Automating neurosurgical tumor resection surgery: Volumetric laser ablation of cadaveric porcine brain with integrated surface mapping.

OBJECTIVE: Current surgical instruments for soft tissue resection including neurosurgical procedures rely on the accuracy and precision of the human operator and are fundamentally constrained by the human hand. Automated surgical action with the integration of intraoperative data sources can enable highly accurate and fast tissue manipulation using laser ablation. This study presents the first experiments with a prototype designed for automated tumor resection via laser ablation. We demonstrate targeted soft tissue resection in porcine brain with an integrated device that combines 3D scanning capabilities with a steerable surgical laser and discuss implications for future automated robotic neurosurgical procedures. STUDY DESIGN AND METHODS: A device consisting of a two-axis galvanometer for steering a cutting laser and a 3D surface profiler is used to perform volumetric removal of tissue of ex vivo porcine brain. Three-dimensional surface profiles are gathered between cuts and used to estimate ablation rate. RESULTS: Volumetric ablation of porcine brain tissue is performed and subsequently surface profiled. The average ablation rates across the area cutting areas were 2.6 mm3 /s and 3.7 mm3 /s for the initial and subsequent cuts, respectively. A Kruskal-Wallis and post-hoc Tukey test show statistical significance between the initial and subsequent cuts. Accuracy between cuts when benchmarked against a human surgeon varied from 47 to 88%. CONCLUSION: A feed-forward volumetric resection is demonstrated with sensing and cutting housed within a single device, thereby opening the potential for automated soft tissue resection as necessary during the surgical removal of pathologic tissues. High variance around target cut depths motivates future work in developing a closed-loop ablation tool as well as characterization of laser-tissue interactions for predictive modelling. Objective Lasers Surg. 50:1017-1024, 2018. © 2018 Wiley Periodicals, Inc.

A blinded study using laser induced endogenous fluorescence spectroscopy to differentiate ex vivo spine tumor, healthy muscle, and healthy bone.

Ten patients undergoing surgical resection for spinal tumors were selected. Samples of tumor, muscle, and bone were resected, de-identified by the treating surgeon, and then scanned with the TumorID technology ex vivo. This study investigates whether TumorID technology is able to differentiate three different human clinical fresh tissue specimens: spine tumor, normal muscle, and normal bone. The TumorID technology utilizes a 405 nm excitation laser to target endogenous fluorophores, thereby allowing for the detection of tissue based on emission spectra. Metabolic profiles of tumor and healthy tissue vary, namely NADH (bound and free emission peak, respectively: 487 nm, 501 nm) and FAD (emission peak: 544) are endogenous fluorophores with distinct concentrations in tumor and healthy tissue. Emission spectra analyzed consisted of 74 scans of spine tumor, 150 scans of healthy normal bone, and 111 scans of healthy normal muscle. An excitation wavelength of 405 nm was used to obtain emission spectra from tissue as previously described. Emission spectra consisted of approximately 1400 wavelength intensity pairs between 450 and 750 nm. Kruskal-Wallis tests were conducted comparing AUC distributions for each treatment group, α = 0.05. Spectral signatures varied amongst the three different tissue types. All pairwise comparisons among tissues for Free NADH were statistically significant (Tumor vs. Muscle: p = 0.0006, Tumor vs. Bone: p < 0.0001, Bone vs. Muscle: p = 0.0357). The overall comparison of tissues for FAD (506.5-581.5 nm) was also statistically significant (p < 0.0001), with two pairwise comparisons being statistically significant (Tumor vs. Muscle: p < 0.0001, Tumor vs. Bone: p = 0.0045, Bone vs. Muscle: p = 0.249). These statistically significant differences were maintained when stratifying tumor into metastatic carcinoma (N = 57) and meningioma (N = 17). TumorID differentiates tumor tissue from normal bone and normal muscle providing further clinical evidence of its efficacy as a tissue identification tool. Future studies should evaluate TumorID's ability to serve as an adjunctive tool for intraoperative assessment of surgical margins and surgical decision-making.