The value of multisequence MR in the morphological evaluation of peripheral lung cancer

Abstract

Background

An increasing number of studies have shown that thoracic magnetic resonance imaging (MRI) has potential value in the early diagnosis of lung cancer. While the ability of T2 fast-BLADE (T2fBLADE), three-dimensional T1-weighted volumetric interpolated breath-hold examination (T1-VIBE) and T1 mapping to display morphological signs of lung cancer has been less well studied. Our study aimed to compare the image quality of stack-of-stars VIBE (Star-VIBE) and T1-VIBE and to investigate the ability of T2fBLADE, T1-VIBE, Star-VIBE and T1 mapping to show the morphological signs of lung cancer.

Methods

Patients with a proposed diagnosis of peripheral lung cancer by computed tomography (CT) were collected prospectively and underwent MR scanning before pathological confirmation. MR scan Star-VIBE, T1-VIBE, T2fBLADE and T1 mapping sequences. Signal to noise ratio (SNR) and contrast to noise ratio (CNR) of Star-VIBE and T1-VIBE images were independently measured and compared by two radiologists. With CT images as the gold standard, the morphological signs (including lobulation, spiculation, pleural retraction, halo, cavitation, air bronchogram, necrosis, pleural effusion and mediastinal lymphadenopathy) and maximum diameters of the lesions on four MR sequences were evaluated and measured. A paired Wilcoxon test was used to compare the SNR, CNR and Likert scale of T1-VIBE and Star-VIBE. The Kruskal‒Wallis H test was used to compare the subjective scores of MR sequences. The intraclass correlation coefficient (ICC) and Cohen’s kappa coefficient (k) were used to compare the agreement of lesion size and subjective scores between the two radiologists.

Results

65 patients with a total of 69 lesions were included. The image quality of Star-VIBE was statistically higher than T1-VIBE (P < 0.05). The overall detection rate of morphological signs of lung cancer by MR was 76.9–100%. MR (Star-VIBE and T1-VIBE) have a high sensitivity and specificity (96.5−100%) for the display of various morphological signs. The detection rates of halo and cavitation by Star-VIBE and T1-VIBE (61.5−83.3%) were higher than those by the T2fBLADE (23.1% and 33.3%). The detection rate of necrosis by the T2fBLADE (84.2%) was higher than that of Star-VIBE and T1-VIBE (73.7% and 63.2%), and the detection rate of residual signs was similar (91.7−100%). Except for lobulation, spiculation and pleural effusion (81.3−100%), the detection rates of all morphological signs by the T1 mapping sequence were low (0−77.4%). The subjective scores of lobulation, spiculation, pleural retraction and mediastinal lymphadenopathy in Star-VIBE and T2fBLADE were higher than in T1-VIBE (P < 0.013).

Conclusions

Star-VIBE can provide higher image quality than T1-VIBE. Combined multisequence MR can better show the morphological signs of peripheral lung cancer.