DTI is a promising technique for determining the responsible lesion of cervical radiculopathy [22], but the selection and delineation of the ROI influence the results [20, 21]. Unlike normal nerve roots, diseased nerve roots are compressed in different directions, resulting in different degrees of edema inside the nerve root (nerve fibers), so the signals of the nerve root cross-section are heterogeneous. Different drawing methods cover different signal areas which may lead to different DTI values. Therefore, it is of clinical significance to discuss the ROI sketching method of diseased nerve roots.
This study aimed to explore the impact of different ROI sketching methods on the repeatability and consistency of DTI measurement values in patients with cervical spondylotic radiculopathy. The results suggest that for the delineation of DTI ROI in patients with cervical spondylotic radiculopathy, the free-hand and single largest circle methods were the most consistent methods.
In this study, we found that when measuring FA and ADC in patients with cervical spondylotic radiculopathy, intra- and interobserver variabilities were dependent upon the methods of ROI delineation. FA and ADC measurements obtained by the largest circle and the free-hand methods were more reproducible than those obtained from the rectangle or four small rounds measurements. Ma et al. [23] found that the ROI size had a considerable influence on the ADC measurements of PDACs and suggested that the largest round ROI yielded the best intra- and interobserver reproducibility. At the same time, Jafari-Khouzani et al. [24] believe that increasing the ROI size can reduce the variance of the FA and ADC values. This is consistent with our results. We think that a larger circle will yield high repeatability and consistency. Of course, the nerve root cross-section is mostly a circle. A round ROI covering the cross-sectional nerve area will include most of the pixels, leading to the highest homogeneity. Moreover, the maximum circular ROI is easier to operate in practice with less time and better controllability, which greatly reduces the measurement errors caused by the circle sketched beyond the actual boundary of the nerve root during the actual operation. Therefore, the ICCs are high. The ROIs area of the free-hand method is greater than the maximum roundness method, nevertheless, due to the time-consuming drawing process, the contour method requires the operator to continuously judge and identify the actual boundary of the nerve roots. Therefore, the result is greatly affected by the operator's subjective factors.
The quadrilateral and four small circles ROI methods have high central variability and lower ICCs. It takes longer to include fewer pixels. Nogueira et al. [20] report that small ROIs show high ADC reproducibility in the DTI diagnosis of breast lesions. Inoue et al. [25, 26] showed that the ROI shape has no marked influence on the ICC in endometrial carcinoma. These are contrary to our conclusions and may be due to the different nature of the subjects and lesions. Moreover, an important advantage of using the largest round ROIs is that its placement is much less time-consuming compared to having to delimit the whole slice. Lambregts et al. [18] and Ma et al. [23] showed that the ROI has a considerable influence on tumor DTI values. Sun et al. [27] showed that the ADC and FA values derived from outline ROIs are higher than those from round ROIs. Inconsistent with those previous studies, we found that there were no significant differences in the FA values and ADC values of the four methods, which may be related to the small sample size. Moreover, FA and ADC values may be correlated with age, sex, and BMI. The cross-sectional areas of the nerve roots included in the four methods were similar, that is, the number of axons did not differ significantly, so there was no significant difference in the measured values. The obtained FA measurements of four ROIs are lower than those reported in the previous literature [17, 22], which may be because included patients had more severe root compression than the reported patients. In this present study, the mean FA values in entrapped nerve roots were lower than they were in intact nerve roots, indicating that diffusion in the tissue had become more isotropic because of edema, in which fluid is trapped in the tissue, creating an isotropic environment and a reduction in FA.
Of course, there are limitations in this study. First, in order to obtain the typical diseased nerve root, we adopted very strict case inclusion criteria, leading to small sample size. Second, due to the limited sample size, we did not group the patients according to sex, age, body mass index, and other factors, and did not study the possible influence of different factors on the results. Therefore, in future studies, we will combine multiple centers to screen and include typical cases, and assess the influence of relevent factors on the results.