Mathematical Method for Enhancing Biochemical information in Micro-FTIR Images of Histopathological Slides

Abstract

Early and accurate cancer diagnosis improves prognosis and reduces treatment invasiveness and costs. Histopathological biopsy analysis is crucial but depends on pathologist expertise. Biochemical information from samples can enable earlier diagnosis due to tissue changes before visible alterations. Optical techniques, particularly micro-infrared spectroscopy (micro-FTIR), are promising for this. However, artifacts from contaminants and baseline distortions can affect data quality. This work develops a method using principal component regression (PCR) to remove these artifacts. Applying this method to 192 micro-FTIR images of thyroid and mouth tissues achieved over 90% correlation with traditional hematoxylin-eosin images, indicating effective artifact removal and accurate results.