Imaging margins of skin tumors using laser-induced breakdown spectroscopy and machine learning

Nowadays, laser-based techniques play a significant role in medicine, mainly in the ophthalmology, dermatology, and surgical fields. So far, they have presented mostly therapeutic applications, although they have considerable potential for diagnostic approaches. In our study, we focused on the application of laser-based spectroscopy in skin cancer assessment. Recently, lengthy and demanding pathological investigation has been improved with modern techniques of machine learning and analytical chemistry where elemental analysis provides further insight into the investigated phenomenon. This article deals with the complementarity of Laser-Induced Breakdown Spectroscopy (LIBS) with standard histopathology. This includes discussion on sample preparation and feasibility to perform 3D imaging of a tumor. Typical skin tumors were selected for LIBS analysis, namely cutaneous malignant melanoma, squamous cell carcinoma and the most common skin tumor basal cell carcinoma, and a benign tumor was represented by hemangioma. The imaging of biotic elements (Mg, Ca, Na, and K) provides the elemental distribution within the tissue. The elemental images were correlated with the tumor progression and its margins, as well as with the difference between healthy and tumorous tissues and the results were compared with other studies covering this topic of interest. Finally, self-organizing maps were trained and used with a k-means algorithm to cluster various matrices within the tumorous tissue and to demonstrate the potential of machine learning for processing of LIBS data.

Automated summary

Laser-based techniques are widely utilized in medicine, with a focus on therapeutic applications and potential for diagnostic approaches. The study explores the application of laser-based spectroscopy in skin cancer assessment, utilizing modern techniques such as machine learning and analytical chemistry. The use of Laser-Induced Breakdown Spectroscopy (LIBS) in combination with standard histopathology shows promise in providing insights into tumor progression and tissue differences, with the potential of machine learning for processing LIBS data.