3D reconstruction of peripheral nerves is one important tool to understand the nerve internal information and functional basis, and it is gradually becoming a hot spot of the peripheral nerve research field [3, 14, 15, 24,25,26,27,28]. 2D acetylcholinesterase staining is a key technique for 3D peripheral nerve reconstruction, which affects the precision and accuracy of the 3D reconstruction images. Ideal 2D staining section images are those that are able to explicitly display the nerve fascicles and the nerve fibers of different functions with clear textural properties, which can facilitate subsequent image recognition, partition and 3D reconstruction. At present, Karnovsky–Roots staining is the commonly employed 2D staining method. However, this staining technique can not explicitly display the nerve fascicles and textural properties of the motor and sensory nerve fibers. Consequently, using the resulting images, it is difficult to establish an accurate 3D model due to the poor effect of image recognition. We have recently explored the 3D reconstruction of the peripheral nerves based on traditional 2D staining method [12, 13, 24, 29]. Improving image texture features is still a difficult problem to solve. Myelin may be a useful structure. In theory, both sensory and motor nerve fibers are myelinated nerve fibers. If the myelin can be stained on the basis of Karnovsky–Roots staining method, the texture characteristics of the inner structure of the nerve may be improved. There are some methods to show myelin, for example, toluidine blue and Ponceau 2R have been used as effective staining methods for the study of myelin morphology in research and clinical work [30]. While these staining methods offer more options for improving image quality, no one was used in research of 3D reconstruction. The use of two staining methods either separately or together based on Karnovsky–Roots staining method to show the myelin sheath has not been reported in the current literature. This study found that counterstaining can improve the image quality. In the current study, the K-B-2R staining method was developed by modifying the conventional acetylcholinesterase staining. This K-B-2R new staining method has been shown to be able to display the axon and myelin sheath simultaneously, which significantly improves the display quality of the nerve fascicle regions, and the motor and the sensory nerve fiber textures. The resulting images of the peripheral nerve have led to a high degree of recognition, image partition and accomplishment of the 3D reconstruction (Figs. 8, 9d and 10d). Importantly, the reconstruction cycle was enormously shortened and the precision of 3D reconstruction model was significantly enhanced. In addition, the 3D printing technology was applied to create the 3D digital model of nerve fascicle (Figs. 8c and 9e). Thus, this new staining technique can facilitate 3D reconstruction and creation of the 3D digital model, which suggests that this new technique can facilitate to rebuild and repair the nerve fascicles when it is used in conjunction with the 3D reconstruction and 3D printing technologies.
Traditional staining method of peripheral nerve sections has several limitations [9, 25]. Firstly, as the sections of nerve specimens are several microns in thickness, it is difficult to successfully prepare adjacent sections in a large number, which causes missing data and reduces the accuracy of images. Secondly, the section’s partial loss during the staining process is different for each section, and the accumulation of loss error during image mergence is likely to lead to distortion of the merged image.
Our previous works [12, 13, 29] have demonstrated that the simultaneous display of both axon and myelin sheath of motor and sensory nerves can avoid the incidence of such errors during image mergence. While the axons and myelin sheaths of motor and sensory nerves are the common parameters for observation and analyses of the physiological pathological states of nerve fibers, various reagents and equipment can display the morphology of axons and myelin sheaths. As the myelin sheath can be explicitly displayed by toluidine blue and ponceau 2R staining, 2D K-B-2R staining method was proposed in the study. The functions, properties and distributions of the nerve fibers can be determined by the morphology of both axon and myelin sheath.
In the current study, 2D K-B-2R staining was found to be able to determine the properties and distributions of nerve fibers. Acetylcholinesterase-positive sites are the unique signs of Karnovsky–Roots staining as also shown in the current study (Fig. 5a), and the changes in the distribution characteristics or quantities of acetylcholinesterase-positive sites might affect the precision of 3D reconstruction. Herein, no evident changes were documented in the distribution characteristics or quantities of acetylcholinesterase-positive sites after K-B-2R staining (Fig. 5c), with the results of having the same positions of the positive sites on the identical specimen section as with Karnovsky–Roots staining. In addition, no statistically significant differences were observed in the quantities of acetylcholinesterase-positive sites identified by the three staining methods. Furthermore, during K-B-2R staining, the myelin sheath staining contributed to the exclusion of the possibility of non-specific dye sediment (Table 1). These results suggest that K-B-2R staining does not significantly affect the Karnovsky–Roots staining results, but it can additionally determine the properties and the distribution sites of the nerve fibers.
In addition, K-B-2R staining is favorable for the subsequent section image stitching step of 3D reconstruction. In Karnovsky–Roots stained images, the staining color is plain. The brown spots formed by acetylcholinesterase-positive sites (Figs. 1 and 5a) are the primary texture signs for image mergence. While approximately 20–30% of overlap is required between adjacent images, even 50% overlap is sometimes required for certain regions. As illustrated in Figs. 3 and 5c and Table 1, K-B-2R staining could explicitly display the nerve fascicle regions and the myelin sheath structures of the nerve fibers. The connective tissues were obviously stained with clear edges, which could be used for image mergence with abundant textures, allowing adjacent images with 15–20% overlap being merged into a panorama image. Due to this connective tissue staining, the K-B-2R staining method is advantageous for nerve specimens with large cross section areas. It not only significantly decreases the frequency of image partition and photography, reduces the workload of image mergence, enhances working efficiency, but also minimizes the accumulated error of manual operation and enhances the precision of image mergence results.
In the current study, the 2D K-B-2R staining method was found to be more suitable for image partition of nerve fascicle regions. Comparisons of the automatic image partition of the nerve fascicle regions from each of the three methods (as performed using the same algorithm method) (Fig. 5) show that K-B-2R staining obviously has significant advantages in displaying nerve textures. Firstly, by comparing the acquired nerve fascicle edges and the actual edges, the acquired nerve fascicle edges revealed in Fig. 5 in Karnovsky–Roots stained images were inconsistent with the actual edges, which required manual processing (Fig. 5a). Compared with Karnovsky–Roots stained images, the nerve fascicle edges acquired in K-B stained images were more satisfied; however, they still required manual processing and adjustment, and the non-specific staining among nerve fascicles was increased, leading to more frequent wrong partition. On the other hand, the nerve fascicle edges acquired in K-B-2R stained images were the most consistent with the actual situations. Since the texture properties of nerve fascicle regions and nerve fascicles significantly differed, despite the existence of some non-specifically stained spots, the probability of wrong image partition was enormously reduced in K-B-2R stained images.
Furthermore, the obtained images by this method required less time for slight manual processing and adjustment (Fig. 5). In Karnovsky–Roots and K-B stained images, evidently wrong partitioning occurred in the acquired nerve fascicle edges, the operation time was long (Table 1), and the partitioning required multiple manual interventions. In contrast, the partition results obtained by the K-B-2R staining were almost identical to the actual nerve fascicle edges, which required few manual interventions and shorter operation time. Thus, the 2D K-B-2R staining method proposed in the current study is more advantageous for image recognition compared with alternative approaches.
The current study shows that the K-B-2R staining method enabled staining of the connective tissues within the nerves especially the nerve fascicle regions, revealing the ring-shaped perineurial structures (Fig. 5c), which significantly contribute to images of nerve fascicle regions. In addition, the myelin sheath edges were explicitly displayed with this staining method, with each myelin sheath corresponding to each nerve fiber, and with a large quantity of myelin sheath edges aggregating in the nerve fascicle regions. These features significantly differed from those of the connective tissues among nerve fascicles. Among the axons wrapped by myelin sheaths, the motor nerve fibers were stained, whereas the sensory nerve fibers were not stained. Hence, with this staining method, it is more convenient to distinguish the nerve fascicles and the nerve fibers of different functions and properties, which contributes to image partition.
Data from the current study suggest that the 2D K-B-2R staining method is applicable for batch staining. Reconstructing and displaying the intact peripheral nervous system will require preparing and staining of an extremely large quantity of sections of several microns in thickness without compromising efficiency and quality. The 2D K-B-2R staining method established in the current study has been shown to cause little specimen loss and yet produce stable staining outcomes. Among the 30 sections of median nerve, 27 intact sections were stained and shown to have a consistent textural property (Fig. 11a). In the repeated experiments, among the 240 median nerve sections stained, 226 sections were found intact with a consistent textural property (Fig. 11b). Respectively from the total 300 sections of distal nerve fascicles of the median nerve, 60 sections of the muscle nerve branch, and 60 sections for the sural nerve, 268, 63 and 61 intact sections were obtained with a complete data set following K-B-2R staining. These data suggest a low section loss/damage rate and a high success rate with a complete data set being obtained by the staining method.
The current study compared the staining qualities among the three different methods. Karnovsky–Roots stained sections revealed the nerve fascicle regions being circle-like structures consisting of brown spots, and with no explicit and continuous nerve fascicle membrane structures being noted. A slight quantity of connective tissues among nerve fascicles was stained (Fig. 11a). No transitional color was noted in between the aggregation regions of sensory and motor nerve fibers and among the typical spots of nerve fibers of different properties (Fig. 11b). After K-B staining, nerve fascicle regions were in pale blue color, with nerve fascicle membrane structure being revealed. However, non-specific staining of the connective tissues among nerve fascicles was increased (Fig. 11a). K-B-2R staining revealed clear nerve fascicle membrane structures, and connective tissues among nerve fascicles were evenly stained and the myelin sheath of nerve fascicles was stained light red. The staining color of nerve fascicle regions and surrounding tissues significantly differed. The blue spots resulted from remnant pigment sediment (Fig. 11a). Discontinuous staining was noted in the connective tissues among nerve fascicles, epineurium and perineurium (Fig. 11a). However, the distribution characteristics of acetylcholinesterase-positive sites were consistent between K-B staining and Karnovsky–Roots staining. Perineurial and epineurial structures were seen at the edges of nerve fascicles and nerve. The acetylcholinesterase-positive sites were stained as black. The typical spots of motor nerve fibers were enveloped by the ring-shaped myelin sheath structures, and the typical spots of sensory nerve fibers were distributed in between the ring-shaped myelin sheath structures (Fig. 11b).