Research Report on Automated Robotic Craniotomy Surgery System

Background

As the central organ of complex life activities, the brain governs all psychological and conscious processes, playing a crucial role in every aspect of life. Entering the 21st century, neuroscience has become one of the fastest-growing and most prominent fields, with animal models playing a critical role in studying the brain and neural functions. However, medical imaging technologies widely used today, such as computed tomography (CT), magnetic resonance imaging (MRI), and functional near-infrared spectroscopy (fNIRS), while promising for observing brain structure and function, lack the resolution to capture single neuron activity. Therefore, optical microscopy techniques with micron-level resolution, such as two-photon microscopy, confocal microscopy, and optical coherence tomography (OCT), have become indispensable tools for advancing neuroscience research.

However, these optical techniques face limited penetration depths, making transcranial imaging challenging. To address this, researchers have developed transparent cranial window techniques for optical brain imaging. Traditional cranial window preparation requires manual operation with a drill to grind and cut the skull of experimental animals, followed by the installation of transparent materials matching the skull defect. This method demands highly skilled operators, consumes time, and has varying success rates depending on operator experience, posing risks of brain tissue damage. Therefore, there is an urgent need for precise, safe, and efficient automated craniotomy platforms.

Research Overview

This study, titled Optical Coherence Tomography Guided Automated Robotic Craniotomy Surgery Platform, was conducted by a research team from the Department of Biomedical Engineering, Southern University of Science and Technology, Shenzhen, China. The paper was published in the February 1, 2025 issue of Biomedical Optics Express (Vol. 16, No. 2). The team, led by Haoyuan Li, Yongchao Wang, Wei Chen, and others, received support from various funding bodies, including the National Natural Science Foundation of China and the Shenzhen Science and Technology Innovation Commission.

Research Process and Methods

Innovative System Design

The research team developed an OCT-Guided Automated Robotic Craniotomy Surgery Platform (OCT-ARC). This platform leverages the non-contact functionality, micron-level resolution, and 3D imaging capability of OCT to obtain accurate structural information of the intact skull. The platform also integrates a computer numerical control (CNC) drilling machine driven by a closed-loop stepper motor to precisely control the path of craniotomy.

OCT-Guided Skull Scanning and Data Analysis

The OCT system, configured with a broadband light source and fiber coupler, creates two optical paths directed to the sample and reference arms. Light reflected from the skull surface generates interference signals, which are collected by a commercial spectrometer. Further data processing—such as DC removal, linear K-space interpolation, and inverse Fourier transform—yields the sectional structural data of the skull.

Automated Drilling Module and Calibration

The drilling tool, with a 0.3 mm diameter tip, achieves high-precision three-axis movement via field-oriented control (FOC) of the stepper motor. The relative position between the drilling tip and the OCT optical focus is dynamically calibrated using pixel resolution and coordinate difference equations (e.g., pixel resolutions for air and skull are 5 µm/pixel and 3.3 µm/pixel, respectively), allowing for real-time depth and path adjustments.

Segmentation Algorithm for Skull Surfaces

The skull’s upper and lower surfaces are segmented from OCT data using image processing algorithms. A combination of Gaussian filtering, Otsu thresholding, and Canny edge detection is used to smooth and identify the upper surface boundaries. For the complex lower surface, the team developed a second-order vertical gradient method based on the Sobel operator and polynomial fitting, which accurately locates the lower boundary within 400 µm.

Workflow for Operation

The OCT-ARC system operates in four steps: 1. Perform a large-area 3D scan (C-scan) of the entire skull to acquire structural data. 2. Identify the cranial window region in the scan image and generate a 2D path, projecting this path to the upper and lower surfaces to create a 3D drilling trajectory. 3. Conduct stepwise drilling operations with real-time depth adjustments based on the drilling path and skull thickness. 4. Complete the craniotomy or skull thinning process, followed by the installation of transparent window materials.

Experiment Design and Sample Preparation

The study used 6-to-8-week-old C57 and BALB/C mice, including both males and females, alternatingly. To verify the system’s reliability, the researchers first conducted precision and repeatability tests on deceased mice before performing craniotomies for circular glass windows (diameter: 4 mm), large PMP windows (8 mm × 4 mm), and thinned skull windows (3.3 mm × 3.3 mm) on live mice.

Key Experimental Results

Accuracy Analysis

Precision tests revealed that the system in closed-loop control mode significantly outperformed open-loop mode. The positioning errors in the X, Y, and Z axes were -0.5 ± 1.1 µm, -1.8 ± 0.9 µm, and -0.3 ± 0.9 µm, respectively. The drilling depth tests showed that preset depths of 40 µm, 80 µm, and 120 µm yielded average actual depths of 34.3 ± 2.0 µm, 81.7 ± 1.5 µm, and 113 ± 3.1 µm, respectively.

Cranial Window Preparation

Multiple cranial window types were successfully prepared, including: 1. Glass Window: Drilling took 3–5 rounds, progressively deepening from 30% to 90% of the skull thickness. A circular 4 mm-diameter window was completed in about 3–5 minutes. 2. Large PMP Window: Similar stepwise drilling was used for the 8 mm × 4 mm window, involving 160 points and requiring 7–12 minutes. 3. Thinned Windows: Skull sections ≈180 µm thick were thinned in a total of 35 minutes to a residual thickness of 24.1 ± 4.4 µm, meeting transparency requirements for imaging.

Vascular Imaging Validation

High-quality brain vascular imaging was achieved through various cranial windows: 1. OCTA (Optical Coherence Tomography Angiography): Clear visualization of vascular networks, including pial arteries, veins, and capillaries. 2. DLS-OCT (Dynamic Light Scattering OCT): Blood flow velocity distribution measured across depths. 3. Power Doppler Imaging (PDI): Entire brain vasculature imaged through PMP windows, demonstrating excellent ultrasound transparency.

Conclusions and Future Directions

This study presents a novel OCT-Guided Automated Craniotomy Surgery Platform (OCT-ARC) that addresses the limitations of traditional craniotomy methods, such as extensive training requirements, operating complexity, and inconsistent success rates. The system’s non-contact, fully automated, and high-precision drilling capabilities significantly improve experimental efficiency while minimizing potential damage to brain tissue.

Key innovations include: - Integration of non-contact OCT and closed-loop stepper motor control for large-area, micron-level precision operations. - Optimized segmentation algorithms for automated upper and lower skull surface detection. - Initial validation of the platform’s high utility in mouse brain research, offering an efficient, safe, and cost-effective research tool for neuroscience.

In future research, the team plans to develop a low-cost OCT system and enhance the platform’s degrees of freedom to better meet advanced surgical system demands, paving the way for broader applications in neuroscience research.