Diffusion-Weighted Magnetic Resonance Imaging and Receiver Operating Characteristic (roc) Analysis for Lung Cancer Diagnosis

Yahui Peng,Yongxia Lei,Xinchun Li,Houjin Chen
DOI: https://doi.org/10.2991/itms-15.2015.163
2015-01-01
Abstract:Diffusion-weighted magnetic resonance (DWI-MR) technique is a promising imaging method for lung cancer diagnosis. The purpose of this study is to investigate whether DWI-MR can differentiate lung cancer from infectious nodule accurately. Approved by institutional review board, this study included DWI-MR images of 38 patients who had solitary lung nodules in chest CT examination. Three different non-zero b-values, 600, 800, and 1000 s/mm(2), were used for DWI-MR imaging and apparent diffusion coefficient (ADC) was calculated for lung cancer detection. Histopathological analysis of biopsy or anti-inflammation treatment results were used as the gold standard. Area under the receiver operating characteristic curve (AUC) was used as the figure of merit for diagnostic accuracy assessment. Results indicated that ADC for lung cancer was statistically significantly smaller than that for infectious nodules. The AUC values of ADC were 0.92 +/- 0.07, 0.82 +/- 0.09, and 0.79 +/- 0.10, for b-values of 600, 800, and 1000 s/mm2, respectively. We conclude that the DWI-MR imaging can differentiate lung cancer from infectious nodules accurately.
What problem does this paper attempt to address?