Abstract
Fluorescence imaging in the near-infrared spectral range has unlocked new imaging possibilities in surgical guidance over the past decade. With the administration of fluorescent agents into the tissue and the use of highly sensitive fluorescence cameras, pathophysiological tissue sites that are invisible to human vision can be identified during a surgical operation. The advent of fluorescence agents that target specific molecular characteristics of disease has facilitated the identification and demarcation of cancerous tissue with greater sensitivity and accuracy over nonspecific fluorescence agents that circulate in the vascular system. Despite the advances in cameras and specificity of fluorescence agents, fluorescence imaging still faces challenges toward its clinical translation. The aim of this work is to propose novel solutions to overcome these challenges.The first challenge is the lack of calibration and standardization of fluorescence imaging systems. There are various fluores ...
Fluorescence imaging in the near-infrared spectral range has unlocked new imaging possibilities in surgical guidance over the past decade. With the administration of fluorescent agents into the tissue and the use of highly sensitive fluorescence cameras, pathophysiological tissue sites that are invisible to human vision can be identified during a surgical operation. The advent of fluorescence agents that target specific molecular characteristics of disease has facilitated the identification and demarcation of cancerous tissue with greater sensitivity and accuracy over nonspecific fluorescence agents that circulate in the vascular system. Despite the advances in cameras and specificity of fluorescence agents, fluorescence imaging still faces challenges toward its clinical translation. The aim of this work is to propose novel solutions to overcome these challenges.The first challenge is the lack of calibration and standardization of fluorescence imaging systems. There are various fluorescence imaging systems across different laboratories and clinics, and each system operates with a combination of parameters such as exposure time, working distance, gain, and magnification. The variety of interlaboratory systems and imaging parameters highlights the need to calibrate the system before the measurement to ensure the best performance and reproducibility of the imaging results for the same target. To tackle this challenge, we present the first and second generation of comprehensive solid phantoms that can simultaneously evaluate multiple system parameters. These parameters include camera sensitivity, fluorescence intensity variations as a function of optical properties and depth, illumination homogeneity, optical and fluorescence resolution, and cross talk from excitation light and parasitic illumination leaking into the fluorescence channel. Moreover, these phantoms can be employed for the correction of inhomogeneous illumination.The second challenge of fluorescence imaging is the distortion of the fluorescence signal due to tissue scattering and absorption. Photon diffusion in tissue distorts image quality. Poor image resolution and contrast make tumor delineation difficult, leading to ambiguity in the interpretation of the image. Thus, incomplete tumor resection or excessive healthy tissue removal can occur, both of which negatively affect patients’ quality of life. In this thesis, a new method for fluorescence imaging enhancement is proposed that takes into account tissue heterogeneity, unlike traditional methods that presume tissue homogeneity. We introduce the concept of scanning the tissue with a point-like beam to acquire spatially variant impulse responses (kernels) that are dependent on the local optical properties. Subsequently, these kernels are used in a fast, spatially variant deconvolution scheme to revert the degradation of image quality.Experimental measurements from phantoms and ex vivo mice were used, and improvements were shown and validated.Finally, the correction of intensity values in fluorescence imaging is discussed in the context of a method that could be applied as an extension of the spatially variant deconvolution method. Overall, a comprehensive framework comprising all correction methods described in this thesis is suggested, and a future outlook is described.
show more