Oral squamous cell carcinoma (OSCC) is a significant contributor to the global cancer burden, with surgical resection being the primary course for treatment. However, achieving clear surgical... Show moreOral squamous cell carcinoma (OSCC) is a significant contributor to the global cancer burden, with surgical resection being the primary course for treatment. However, achieving clear surgical margins remains a challenge owing to the complex anatomy of the head and neck, and it is estimated that as many as 30% of patients are sent home after surgery with residual cancer. Current surgical margin assessment techniques, such as post-operative histopathology and intraoperative frozen section analysis, either take too long to complete in an intraoperative time frame or have limitations in accuracy and efficiency. Fluorescence-guided surgery (FGS) has emerged as a promising approach to enhance intraoperative surgical margin assessment, with a few clinical research groups in the US and the Netherlands demonstrating the potential of epidermal growth factor receptor (EGFR)-targeted FGS in OSCC to identify at least a fraction of insufficient margins, intraoperatively. However, existing methods have demonstrated limited accuracy in detecting so-called “close” margins (cancer is found between 1-5 mm from the surgical margin), which are associated with an increased risk of local recurrence compared to “clear margins (>5 mm of healthy tissue at the surgical margin).This thesis evaluated the potential for a dual-aperture fluorescence ratio (dAFR) imaging approach to be used to improve the accuracy of close margin detection and localization in OSCC surgical margins. The dAFR technique involves dividing high numerical-aperture (NA) fluorescence images by low NA images taken from the same point-of-view, creating a ratio image that has been hypothesized to be correlated with depth of the fluorescence inclusion, in a manner that is insensitive to variability in tissue optical properties, in sub-diffuse photon propagation regimes. The validity of the hypothesis that dAFR signals can be correlated directly to depth of fluorescence in OSCC surgical margins was first explored in a set of Monte Carlo photon propagation simulations. The Monte Carlo simulation framework was also used to (1) guide the selection of the optimal range of numerical apertures (NAs) to use for the wide and narrow apertures for dAFR, (2) for evaluating the accuracy and precision of dAFR depth estimation across varying tissue optical properties, and (3) for evaluating dAFR depth estimation accuracy when tissue optical properties were heterogeneous, as would be expected in clinical OSCC margin imaging applications. A prototype dAFR-capable imaging system was designed based on the findings of the simulation work and was constructed considering factors such as accessibility, rapid imaging capability, safety, field-of-view, and image quality. The performance of the constructed system was then first evaluated through phantom experiments, where resinous materials with optical properties matching those of biological tissue and fluorescence at various depths were used to experimentally demonstrate the depth sensitivity of dAFR compared to single-aperture fluorescence (sAF) imaging. The dAFR system was then deployed to a clinic at the University Medical Center Groningen (UMCG) in The Netherlands to be tested out in a clinical pilot study. There, the surgical margins of 3 OSCC patients were imaged with our dAFR approach and the more conventional sAF approach and the correlations between dAFR and sAF measurement to histopathology measured margin thickness was evaluated at 12 different margin locations from 3 patients. In this pilot group of patients, dAFR provided significantly higher accuracy in detecting close margins compared to sAF (p < 0.02), with an area under the receiver operating characteristic curve (AUC of the ROC) of 1.0 for close margins. These results embody the first clinical demonstration of close margin detection in an intraoperative timeframe (< 2 min of imaging). Though the sample size was small (n = 3), these preliminary results have been used to leverage funding of the ongoing development of a second system prototype and the commencement of an 80-patient clinical study over the next 5 years. Other future research directions include the optimization of imaging hardware, correction for surface topography, expansion of clinical studies to other cancer types, integration with other imaging modalities, development of user-friendly interfaces, automated margin detection and localization of insufficient margins, and enhanced co-registration of margin localization in excised margins with in vivo anatomical structures to guide potential extended resection. Show less
