Abstract
Collagen is a key protein in mammals maintaining structural integrity within tissues. A failure in fibrillar collagen reorganization may induce cancer or fibrosis formation, such as in spinal cord injury (SCI), where the healing process after the initial trauma leads to the formation of scar tissue, which includes fibrosis. There is no current treatment targeting the fibrotic process directly, thus a better understanding of collagen properties can help to apprehend malignant states. Characterization of collagen fibers has been widely explored on second-harmonic generation (SHG) images, due to the label-free nature of the SHG imaging technique. Various fibers’ extraction methods have been performed such as curvelet transform (CT) implemented in the open-source software CurveAlign. However, when it comes to investigating collagen fibers that are still under reorganization – undergrowth fibers – as observed in SCI, the CT method becomes complex to tune for non-advanced users in order to properly segment the fibers. To improve collagen detection in the case of undergrowth fibers, we propose a methodology based on the fingerprint enhancement (FP-E) algorithm that requires fewer user input parameters and is less time-consuming. Our method was extensively tested on SHG data from injured spinal cord samples. We obtained metrics that depicted changes in collagen organization over time, particularly a significant increase in fiber density, demonstrating the FP-E algorithm was properly adapted to extract the fibers and was a promising tool to address the evolution of collagen properties after SCI. Besides, the tuning of the method is simpler compared to commonly used software, and the combination with further characterization of the extracted fibers could lead to consider fibrillar collagen as a biomarker in diseases where fibers are under development.