A key step for mitigating tumor recurrence for patients with head and neck cancer is adequate surgical margin delineation. Presently available techniques however limit accurate tumor margin detection during surgery. Herein, we report on tumor visualization using deep learning by combining autofluorescence images acquired by a fiber-based fluorescence lifetime imaging (FLIm) system and white light images (WLI) obtained by surgical cameras. To accomplish accurate registration between FLIm and WLI, a tissue motion correction algorithm was employed as a pre-processing step. The trained model was applied to differentiation of healthy and cancerous tissues in a 50 head and neck cancer patients dataset (ROC-AUC : 0.87).
Robotic and endoscopic surgery is increasingly used in clinical practice and typically relies on stereoscopic vision to enable 3D visualization of the surgical field. We combined this capability with a FLIm acquisition system suitable for the identification of tumor tissue to generate a 3D map of the surgical field that comprises both FLIm and white light image information. This result is achieved using semi-global matching and deep stereo matching neural network. In addition to the generation of a 3D model of the surgical cavity, this approach leads to a more realistic rendering of FLIm maps by including tissue shading.
Accurate cancer margin assessment prior to surgical resection is a key factor influencing the long-term survival of oral and oropharyngeal cancer patients. This leads to the need for additional guidance tools for real-time delineation of cancer margins. In this work, fiber-based fluorescence lifetime Imaging (FLIm) was combined with machine learning to perform intraoperative tumor identification. The developed classifier achieved a measurement-level ROC-AUC of 0.89±0.03 on an N=62 patient dataset. A transparent overlay of classifier output was augmented onto the surgical field and updated through tissue motion correction, ensuring co-registration between tissue and spectroscopic data/classifier output was maintained during imaging..
Oral cavity and oropharyngeal cancers are leading pathologies, representing 3% of all new cancer cases in the United States. Adequate intraoperative marginal clearance of these malignancies is essential for long-term survival; however, presently available techniques limit precise instantaneous tumor margin characterization. Herein, we report the clinical validation of a fiber-based fluorescence lifetime imaging device for real-time intraoperative tumor delineation. Results from 72 human patients are reported (autofluorescence trends, ROC-AUC), including diverse cancer histologies, anatomic sites (e.g. tongue, tonsil, etc.), and patient medical histories. Emphasis is placed on results governing the detection of unknown primary tumors from 4 patients, as well as data from 5 patients presenting with residual carcinoma.
In this work, we evaluate the potential for Fluorescence Lifetime Imaging (FLIm) to complement a surgeon's visual, endoscopic, and pathologic assessment of the adequacy of intraoperative tumor resection in clinical cancer applications of the oral cavity and oropharynx. Using a custom-built FLIm instrument during both non-robotic and robotic assisted surgical procedures, we show that intrapatient contrast between healthy and tumor tissue can be achieved intraoperatively in vivo prior to cancer resection with statistical significance (p<0.001) in 9/9 patients using at least 1/6 FLIm parameters, and ex vivo for surgically excised specimens (p<0.001) for 8/9 patients. We employ a multi-parameter linear discriminant analysis approach to demonstrate superior pathology discrimination ability through leveraging a weighted combination of all FLIm metrics. We also highlight interpatient comparisons to evaluate how FLIm signatures vary across different patients and disparate tissue anatomies.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.