Osteotomy in spine and skull base surgery is a highly demanding task that requires very high precision. Compared to conventional surgical tools, laser allows contactless hard tissue removal with fewer traumas to the patient and higher machining accuracy. However, a key issue remains unsolved: how to terminate the ablation while the underlying critical soft tissue is reached?

Our research group has realised a closed-loop control of a CO₂-laser osteotomy system under the guidance of an optical coherence tomography (OCT). The OCT provides three-dimensional information about the microstructures beneath the bone surface with a resolution on micrometre scale and an imaging depth of about 0.5 mm. The OCT and CO₂-laser systems are integrated using a coaxial setup and a registration between their working spaces (mean absolute error 19.6 μm) was performed.

The laser ablation and OCT scan are performed in turn. After correction of image distortions and speckle noise reduction, the position of the critical structure can be segmented in the enhanced OCT scans. The laser parameters for the next round of ablation are foresightedly planned based on the overlying residual bone thickness. After patient motion compensation by tracking artificial landmarks in the OCT scans (accuracy: RMS 27.2 μm), the ablation pattern can be precisely carried out by the CO₂-laser. The system was evaluated by performing laser cochleostomy on native porcine cochlea and mean ablation accuracy of 30 μm has been achieved.

However, for narrow incisions that are only several tens of micrometres wide, very few pixels are visible beneath the incision bottom in the OCT and a robust segmentation of the critical structure is impossible. We are now developing a hybrid control system, which monitors the ablation-induced acoustic emission (AE) as a secondary control mechanism in addition to the OCT.

When a pre-defined “switching” depth is reached, the AE-based control module is activated. Instead of analysing the acquired signals with conventional Fourier transform, a wavelet transform-based approach has been developed, which compares the correlation coefficients of the wavelet spectra of successive laser pulses. At the transition from bone tissue to the underlying soft tissue layer, a significant change in the coefficients can be observed, which is regarded as the signal for terminating the ablation. In order to keep the injury to the soft tissue layer to a minimal level, the laser energy is reduced after the switching. Preliminary experiments revealed that the wavelet-based approach is capable of controlling the ablation using pulses with extremely low energy down to 0.04mJ/pulse, resulting in an injured tissue layer of less than 10 μm.

We expect to achieve the ablation accuracy on tens of micrometre scale using the proposed hybrid control mechanism.