Clinical Report: Reading Skin Cancer’s Molecular Fingerprint

Overview

In a preliminary ex vivo study, handheld Raman spectroscopy paired with machine-learning models classified normal skin, basal cell carcinoma (BCC), and squamous cell carcinoma (SCC) with accuracy up to approximately 84%. Models distinguished normal tissue from cancer more successfully than they separated BCC from SCC, and the study evaluated analytical performance rather than clinical diagnostic accuracy.

Background

BCC and SCC can resemble benign or precancerous lesions, while biopsy followed by microscopic examination remains the standard diagnostic approach. Raman spectroscopy uses laser light to generate spectra reflecting tissue chemistry, and machine-learning methods can identify patterns within those spectra. Researchers are investigating noninvasive approaches that might help assess skin lesions, but the supplied guideline context states that histologic confirmation remains necessary for suspected cutaneous SCC and that evidence for noninvasive imaging is insufficient for routine diagnostic use.

Data Highlights

Key Findings

• The mobile Raman system used a handheld probe and a 785-nm laser to examine removed normal skin, BCC, and SCC tissue.
• K-nearest neighbors and support vector machine models had the highest reported overall accuracy, at approximately 84%; the support vector machine sensitivity was 78.7% and specificity was 88.6%.
• The shallow neural network had 80.8% accuracy and the highest reported AUC, 0.910.
• Partial least squares discriminant analysis had 95.5% specificity but 66.5% sensitivity, while principal component analysis with quadratic discriminant analysis had sensitivity and specificity of 77.8% and 77.7%, respectively.
• Normal skin was more readily distinguished from cancerous tissue than BCC was from SCC; overlapping molecular patterns between the cancers contributed to classification errors.
• Cancer samples generally showed stronger protein-related signals, whereas normal skin showed stronger lipid-related signals. The study used ex vivo specimens and assessed analytical performance, not diagnostic accuracy in patients.

Clinical Implications

The reported results do not establish Raman spectroscopy as a replacement for biopsy and histopathology: testing was performed on removed tissue, the sample set was small, and performance remained below the greater-than-90% levels the researchers considered desirable for clinical use. The supplied guideline context continues to identify biopsy with histologic confirmation as the diagnostic standard for suspected cutaneous SCC.

Conclusion

Handheld Raman spectroscopy combined with machine learning identified biochemical patterns associated with normal skin, BCC, and SCC in a preliminary tissue study. Its clinical diagnostic performance remains unestablished in the supplied evidence.

Related Resources & Content

1. The Analytical Scientist, 2026 — Reading Skin Cancer’s Molecular Fingerprint
2. European consensus-based interdisciplinary guideline for invasive cutaneous squamous cell carcinoma, Part 1: Diagnostics and prevention, Update 2026
3. American Academy of Dermatology — Squamous cell carcinoma: From symptoms to treatments
4. the pathologist — Molecular Clues to Skin Cancer
5. ASCO Publications — Deep sequencing of metastatic cutaneous basal cell and squamous cell carcinomas to reveal distinctive genomic profiles and new routes to targeted therapies.
6. ASCO Publications — Prognostic insights in skin melanoma: A study of differential diagnostic tissue markers.
7. Reading Skin Cancer’s Molecular Fingerprint
8. European consensus-based interdisciplinary guideline for invasive cutaneous squamous cell carcinoma, Part 1: Diagnostics and prevention — Update 2026
9. Squamous cell carcinoma: From symptoms to treatments
10. NCCN National Comprehensive Cancer Network®NCCN Cl
11. Stages and Risk Groups of Basal Cell Carcinoma | American Cancer Society
12. Separating Skin Cancer From Normal Tissue Using Handheld Raman Spectroscopy | Spectroscopy Online
13. Noninvasive Multimodal Imaging and Its Role in Diagnosing Skin Lesions in Dermatology: A Systematic Review and Meta-Analysis.
14. Performance of Artificial Intelligence in Skin Cancer Detection: An Umbrella Review of Systematic Reviews and Meta‐Analyses - Karimzadhagh - 2026 - International Journal of Dermatology - Wiley Online Library