Session Item

Cone-beam CT studies have shown that extremity soft-tissue sarcoma (ESTS) volumes may change substantially during radiotherapy (RT), leading to plan adaptations. MRI-guided RT is expected to improve the targeting precision in ESTS patients, as it provides superb soft-tissue contrast. Closed-bore high-field (HF) MRI scanners are most adopted in clinical settings. However, the versatility of open low-field (LF) MRI scanners makes them a worthwhile candidate to be investigated for MRI-guided proton therapy (PT). The present study aims to evaluate the setup reproducibility for the upper extremity positioning in both LF and HF MRI scanners.

Five healthy volunteers (BMI 20.5±2.2) laid both supine and prone with neutral and pronation forearm positions in the HF and LF MRI scanners. Participants laid with their arm stretched above the head on the HF scanner and abducted 90° on the LF scanner. Additionally, for two of them, supine positioning with the arm next to body in the HF MRI was possible. An in-house developed arm holder with a 180° hand rotation range allowed reproducible positioning on the MRI couches. In a systematic study, setup positioning comparison within and between HF scans and LF scans was performed. Position differences, soft-tissue deviations, and Dice similarity coefficients (DSC) were derived after image registration. Setup differences were expressed as two couch shifts and rotation values (Δx, Δy, Δα). Soft-tissue deviations were defined as skin-to-bones distances at angles of 30° (Δdsb) (Figure 1).

A traffic light protocol was developed to evaluate positioning scenarios. Comparisons of LF and HF MRI scans proved an overall good registration performance for participants lying in the same position on both scanners (92.3% green light, mean Δdsb 1.4±0.8 mm). Moreover, different lying positions lead to worse registration performance (52.4% yellow light, mean Δdsb 2.4±1.6 mm), especially for higher BMI values. Comparisons of LF MRI scans of repositioned arms resulted in excellent registration performance (100% green light, mean Δdsb 0.6±0.4 mm). Lastly, comparison of HF MRI scans on different body positions demonstrated basically good registration performance (62.5% green light, mean Δdsb 1.2±0.8 mm), especially for lower BMI values (Table 1).

Different upper extremity positioning in LF and HF MRI scanners allowed for reproducible setup scenarios in both scanners. Comparisons of LF and HF MRI scans with the same lying position and different arm orientations suggest that LF MRI can be further explored as an alternative to HF MRI. Comparisons of repositioned arm LF scans encourage applicability in patient setup verification. Therefore, the unique arm abducted 90° positioning in an open LF MRI scanner, which enables a free path for charged particle beams, would allow for in-beam MRI-guided PT of ESTS. Finally, HF scans comparisons advise patient positioning on the HF MRI scanner based on the most comfortable position.