Supplementary material for “High-resolution depth measurements in digital microscopic surgery”

Supplementary material for “High-resolution depth measurements in digital microscopic surgery”

Raw sensor and RBF data of various tissue samples proving the feasibility of the proposed method for carrying out high-resolution depth measurements based on laser triangulation as a supporting technology in the digital microsurgical workflow.

Fully digital microscopes are becoming more and more common in surgical applications. In addition to high-resolution stereoscopic images of the operating field, which can be transmitted over long distances or stored directly, these systems offer further potentials by supporting the surgical workflow based on their fully digital image processing chain. For example, the image display can be adapted to the respective surgical scenario by adaptive color reproduction optimization or image overlays with additional information, such as the tissue topology. Knowledge of this topology can be used for computer-assisted or AR-guided microsurgical treatments and enables additional features such as spatially resolved spectral reconstruction of surface reflectance.

In this work, we proposed a new method for high-resolution depth measurements in digital microsurgical applications, which is based on the principle of laser triangulation. Part of this method was a sensor data fusion procedure to properly match the laser scanner and camera data. In this context, a strategy based on radial basis function (RBF) interpolation techniques was applied to handle missing or corrupt data, which, due to the measuring principle, can occur on steep edges and through occlusion. The proposed method was eventually used for the acquisition of high-resolution depth profiles of various organic tissue samples. The corresponding results are depicted in Sliders 1 to 4. In each case, the fusion of laser scanner and camera data yielded a data matrix containing both image and depth information at each given pixel location.

Slider 1 – Visual representation of the laser-triangulation depth measurement results for the fractured bone sample. An RBF technique was applied for the interpolation of missing data points. For convenience, the gray-scale image of the tissue sample as acquired by the monochrome camera system is compared to a color map overlay visualizing the tissue topology. Additionally shown is a three-dimensional depth reconstruction as obtained from the sensor-fusion data matrix. The corresponding raw and RBF data can be downloaded here.

Slider 2 – Visual representation of the laser-triangulation depth measurement results for the muscle tissue sample. An RBF technique was applied for the interpolation of missing data points. For convenience, the gray-scale image of the tissue sample as acquired by the monochrome camera system is compared to a color map overlay visualizing the tissue topology. Additionally shown is a three-dimensional depth reconstruction as obtained from the sensor-fusion data matrix. The corresponding raw and RBF data can be downloaded here.

Slider 3 – Visual representation of the laser-triangulation depth measurement results for the tissue sample of a pig’s lung. An RBF technique was applied for the interpolation of missing data points. For convenience, the gray-scale image of the tissue sample as acquired by the monochrome camera system is compared to a color map overlay visualizing the tissue topology. Additionally shown is a three-dimensional depth reconstruction as obtained from the sensor-fusion data matrix. The corresponding raw and RBF data can be downloaded here.

Slider 4 – Visual representation of the laser-triangulation depth measurement results for the muscle tissue of a pig’s tongue. An RBF technique was applied for the interpolation of missing data points. For convenience, the gray-scale image of the tissue sample as acquired by the monochrome camera system is compared to a color map overlay visualizing the tissue topology. Additionally shown is a three-dimensional depth reconstruction as obtained from the sensor-fusion data matrix. The corresponding raw and RBF data can be downloaded here.

As can be seen, the proposed method of high-resolution depth measurements based on laser-triangulation works quite well for all tissue samples considered in this work. In all cases, problems of incorrect data representations because of failed RBF interpolation are only observed at the edges of the tissue samples, where, due to occlusions or overly steep gradients, the scattered light is not reflected back onto the photosensitive detector unit of the laser scanner. Besides these small, edge-related disturbances, no further severe measurement errors are observed. In particular, an equally good performance of the proposed method can be concluded when assessing the inner structure of the tissue surfaces, i.e., all parts that are sufficiently far from the samples' edges. As can be seen, the tissue fine structures are resolved properly in all cases without showing any non-smooth perturbations. This basically indicates that neither residual surface moisture nor a specific tissue color prevent the proposed method from providing accurate depth measurements when being applied to different kinds of organic tissue samples and, consequently, confirms its feasibility as a supporting technology in the digital microsurgical workflow.

Contact: