Impact of different ROI selection methods on histogram analysis parameters based on enhanced T1WI of high-grade gliomas / 中国医学影像技术

ABSTRACT

Objective: To investigate the repeatability of different ROI seletion methods on histogram analysis parameters of enhanced T1-weighted images of high-grade gliomas (HGGs). Methods: Four ROI selecting methods (small ROI method, contour ROI method, threshold ROI method and volume ROI method) were used to measure the histogram parameters on enhanced T1WI of 45 HGGs patients. The consistency between 2 observers and the differences of histogram parameters were evaluated. Results: The consistent of the mean, standard deviation and kurtosis values measured using volume, contour and threshold ROI methods between 2 observers were good (ICC≥0.80), and of the skewness values were moderate to good (ICC=0.73-0.90). The consistent of histogram characteristic parameters measured using small ROI method between 2 observers were low to medium (ICC=0.30-0.69). Histogram characteristic parameters measured with volume ROI method and contour ROI method had no significant difference (all P>0.05), but the average value, skewness value and kurtosis value measured with threshold ROI method were higher than those with volume and contour ROI method (all P<0.05), and the standard deviation of threshold ROI method was lower than that of volume ROI method and contour ROI method (both P<0.05). Standard deviation measured with small ROI method was lower than that of volume ROI method and contour ROI method (all P<0.05). Conclusion: The consistency of histogram characteristic parameters of HGGs on enhanced T1WI measured with volume ROI method is the highest, and different ROI selection methods have some impact on the measurement results of histogram characteristic parameters of enhanced T1WI.