Rapid Automated Analysis of Skull Base Tumor Specimens Using Intraoperative Optical Imaging and Artificial Intelligence

BACKGROUND

Accurate specimen analysis of skull base tumors is essential for providing personalized surgical treatment strategies. Intraoperative specimen interpretation can be challenging because of the wide range of skull base pathologies and lack of intraoperative pathology resources.

OBJECTIVE

To develop an independent and parallel intraoperative workflow that can provide rapid and accurate skull base tumor specimen analysis using label-free optical imaging and artificial intelligence.

METHODS

We used a fiber laser-based, label-free, nonconsumptive, high-resolution microscopy method (<60 seconds per 1 × 1 mm2), called stimulated Raman histology (SRH), to image a consecutive, multicenter cohort of patients with skull base tumor. SRH images were then used to train a convolutional neural network model using 3 representation learning strategies: cross-entropy, self-supervised contrastive learning, and supervised contrastive learning. Our trained convolutional neural network models were tested on a held-out, multicenter SRH data set.

RESULTS

SRH was able to image the diagnostic features of both benign and malignant skull base tumors. Of the 3 representation learning strategies, supervised contrastive learning most effectively learned the distinctive and diagnostic SRH image features for each of the skull base tumor types. In our multicenter testing set, cross-entropy achieved an overall diagnostic accuracy of 91.5%, self-supervised contrastive learning 83.9%, and supervised contrastive learning 96.6%. Our trained model was able to segment tumor-normal margins and detect regions of microscopic tumor infiltration in meningioma SRH images.

CONCLUSION

SRH with trained artificial intelligence models can provide rapid and accurate intraoperative analysis of skull base tumor specimens to inform surgical decision-making.

Off-Peak Near-Infrared-II (NIR-II) Bioimaging of an Immunoconjugate Having Peak Fluorescence Emission in the NIR-I Spectral Region for Improving Tumor Margin Delineation

The primary treatment for malignant tumors remains to be resection. The strongest predictor of recurrence and postoperative prognosis is whether diseased tissue/cells remain(s) at the surgical margin. Cancer surgery entails surgeons having the capability to visually distinguish between subtle shades of color in attempts of differentiating between diseased tissue and healthy tissue under standard white-light illumination, as such tissue states appear identical at the meso-/macroscopic level. Accordingly, enhancing the capability of surgeons to do so such that they can accurately delineate the tumor margin is of paramount importance. Fluorescence-guided surgery facilitates in enhancing such capability by color-coding the surgical field with overlaid contrasting pseudo-colors from real-time intraoperative fluorescence emission via utilizing fluorescent constructs in tandem. Constructs undergoing clinical trials or that are FDA-approved provide peak fluorescence emission in the visible (405 - 700 nm) or near-infrared-I (NIR-I) spectral region (700-900 nm), whereby differentiation between tissue states progressively improves in sync with using constructs that emit longer wavelengths of light. Here, we repurpose the usage of such fluorescent constructs by establishing feasibility of a tumor-targeting immunoconjugate (cetuximab-IRDye800) having peak fluorescence emission at the NIR-I spectral region to provide improved tumor margin delineation by affording higher tumor-to-background ratios (TBRs) when measuring its off-peak fluorescence emission at the near-infrared-II (NIR-II) spectral region (1000-1700 nm) in in vivo applications. We prepared murine tumor models, administered such immunoconjugate, and imaged such models pre-/post-administration via utilizing imaging systems that separately afforded acquisition of fluorescence emission in the NIR-I or NIR-II spectral region. On doing so, we determined in vivo TBRs, ex vivo TBRs with/-out skin, and ex vivo biodistribution, all via measuring the fluorescence emission of the immunoconjugate at tumor site(s) at both spectral regions. Collectively, we established feasibility of using the immunoconjugate to afford improved tumor margin delineation by providing 2-fold higher TBRs via utilizing the NIR-II spectral region to capture off-peak fluorescence emission from a fluorescent construct having NIR-I peak fluorescence emission.