforming one reference phase (EOE) to 9 other phases via DIR from the original 4D-MRI. 5D-MRI were generated by applying 4D-DVF to each of the new contrast 3D-MRI. The 5D-MRI technique was examined on digital human phantom (XCAT) infused with patient data and seven liver cancer patients. Contrast versatility, image quality (tumor CNR and liver SNR) improvement and tumor motion accuracy were evaluated. Tumor motion trajectories mean difference in displacement (D) and correlation coefficient (CC) to the reference cine MRI were evaluated.
Results: 5D-MRI or new 4D-MRIs with versatile image contrasts were generated in both XCAT phantom and liver cancer patients. As one of many image quality improvement examples, compared to the highest tumor contrast standard MRI (DWI, tumor CNR: 7.53Â±2.65), 5D-MRI-A achieved a tumor CNR of 8.67Â±3.22, while significantly improved liver SNR from 4.93Â±2.36 (DWI) to 11.93Â±1.18 (5D-MRI-A) (p<0.0001). Tumor motion trajectories of 5D-MRI closely matched with that of the reference cine MRI: D and CC ranged from 0.402 to 0.561 mm (mean=0.463 mm) and from 0.980 to 0.987 (mean=0.986), respectively.
Conclusion: A novel 5D-MRI technique capable of producing versatile new contrast 4D-MRI images and improved image quality was developed. It holds promises in improving versatility and quality of 4D-MRI for radiation therapy.
Funding Support, Disclosures, and Conflict of Interest: This work was supported by NIH grant R21CA165384.