High-Resolution Free-Breathing Chemical-Shift-Encoded MRI for Characterizing Lymph Nodes in the Upper Abdomen.

Abstract

Objectives: Accurate lymph node (LN) staging is crucial for managing upper abdominal cancers. Ultrasmall superparamagnetic iron oxide (USPIO)-enhanced magnetic resonance imaging effectively distinguishes healthy and metastatic LNs through fat/water and -weighted imaging. However, respiratory motion artifacts complicate detection of abdominal LNs. This study evaluates if a free-breathing radial stack-of-stars acquisition can match or outperform Cartesian reference scans to visualize LNs and depict uptake of USPIO nanoparticles.

Materials and methods: Five volunteers with USPIO and 20 patients without USPIO were scanned using radial stack-of-stars, Cartesian dual-echo, and fat-saturated Cartesian multiecho sequences for fat/water imaging and estimation. Reconstructed images from radial and Cartesian patient data underwent qualitative comparison by 2 radiologists. LNs were identified in all fat/water images, LN short-axis sizes were measured, and relaxation rates were analyzed using linear correlations and Bland-Altman plots.

Results: Radial imaging provided better image quality than the Cartesian reference standard, according to both readers. Substantially, more LNs were identified in radial compared with Cartesian datasets (349 vs 202). Median short-axis diameters showed a significant difference, measuring 2.7 mm (interquartile range [IQR]: 2.7-4.6 mm) for radial images and 4.5 mm (IQR: 3.7-5.6 mm) for Cartesian images ( P < 0.0001). Relaxation rates measured in radial data showed a significant linear correlation with the Cartesian reference (Pearson correlation coefficient: 0.90 with P < 0.0001). Bland-Altman plots indicated a slight bias with a mean difference (MD) of 3.9 s -1 and limits of agreement at MD ± 16.4 s -1 .

Conclusions: This work presents a promising magnetic resonance imaging method to depict upper abdominal LNs and to visualize their USPIO uptake. Instead of multiple Cartesian breath-hold scans, all relevant contrasts and parameters are obtained from a single free-breathing radial acquisition. The proposed method yielded higher image quality and more sensitive detection of small LNs. value analysis showed a strong linear correlation with the reference, albeit with minimal biases.