Apparent tissue sodium concentration quantification across reconstruction methods with different partial volume effect reduction - A preliminary study.

Abstract

Introduction: In sodium (²³Na) magnetic resonance imaging (MRI), partial volume effects (PVE) are one of the most common causes of errors in the in vivo quantification of apparent tissue sodium concentration (aTSC). Advanced image reconstruction algorithms, such as compressed sensing (CS), have the potential to reduce PVE. This study aimed to investigate the feasibility of CS-based reconstruction techniques for improving image quality, and to assess whether differences in aTSC exist when images are reconstructed using different methods.

Subjects and methods: In this study, three healthy participants and 12 female participants with breast cancer were examined on a 7T MRI scanner. ²³Na-MRI images were reconstructed using weighted total variation (wTV), directional total variation (dTV), anatomically guided total variation (AG-TV) and adaptive combine (ADC) methods. The consistency of tumor volume delineations based on sodium data was assessed using the Dice score, and aTSC quantification was performed for various image reconstruction methods. Pairwise comparisons of aTSC values between reconstruction methods (wTV, dTV, AG-TV, and ADC) were conducted using the Wilcoxon signed-rank test.

Results: All methods provided breast MRI images with well-preserved sodium signal and tissue structures. The mean Dice scores for wTV, dTV, and AG-TV were 65±11%, 72±9%, and 75±6%, respectively. Mean aTSC (in mmol/L) values in glandular tissue were 52.0±4.0 for wTV, 55.0±4.0 for dTV, 63±3.0 for AG-TV, and 69.0±3.0 for ADC reconstructions. In adipose tissue, mean values of aTSC were 32.0±4.0, 29.0±3.0, 19.0±2.0, and 23.0±3.0 for wTV, dTV, AG-TV, and ADC, respectively. In tumors, mean aTSC were as follows: 80.0±5.0 for wTV, 85.0±5.0 for dTV, 99.0±5.0 for AG-TV, and 105.0±7.0 for ADC. Post-hoc Wilcoxon signed-rank tests showed that AG-TV differed significantly from both wTV and dTV across all tissues (all p< 0.001), and that wTV also differed significantly from dTV (p=0.038 in gland, p=0.027 in adipose, and p=0.009 in tumor).

Conclusion: Tumor appearance and aTSC estimations depend on the reconstruction method and parameter selection, likely due to differences in PVE reduction. Among the evaluated approaches, AG-TV showed the most favorable overall performance for ²³Na-MRI of breast tumors under the studied conditions.