Design, calibration, and preliminary testing of an additive-manufactured Raman spectrometer for skin cancer screening

Skin cancer is one of the most common forms of malignant neoplasms worldwide. Currently, the primary method for diagnosing skin cancer is invasive histopathological examination. A non-invasive alternative is Raman spectroscopy, which detects the inelastic scattering of light by tissue. The objective of this project was to build, calibrate, and test a portable, low-cost Raman spectrometer. The construction involved the precise arrangement of optical components within an additive-manufactured housing, configured as an LGL refractive spectrometer. Calibration was performed using a paracetamol sample, as this substance exhibits a large number of characteristic peaks in the same regions of the skin under study. To test the equipment, a preliminary study was conducted by collecting Raman signals from regions with and without lesions on a participant's skin to determine whether there was any significant difference in the information derived from these samples. Features were extracted from the collected signals and fed into an SVM classifier, achieving accuracies of 89.3%, 85.2% and 83.2%, respectively, in binary distinguishing between keratosis lesion and healthy skin, lesion suspected of malignancy and keratosis lesion, and lesion suspected of malignancy and healthy skin. A preliminary validation was performed for the pigmented lesions classification model using signals from eight participants, achieving an accuracy of 75%. Although further study with a larger sample size is needed, the results suggest that the device is capable of capturing intrinsic skin information (both with and without lesions).