Automated laser retraction for targeted glioblastoma coverage during laser interstitial thermal therapy

Abstract Background Current magnetic resonance guided laser interstitial thermal therapy (MRgLITT) requires physician input to incrementally retract the laser probe during glioblastoma (GBM) treatment, to achieve maximum safe lesion coverage (LC). Objective Through this computational study, we propose an automated MRgLITT system based on temperature and thermal damage feedback to enable precise laser probe positioning and achieve the target LC during GBM treatment. Methods Our goal was to design a cascaded proportional‐integral‐derivative (PID) controller with a fuzzy logic controller to achieve thermal damage, Ω ≥ 0.99 within 90% of the target tumor volume. Two modelling approaches were explored for PID control system to maintain the tumor boundary temperature at: (i) constant optical properties and (ii) variable optical properties. A fuzzy‐logic controller was used to incrementally retract the laser probe once the thermal damage at the corresponding boundary setpoint reached the target (Ω = 0.99). The PID and fuzzy logic controller was designed in the system dynamic modeling software, MATLAB Simulink, and bioheat transfer physics was modelled using the finite element analysis (FEA) software, COMSOL Multiphysics. FEA simulations were performed on a de‐identified patient voxel model consisting of domains such as the skull, CSF, brain and tumor. The laser probe was modeled as a cylinder with a diameter of 1.65 mm and tip length of 5 mm. The temperature and thermal damage measuring probes were modeled to represent the thermal imaging planes from clinically approved LITT systems. Results The results indicated that the automated control framework maintained the tumor boundary temperature at 60°C. The designed automated control framework did not achieve the target LC but increased the LC by ∼25% with reduction in treatment time. Conclusion The results of this computational study indicate that the designed automated MRgLITT approach has potential to improve the LC, reduce treatment time and operator specific variability. Future efforts will focus on developing and validating the proposed approach.