Year: 2026
Type: preprint
Abstract: Raman spectroscopy is often reported to stage colorectal neoplasia, but validation is usually done at the level of spectra rather than subjects. We reanalysed public spectra from a p53 deletion mouse model (47 mice; 485 spectra of normal mucosa, hyperplasia, adenoma and carcinoma) with leave-one-mouse-out validation, covariate tests and within-scan paired contrasts. Lesions were separated from normal mucosa with a balanced accuracy of 0.79 (95% CI 0.74–0.84). Adenoma and carcinoma were not separated (0.48), although spectrumlevel validation suggested 0.75. Within animals, a progression score ordered normal mucosa, hyperplasia and adenoma, and placed adenoma above adjacent normal tissue in all 20 paired mice, whereas carcinoma did not exceed adenoma. Genotype and sex left no trace in normal mucosa, but spectra drifted with acquisition order. In this model, Raman spectra track neoplastic change up to adenoma, while the apparent malignancy signal reflects animal identity.
Source: ChemRxiv
Author: Seyedeh Rahemeh Amiri
Institution: University at Albany, State University of New York
Language: English
Subfield: Biophysics
Domain: Life Sciences
Open Access status: green
OpenAlex ID: W7225895034
