Fast robot-to-iOCT calibration using conical geometry and 2D cross-scan OCT

Accurate spatial registration between optical coherence tomography (OCT) and robotic coordinate systems is critical for OCT-guided microsurgery, where precise tool localization is essential for safe, reliable intervention.

We propose and validate a practical and accurate method for calibrating the transformation between OCT and robot coordinate systems in robotic microsurgical systems.

A conical calibration tool rigidly attached to the robot wrist was scanned using fast 2D OCT cross-sections. After denoising, edge extraction, and two-line fitting, the cone tip was localized, and the resulting coordinates were incorporated into a vector-based three-point calibration to recover the OCT-to-robot transformation. The method supports both manual and automated operation, including pose adjustment to improve scan-plane alignment.

Experiments with a swept-source OCT system, a Meca500 robot, and a 3D-printed calibration tool showed stable registration across the OCT working range. The method also remained robust in the presence of OCT noise and moderate manufacturing imperfections of the cone geometry. The single cone-tip localization repeatability was estimated to be $34.7\mu m$ , where calibrating over 16 mm of OCT X-Y lateral field-of-view range translates to 0.31% of linear calibration uncertainty in a bench-top environment.

The proposed cone-based framework provides a practical and accurate solution for routine robot-to-OCT calibration in high-precision surgical systems.