Thorsten Vollborn

A voice-coil actuated ultrasound micro-scanner for intraoral high resolution impression taking

Silicone based impression-taking of prepared teeth followed by plaster casting is well-established but potentially less reliable, error-prone and inefficient for newly emerging techniques such as computer aided design and manufacturing (CAD/CAM) of dental prosthetics. Intra-oral optical scanners have been introduced to increase efficiency of CAM but no breakthrough occurred so far. Oral liquids such as saliva, blood and sulcular fluid are still one of the main problems since the preparation area must be completely dry. Moreover, sub-gingival preparations need to be uncovered invasively prior to scanning and a reflecting powder coating is required in some cases. High frequency ultrasound (HFUS) has been recently introduced as an alternative to optical scanning. Ultrasound is less sensitive against oral fluids and in principal able to penetrate gingiva in a patient-friendly and cost-effective way. Although HFUS systems have been introduced for ophthalmology, dermatology or small animal imaging, none of them suits the challenging requirements and high accuracy demands for intra-oral micro-scanning of prepared teeth. For this reason, we conceived a new ultrasonic micro-scanning device based on a voice-coil actuated spherically focused HFUS transducer for intra-oral use. The system, which is designed for both highly dynamic accurate positioning and micrometer-resolution, is supplied with a sensor providing position feedback for motion control as well as the ultrasound trigger engine. In this contribution, we describe the set-up and evaluate the lateral displacement of the micro-scanners end-effector with respect to the oscillation rate using laser triangulation. The results are in good agreement to the requirements of an intra-oral ultrasound based micro-scanner.

Accuracy assessment of high frequency 3D ultrasound for digital impression-taking of prepared teeth

Silicone based impression-taking of prepared teeth followed by plaster casting is well-established but potentially less reliable, error-prone and inefficient, particularly in combination with emerging techniques like computer aided design and manufacturing (CAD/CAM) of dental prosthesis. Intra-oral optical scanners for digital impression-taking have been introduced but until now some drawbacks still exist. Because optical waves can hardly penetrate liquids or soft-tissues, sub-gingival preparations still need to be uncovered invasively prior to scanning. High frequency ultrasound (HFUS) based micro-scanning has been recently investigated as an alternative to optical intra-oral scanning. Ultrasound is less sensitive against oral fluids and in principal able to penetrate gingiva without invasively exposing of sub-gingival preparations. Nevertheless, spatial resolution as well as digitization accuracy of an ultrasound based micro-scanning system remains a critical parameter because the ultrasound wavelength in water-like media such as gingiva is typically smaller than that of optical waves. In this contribution, the in-vitro accuracy of ultrasound based micro-scanning for tooth geometry reconstruction is being investigated and compared to its extra-oral optical counterpart. In order to increase the spatial resolution of the system, 2nd harmonic frequencies from a mechanically driven focused single element transducer were separated and corresponding 3D surface models were calculated for both fundamentals and 2nd harmonics. Measurements on phantoms, model teeth and human teeth were carried out for evaluation of spatial resolution and surface detection accuracy. Comparison of optical and ultrasound digital impression taking indicate that, in terms of accuracy, ultrasound based tooth digitization can be an alternative for optical impression-taking.

High Frequency (75MHz) Ultrasound based Tooth Digitization using Sparse Spatial Compounding

Over the last decade, extra- and intraoral optical scanning for computer integrated manufacturing (CIM) of dental restorations became the focus of interest. Despite the fact that intraoral systems are getting more and more accurate, their invivo accuracy is influenced by the presence of oral fluids. Moreover, subgingival preparation margins need to be uncovered invasively prior to the scan and powder may be required to cope with different translucency and reflectivity of target materials. High frequency ultrasound (HFUS) based intraoral micro-scanning could be an alternative technology for optical impression taking. However, for accurate 3D teeth geometry reconstruction, homogenously distributed echoes of the occlusal and lateral tooth surfaces as well as margins are required. Whereas HFUS phased array technology for 3D image compounding is not yet available, mechatronic single element based transceiver concepts at most require 5 degrees of freedom (4 in case of synthetic aperture focusing). To overcome these drawbacks, a sparse spatial compounding (SSC) technique is being investigated which makes use of only a limited number of additional scanning directions under a fixed incidence angle allowing for simplification of the final micro-scanning kinematic. Measurements with extracted prepared human molar teeth have been carried by using an extraoral HFUS-SSC scanner. The results demonstrate that with SSC almost homogenously distributed spatial data coverage of tooth surface points can be achieved.

An ultrasonic micro-scanner for thickness assessment of the vestibular jawbone: In-vitro accuracy evaluation

Progressive peri-implant bone loss may lead to implant failure. Conventionally, the vestibular jawbone thickness (VJT) is monitored via cone beam computed tomography. Ionizing radiation and artifacts due to metallic implants, as well as superstructures, are major drawbacks of x-ray-based imaging techniques. As a non-invasive and patient-friendly alternative, high frequency ultrasonic (HFUS) micro-scanning can be used to assess the jawbone surface. In this study, we present a HFUS scanner and algorithm for assessing the jawbone thickness which is based on a priori information of the superstructure surface and which does not require ultrasound penetration of the jawbone. Four implants were inserted into bovine ribs. Prior to mounting the polymer superstructures, the position and orientation of the implant relative to the superstructure surface was determined based on optical 3-D scans. Subsequently, porcine gingiva samples were attached to the ribs. The specimens were fixed in a water basin filled with isotonic saline solution. Ultrasound data was acquired with a HFUS micro-scanner (center frequency: > 50 MHz, relative bandwidth: > 70%, aperture 4 mm, focus 8 mm) designed for intra-oral use supporting highly dynamic accurate positioning in micrometer-resolution. Within this study, the ultrasound trigger spacing was set to 39 μm. Bone and superstructure surfaces were segmented out of the ultrasound data and converted to polygon meshes. These meshes were matched to an a priori acquired 3-D model of the superstructure. Finally, the vestibular bone thickness was calculated from the matched 3-D model. For evaluation of the proposed ultrasonic technique all specimens were cut into slices (thickness approx. 1 mm), examined using a stereo-microscope and compared to the calculated bone thickness. The overall error of ultrasound based bone thickness determination was 38±99μm (max: 260μm, min: -240μm) which is in good agreement to clinical requirements and which outperforms conventional x-ray based cone beam computed tomography.

Robot integrated ultrasound geometry-scanning for trackerless bone cement detection in RTHR

Detection and removal of intra-femoral bone-cement in the context of revision total hip replacement (RTHR) can be a time-consuming and risky intervention. Within the framework of computer and robot assisted surgery a system for integrated intraluminal ultrasound based cement detection using a modular mini-robot is being developed. One challenge of an ultrasound based approach is that refraction and wave mode conversion at the water-cement interface will cause refraction distortion and artefacts respectively. In order to minimize measurement errors a scanning concept aiming on an almost perpendicular beam direction with respect to the local water cement interface scanning has been developed. Bone cement prepared ex-vivo femora have been ultrasound scanned and the results have been compared with golden standard CT data. The ultrasound integrated geometry scanner, the detection algorithms as well as the ex-vivo evaluation is presented as part of this contribution.

Distortion reduction for a dental HFUS microscanning device

Silicone impression-taking of teeth is an established but inefficient technique for computer-aided manufacturing of dental prosthetics. Hence, intra-oral scanners based on optical technologies have been developed to optimize the process of impressioning and to enable a complete chairside workflow. However, during the scan, the presence of blood or saliva can cause critical model defects and the need to invasively expose subgingival areas remains essential for scanning. We introduced high frequency ultrasound (HFUS) as a new method for scanning dental structures. Ultrasound is less sensitive to oral fluids and is inherently able to penetrate gingiva non-invasively. HFUS-based intra-oral microscanning (USM) of teeth requires both a high spatial resolution and a mechanical precision to finally achieve an optimal impression detail. We designed a microscanner based on a direct drive mechanism with 2 degrees of freedom. In a preliminary study we tested the systems 2D-distortion by scanning a ball grid array as a reference body (RB) in degassed and tempered water. The ball grids accuracy was measured by a commercial optical reference scanner (OS). The Euclidean distance (ED) between ball centers of the RB and the USM were calculated. Afterwards, we implemented a 2D-distortion reduction method (DRM). The scaling factors (coefficients of the compensation matrix) were calculated by a linear regression analysis on the EDs. Finally, the systems precision was evaluated by scanning the occlusal surface of a molar tooth with the USM and applying the DRM. The acquired model was aligned and compared to a scan obtained with the OS by using a best-fit algorithm. Without using the DRM we measured a mean deviation of 14.8 microns (positive) and 21.2 microns (negative) (SD 22.7 microns). After the application of the DRM we achieved a mean deviation of 7.6 microns (positive) and 19.9 microns (negative) (SD 16.9 microns). The scan quality of the USM was improved by the developed DRM and thus reached the anticipated accuracy range for intra-oral impressioning.

Model based assessment of vestibular jawbone thickness using high frequency 3D ultrasound micro-scanning

Endosseous implants are well-established in modern dentistry. However, without appropriate therapeutic intervention, progressive peri-implant bone loss may lead to failing implants. Conventionally, the particularly relevant vestibular jawbone thickness is monitored using radiographic 3D imaging methods. Ionizing radiation, as well as imaging artifacts caused by metallic implants and superstructures are major drawbacks of these imaging modalities. In this study, a high frequency ultrasound (HFUS) based approach to assess the vestibular jawbone thickness is being introduced. It should be emphasized that the presented method does not require ultrasound penetration of the jawbone. An in-vitro study using two porcine specimens with inserted endosseous implants has been carried out to assess the accuracy of our approach. The implant of the first specimen was equipped with a gingiva former while a polymer superstructure was mounted onto the implant of the second specimen. Ultrasound data has been acquired using a 4 degree of freedom (DOF) high frequency (<50MHz) laboratory ultrasound scanner. The ultrasound raw data has been converted to polygon meshes including the surfaces of bone, gingiva, gingiva former (first specimen) and superstructure (second specimen). The meshes are matched with a-priori acquired 3D models of the implant, the superstructure and the gingiva former using a best-fit algorithm. Finally, the vestibular peri-implant bone thickness has been assessed in the resulting 3D models. The accuracy of this approach has been evaluated by comparing the ultrasound based thickness measurement with a reference measurement acquired with an optical extra-oral 3D scanner prior to covering the specimens with gingiva. As a final result, the bone thicknesses of the two specimens were measured yielding an error of −46±89μm (first specimen) and 70±93μm (second specimen).

Sensitivity analysis of synthetic aperture focusing based on the virtual source point for high-frequency ultrasound imaging

In a previous work we investigated the use of the synthetic aperture focusing technique based on a virtual point source (VSAFT) for high frequency ultrasound-based tooth surface reconstruction in the context of CAD/CAM-based dental restorations. However, discrepancies between real and assumed values of the transducers and coupling medias parameters could affect the performance of the VSAFT. In the framework of this work an analysis has been conducted to investigate the influence of wrong assumptions concerning speed of sound and location of the virtual point source (VPS) with respect to spatial resolution and SNR after processing with VSAFT in case of using a focused (f#=2) high frequency (75MHz) ultrasound probe. Additionally, a comparison between common apodization windows (Hamming, triangle and cosine) and the coherence factor method (CF) has been carried out. The results show that already 3% deviation of the VPS position from their real value causes a significant deterioration of the achieved 6-dB lateral resolution after VSAFT processing. The best lateral resolution and SNR have been achieved with the CF method. Furthermore, the acoustic focus is more appropriate for VSAFT compared to the geometric pendant.