Wound regeneration is a complex process that is regulated by orchestrated mechanisms, influenced by chemical, cellular, and molecular factors1–2. The healing process begins at the time of injury and eventual maturation could continue for months or even years until the wound completely heals and is structurally and functionally similar to uninjured skin3. The four overlapping phases of wound healing are homeostatic, inflammatory, proliferative, and remodeling (Fig. 1). The homeostatic phase lasts a few hours producing a fibrin plug followed by an inflammatory phase, which can last between hours and days, during which aggregated platelets and cells release pro-inflammatory mediators4. The early inflammatory phase is succeeded by the proliferative phase lasting a few weeks during which macrophages and fibroblast cells invade the wound bed forming granulation tissue and active migration of the wound epithelial cells occurs5. The last phase of wound healing is the remodeling phase, which is characterized by proliferative cell apoptosis, adjustment of extracellular matrix (ECM), and replacement of type 3 with type 1 collagen in the dermis. It can last for weeks to years3.
During healing phases, a strong interaction between the nervous system and skin involving a variety of neuromodulators, cytokines, hormones, and other effector molecules has been reported6–7. The nervous system can be influenced both at the local and central levels by the stimuli at the skin and vice versa. The brain can alter skin function during the pathophysiological state and skin can modulate the nervous system by releasing a variety of neuropeptides8. The cutaneous nerves therefore positively affect all the stages of wound healing9. Numerous neuropeptides e.g. substance P (SP) released from cutaneous nerves have been reported to activate vital mechanisms during the inflammatory phase10. Similarly, neuropeptides released by cutaneous nerves influence the proliferation phase. They can promote the proliferation of fibroblasts, keratinocytes, and endothelial cells by stimulating DNA synthesis, can stimulate angiogenesis, support granulation tissue remodeling, and many more 11–15. The effect of innervation on the remodeling stage has also been studied in past. It has been demonstrated that a significantly higher number of nerve fibers correlate with normotrophic scars in comparison to hypertrophic scars during the remodeling phase16–17.Hence, the literature strongly suggests a regulatory role for skin nerves in wound healing and any impairment in skin innervation is one of the leading causes of occurrence of chronic wounds e.g., diabetic neuropathy with their foot ulcers and plegias with their sacral and trochanteric pressure sores18–19.
Previously, numerous studies have been conducted on quantifying skin innervation20–26. However, most such studies are not fully automated, have manual counts of IHC-stained structures that are prone to user errors and variation, require multiple observers, and are complex and labor-intensive. One such example is dendrite analysis involving manually tracing neurons using the simple neurite tracer plug-in of ImageJ software 21–22. While this semi-manual approach has been proven effective, it involves identifying the beginning and end points of dendrites and digitally drawing individual branch segments throughout the entire neuron, making it a labor-intensive and time-consuming process. Another example is semi-automated Sholl analysis for quantifying changes in the growth and differentiation of neurons and glia23. The method offers several advantages over conventional manual quantification, including faster analysis time and increased statistical sensitivity, however, has some limitations, such as reliance on manual input from the user, which introduces a risk of user error and variation impacting the accuracy and reliability of the results. Additionally, the semi-automated Sholl method is complex and time-consuming to set up initially, which could act as a barrier for researchers who do not have the technical expertise or resources to implement the method effectively. In an effort to quickly, objectively, and reproducibly quantify cutaneous innervation we developed a fully automated Matlab-assisted image analysis tool aided by powerful deep neural network, DnCNN, for pre-processing (de-noising) of the IHC-images. This network can detect and remove high-frequency image artifacts and increase image resolution, noise is minimized, resulting in higher quality images that can be more accurately analyzed. The DnCNN is particularly developed for image processing27, and has shown effectiveness in a wide range of applications, including medical imaging28.
Utilizing automated Matlab-assisted tool aided with DnCNN we quantified skin innervation during wound healing stages at days 3,7,10 and 15. The data show a positive correlation between the increase in nerve fiber density and re-epithelization.