Premium
Apparent diffusion coefficient measurement in glioma: Influence of region‐of‐interest determination methods on apparent diffusion coefficient values, interobserver variability, time efficiency, and diagnostic ability
Author(s) -
Han Xu,
Suo Shiteng,
Sun Yawen,
Zu Jinyan,
Qu Jianxun,
Zhou Yan,
Chen Zengai,
Xu Jianrong
Publication year - 2017
Publication title -
journal of magnetic resonance imaging
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.563
H-Index - 160
eISSN - 1522-2586
pISSN - 1053-1807
DOI - 10.1002/jmri.25405
Subject(s) - kurtosis , effective diffusion coefficient , nuclear medicine , standard deviation , intraclass correlation , percentile , region of interest , reproducibility , coefficient of variation , medicine , volume (thermodynamics) , receiver operating characteristic , correlation coefficient , skewness , mathematics , statistics , radiology , magnetic resonance imaging , physics , quantum mechanics
Purpose To compare four methods of region‐of‐interest (ROI) placement for apparent diffusion coefficient (ADC) measurements in distinguishing low‐grade gliomas (LGGs) from high‐grade gliomas (HGGs). Materials and Methods Two independent readers measured ADC parameters using four ROI methods (single‐slice [single‐round, five‐round and freehand] and whole‐volume) on 43 patients (20 LGGs, 23 HGGs) who had undergone 3.0 Tesla diffusion‐weighted imaging and time required for each method of ADC measurements was recorded. Intraclass correlation coefficients (ICCs) were used to assess interobserver variability of ADC measurements. Mean and minimum ADC values and time required were compared using paired Student's t‐tests. All ADC parameters (mean/minimum ADC values of three single‐slice methods, mean/minimum/standard deviation/skewness/kurtosis/the10 th and 25 th percentiles/median/maximum of whole‐volume method) were correlated with tumor grade (low versus high) by unpaired Student's t‐tests. Discriminative ability was determined by receiver operating characteristic curves. Results All ADC measurements except minimum, skewness, and kurtosis of whole‐volume ROI differed significantly between LGGs and HGGs (all P < 0.05). Mean ADC value of single‐round ROI had the highest effect size (0.72) and the greatest areas under the curve (0.872). Three single‐slice methods had good to excellent ICCs (0.67–0.89) and the whole‐volume method fair to excellent ICCs (0.32–0.96). Minimum ADC values differed significantly between whole‐volume and single‐round ROI ( P = 0.003) and, between whole‐volume and five‐round ROI ( P = 0.001). The whole‐volume method took significantly longer than all single‐slice methods (all P < 0.001). Conclusion ADC measurements are influenced by ROI determination methods. Whole‐volume histogram analysis did not yield better results than single‐slice methods and took longer. Mean ADC value derived from single‐round ROI is the most optimal parameter for differentiating LGGs from HGGs. Level of Evidence: 3 J. Magn. Reson. Imaging 2017;45:722–730.