The gray values of different spinal cord elements including the maximal compressive level, compression-free level and the dural sac were measured, and the GVRs of maximal compressive level/compression free level and maximal compressive level/dural sac were calculated. The results of intra- and inter-observer reliability prove that the GVR of maximal compressive level/dural sac is more consistent for both inter- and intra-observer reliability.
In DCM, owing to the chronic compression, many pathological changes occur in the spinal cord, including ischemia, edema, neuron apoptosis, spinal cord atrophy and cystic necrosis . Consistently, these pathological changes are always reflected as ISI on T2W MRI. This characteristic of MRI may cause the deviation when the signal intensity of T2W MRI ISI is used to evaluate the status and predict the neurological recovery of DCM. The parenchymatous degeneration and cystic necrosis indicate different severities of spinal cord impairment [16-17]. The cystic necrosis of the spinal cord can also be reflected as T1W hypointensity, but the positive rate of T1W hypointensity is not high in DCM. This characteristic limits the application of T1W MRI in DCM .
The echogenicity of ultrasonography is based on the different densities of tissues (Walker 2004). For DCM, the chronic compression derived pathological changes may cause different densities of the spinal cord . Then the different densities of the spinal cord are reflected as different levels of echogenicity on IOUS. Previously, we quantified the spinal cord hyperechogenicity as gray value, and revealed the negative correlation between hyperechogenicity intensity and postoperative neurological recovery [7-8]. Besides, ultrasound has a significant image difference between liquid and parenchymal tissues, and thus can easily identify the pathology of cystic and parenchymatous changes. All those features of ultrasonography are unfulfilled on the clinical application of MRI or CT, and determine the application value and irreplaceability of IOUS.
According to the ultrasonography principles, differences in the density of adjacent tissues result in different echoes and are reflected as different gray values on ultrasound images . The spinal cord hyperechogenicity in DCM patients results from the chronic compression subjected to the static and dynamic mechanical forces acting on the spinal cord [21-22]. The chronic compression leads to fibrin deposition, and even fibrosis in the compression region, accompanied with the loss of nerve cells, proliferative fibroblasts and capillary endothelial cells [23-24]. Eventually, the uneven density of the spinal cord will occur and appears as hyperechogenicity on IOUS [25-26]. As shown in the typical case, spinal ultrasound can still detect signs that are not observed by MRI. When MRI presents only high signal changes, ultrasound can still return additional signs, such as syringomyelia signs. The intensity of spinal cord hyperechogenicity is considered as a potential predictor of neurological recovery in DCM after decompression. Therefore, the reliabilities of two methods to evaluate the hyperechoic intensity of DCM spinal ultrasound in the same observer and between observers were compared to find out a more reliable evaluation method.
The method used to calculate GVR by comparing the gray values of spinal cord hyperechogenisity and dural sac (GVR-B) is highly consistent for both inter- and intra-observer reliabilities, which are obviously higher than those of GVR-A. We attributed this improvement to the selection of ROIs, which, for intraoperative ultrasonic signal measurement, may be an important influence factor on GVR. Gcompression (gray value of circle 1) has a one-to-one correspondence with Gsac (gray value of circle 3), but Gnorml (gray value of circle 2) does not have such correspondence with Gsac. The principle of ROI selection and the GVR-based calculation method reflect that Gnorml is more random and highly variable, which also result in lower reliability of GVR-A. Conversely, when GVR-B was used to calculate the same Gcompression, more significant intra- and inter-observer reliability was obtained, indicating its consistency is better with the dura mater as a reference. Based on this finding, we suggest to use method B to evaluate the intensity of spinal cord hyperechogenicity in DCM.
There are several main limitations. First, as an exploratory prospective study, the sample size was relatively small. Second, no patient-based outcomes were evaluated. Moreover, the lack of multicenter study can lower the reliability of our statistical analysis.