An automated digital fluorescence imaging system of tumor margins using clustering-based image thresholding

Abstract

An optical system for efficient fluorescence imaging of cancer margins aiming at enhanced discrimination of the tumor area from the surrounding normal tissue, is presented. Fluorescence imaging was used to acquire grayscale images of brain tumor samples of 10 μm slice thickness. The tumor cells are characterized as Gli36Δ5 cells expressing Green Fluorescent Protein (GFP). An image processing technique involving the clustering-based concept of Otsu segmentation was applied to enhance the contrast and difference between the tumor and the rest of the tissue for improved visualization of tumor margins. Edge detection was performed on these processed images to mark the boundaries of the tumor area. The fluorescence imaging results depict clear demarcation of tumor boundary and a substantial improvement of the contrast, post processing.