Purpose: MR combined with kVCT carries potential to accurately determine stopping power ratio (SPR) and I-value. To date, SPR accuracy has been limited by kVCT-based electron density (ED) uncertainties. This work aims to overcome this limitation by developing a novel multi-modal method to calculate ED and experimentally validate it in the determination of SPR in animal tissues
Methods: SPR was determined by Bethe-Bloch equation using a novel combination of previously published combined MR and CT (cMaC) methods to calculate I-value and ED. I-value, ED, and SPR were determined in ex-vivo animal tissue phantoms (brain and liver) by cMaC and compared against the corresponding ground truth (GT) values. Also, an ex-vivo intact porcine shoulder phantom was created that included soft tissue (muscle/organ), adipose, and bone components. Finally, cMaC-based and stoichiometric calibration-based SPR calculations were compared in scans from a head-and-neck cancer patient. In all cases, kVCT scans (120 kVp) were registered to proton density-weighted VIBE Dixon water/fat and UTE MRI scans, which were acquired on 3T MRI (MAGNETOM Vida, Siemens Healthcare, Erlangen, Germany) using a UTE prototype sequence. To externally validate ED accuracy in the porcine shoulder phantom, it was also scanned with MVCT.
Results: In the brain and liver phantoms, ED(cMaC) and SPR(cMaC) values were within 0.2% of GT, and I-value(cMaC) values differed from GT by <1.0eV. In the porcine shoulder phantom, ED(cMaC) and ED(MVCT) values had average differences of 1.4%, -2.6%, and 0.5% for bone, adipose, and soft tissue, respectively, compared to differences of 3.8%, -2.9%, and 0.1% between ED(kVCT) and ED(MVCT). In the head-and-neck data, SPR(cMaC) values deviated from SPR(Stoichiometric) by 3.8%, 0.2%, and 4.2% for bone, adipose, and soft tissue, respectively.
Conclusion: MRI, when combined with kVCT, has the potential to accurately determine ED, I-value, and SPR, which may lead to reduced range uncertainty in proton therapy.