In vivo Laser‐induced Fluorescence Imaging of a Rat Pancreatic Cancer with Pheophorbide‐a

Publisher Website
Google Scholar
Abstract
Laser‐induced fluorescence (LIF) of pheophorbide‐a (Ph‐a) was used for imaging of a rat pancreatic tumor. Using a dimensionless function (the ratio of Ph‐a fluorescence by bluish autofluorescence), the fluorescence contrasts between excised tumors and their paired pancreas were investigated up to 48 h after a 9 mg kg‐1 Ph‐a intravenous administration. Among five tested excitation wavelengths, 355 and 610 nm excitations gave the best distinctive contrasts, both 48 h after dye injection. The LIF imaging of six intrapancreatic tumors and six healthy pancreas was carried out in vivo using two laser excitations: 355 nm (Nd:YAG + tripling) for bluish autofluorescence and 610 nm (rhodamine 6G dye) for reddish autofluorescence and dye emission. Images were recorded through bandpass filters at 470 and 640 nm (autofluorescence) and at 680 nm (dye + autofluorescence) with an intensified charged‐coupled device camera. Autofluorescence as Ph‐a fluorescence images did not allow accurate LIF diagnosis of pancreatic carcinoma. An image processing, including for each pixel a computed division of Ph‐a fluorescence (after subtraction of reddish autofluorescence) by bluish autofluorescence intensity generated poorly contrasted tumor images in five of six and false tumor localization in one of three of the tumor‐bearing pancreas. A fitting of the digital 640 nm autofluorescence up to the mean 680 nm fluorescence intensity in pancreas prior to subtraction allowed a safe diagnosis to be made with well‐contrasted tumor images. To assess automation ability of the processing, a same fitting coefficient (mean of individual values) was applied. In this way, false‐negative (one of six) and false‐positive (two of six) images were present in tumor‐bearing animals as false‐positive in one‐half of the controls. A successful standardized procedure was then applied with a normalization of 640 and 680 nm pancreas intensities to a same set threshold prior processing. In opposition to thin‐layered hollow organs, such as bronchial tube or digestive tract, LIF imaging of carcinoma inserted in a compact organ is exhausting. The use of a dye excitable in the red wavelength range (610 nm for Ph‐a) may partly solve this problem, rendering LIF imaging more accurate and potentially automated.