Tumor border delineation in basal cell carcinoma using Fourier analysis

Basal cell carcinoma (BCC) is the most common malignancy in Caucasians. Surgical excision is the gold standard for treating BCC; however, poorly defined borders often challenge complete removal. Therefore, there is a need for efficient imaging techniques to enable visual and quantitative evaluation of BCC tumor borders before and during surgery. In this study, we aimed to develop a mathematical algorithm for tumor border delineation using nonlinear optical microscopy. Two-photon excitation auto-fluorescence (2PEF) and second-harmonic generation (SHG) images were acquired from nodular, micronodular, and infiltrative BCC skin sections. Fast Fourier transform (FFT) analysis was executed on two-dimensional SHG images, to characterize the collagen fibre structure. Based on differences in the collagen fiber network between healthy tissue and the tumor microenvironment, a heatmap was generated to visualize the predicted tumor borders of BCC skin sections. Spatial grid resolution proved to be a critical parameter for FFT analysis, with the highest border delineation accuracy achieved using a unit cell size of 0.14 × 0.14 mm². Smaller unit cells led to the omission of small tumor nests, whereas larger unit cells reduced sensitivity by including normal anatomical structures, such as hair follicles, within the predicted tumor border.