Year: 2026
Type: dissertation
Abstract: Early detection of liver fibrosis is critical to reducing the global disease burden associated with this condition.Near-infrared (NIR) spectroscopy (NIRS) has the potential to provide a costeffective and rapid label-free assessment of underlying biochemical variations in liver tissue.As a non-invasive tool, it may complement existing clinical imaging modalities, such as ultrasound elastography and magnetic resonance elastography.By examining tissue absorption, scattering, and water content, NIRS can provide relevant information about the structural patterns that can aid disease stratification.This study evaluates the relationship between near-infrared spectral features and histopathological assessment derived from digital whole-slide images through the development of predictive models, including both regression and classification multivariate analyses.Preliminary results obtained from a sample size of twelve (n = 12) from a single porcine subject indicate that near-infrared spectroscopy combined with machine learning can achieve promising levels of predictive accuracy.The regression analysis achieved R 2 values of 0.84 and 0.64 on the test sets for masked percentage and target max mean ratio, respectively, whereas of the corresponding classification accuracies were 1.00 and 0.90.These findings suggest near-infrared spectroscopy may serve as a valuable non-invasive adjunct tool for the characterization of liver fibrosis.Ultimately, this approach may support prognosis by providing insight into underlying tissue biochemical changes.
Author: Oyewo, Adenike
Institution: University of Eastern Finland
Language: English
Subfield: Radiology, Nuclear Medicine and Imaging
Field: Medicine
Domain: Health Sciences
Open Access status: green
OpenAlex ID: W7215386917
