Model-based Approaches to Diffuse Optical Imaging and Dosimetry

Sammanfattning: The work within this thesis investigates Photodynamic therapy and Fluorescence imaging for therapeutics and visualization of deeply embedded lesions. Photodynamic therapy (PDT) is a cancer treatment modality that can eradicate tumors when light, a photosensible drug and oxygen are present. In recent decades there has been an interest in adapting this modality to deep-seated solid tumors. This requires the use of interstitially placed fibers for light delivery. The technique has been shown to be safe with minor complications but the treatment outcome show substantial inter- and intra-patient variations. This is inherited from the intrinsically complex interactions between the three components. In order to find a remedy to these problems dosimetry is needed. PDT-dosimetry adheres to the determination of parameters such as treatment time and light source positions in order to induce treatment response in the target volume, while sparing healthy tissue. In this thesis a dosimetry scheme for interstitial PDT (IPDT) of prostate cancer has been developed. It was implemented on an instrument utilizing 18 optical fibers for light delivery and monitoring. Measurements are performed before, during and after PDT in order to evaluate the optical properties. The calculation of a light dose, based on these properties, allows patient-specific and realtime treatment dosimetry. Experimental validation of the instrument and dosimetric tools confirmed the capability to tailor a light dose to a specific target volume as well as compensate for potential variations in the light propagation. The same rationale was applied in a clinical trial incorporating four patients. A finding from the trial was that the patients were subject to undertreatment. One postulated explanation is that local absorbers and heterogenous tissue in the vicinity of the fibers will decrease the therapeutic light power. In addition the threshold dose, i.e. the light dose required for tumor eradication, was too low. This was also confirmed in a pre-clinical canine study where the light-dose was escalated. The dosimetry scheme above only takes light into account. Work within the thesis has also been devoted to extend the dose model so that the concentration of the photosensible drug can be included. A method has been implemented that relies on the drug fluorescence in order to assess the spatial and temporal distribution, in connection to prostate-PDT. The feasibility was concluded in an experimental study. The method has also been applied to data from the clinical trial. Initial work has also been performed using another approach to dosimetry. During treatment the drug concentration can decrease, referred to as photobleaching. The photobleaching rate is dependent on several factors such as light and oxygen. Hence by assessing the photobleaching throughout the target volume treatment assessment can potentially be performed. This was investigated with measurements acquired from the clinical trial. Potential correspondence between treatment assessment using MRI, two weeks post-PDT, and the photobleaching dose is discussed. The determination of the drug distribution within the prostate using fluorescence can be referred to as fluorescence enhanced diffuse optical tomography (FDOT). Another vast field of research where such methods are applied, is the study of fluorescent compounds inside small animals. Then, measurements are performed on the surface of the animal followed by the execution of a theoretical reconstruction method, called an inverse model. Two approaches have been taken with the common goal to improve tomographic fluorescence imaging. By acquiring fluorescence measurements in several spectral bands, means arise for localization of several fluorophores simultaneously inside the body. The problem is that the theoretical calculation is challenging and requires ample amount of computational power. In order to alleviate this problem a tomographic imaging algorithm was implemented that decrease the computational burden. Another method, also relying on multispectral measurements, was developed with the intent to increase the robustness of the theoretical problem. Experimental work on optical phantoms has been performed to verify these methods. A problem in fluorescence imaging and tomography is the ever present autofluorescence arising from the tissue, surrounding a fluorophore. This problem was targeted using upconverting nanocrystals as fluorescent agents. In these nanocrystals, fluorescence is emitted at a shorter wavelength, compared to the excitation light. Autofluorescence, on the other hand, will be induced as normal, i.e. at longer wavelengths. Hence, autofluorescence-free imaging can be performed due to the spectral separation between the auto- and nanocrystal fluorescence. This was verified in both transillumination imaging and FDOT within an experimental phantom study.