Susanne Winter

Registration of CT and Intraoperative 3-D Ultrasound Images of the Spine Using Evolutionary and Gradient-Based Methods

A system for the registration of computed tomography and 3-D intraoperative ultrasound images is presented. Three gradient-based methods and one evolutionary algorithm are compared with regard to their suitability to solve this image registration problem. The system has been developed for pedicle screw insertion during spinal surgery. With clinical preoperative and intraoperative data, it is demonstrated that precise registration is possible within a realistic range of initial misalignment. Significant differences can be observed between the optimization methods. The covariance matrix adaptation evolution strategy shows the best overall performance, only four of 12 000 registration trials with patient data failed to register correctly.

Phase difference based RFID navigation for medical applications

RFID localization is a promising new field of work that is eagerly awaited for many different types of applications. For use in a medical context, special requirements and limitations must be taken into account, especially regarding accuracy, reliability and operating range. In this paper we present an experimental setup for a medical navigation system based on RFID. For this we applied a machine learning algorithm, namely support vector regression, to phase difference data gathered from multiple RFID receivers. The performance was tested on six datasets of different shape and placement within the volume spanned by the receivers. In addition, two grid based training sets of different size were considered for the regression. Our results show that it is possible to reach an accuracy of tag localization that is sufficient for some medical applications. Although we could not reach an overall accuracy of less than one millimeter in our experiments so far, the deviation was limited to two millimeters in most cases and the general results indicate that application of RFID localization even to highly critical applications, e. g., for brain surgery, will be possible soon.

Toward Registration of 3D Ultrasound and CT Images of the Spine in Clinical Praxis: Design and Evaluation of a Data Acquisition Protocol

Recent work has demonstrated the accuracy and operational viability of an algorithm proposed by the authors that successfully registers 3-D ultrasound data with CT or MRI data. The successful application of this method to intraoperative navigation, however, depends critically on the quality of the acquired ultrasound data. This gives rise to two questions concerning the usability of the algorithm in clinical praxis. First, how can one guarantee high-quality, user-independent ultrasound registration data with this procedure? Second, can this approach work reliably in clinical practice, namely within the operating theater? To address both of these questions, we present an ultrasound data acquisition protocol that leads the user through the data acquisition process and also provides the criteria to adjust the relevant ultrasound parameters. We also evaluated criteria for the visual inspection of the suitability of the ultrasound data for the registration process. Results for this evaluation show that these visual criteria can be used to decide preoperatively if an ultrasound registration will be successful in a patient. The intraoperative evaluation of the protocol showed that high-quality registrations can be achieved under realistic conditions. This protocol and the visual inspection criteria, together with the ultrasound registration algorithm, provide a surgical team with a means of performing precise, cost-effective navigation in patients for whom a navigated intervention was previously impossible. We evaluated the proposed procedure in clinical practice.

Registration of bone surfaces, extracted from CT-datasets, with 3D ultrasound

An essential task of computer assisted surgery is the registration of preoperative image data with the coordinate system of the operating room. This can be reached by using intraoperative imaging and registrating preoperative and intraoperative datasets. For intraoperative imaging ultrasound is a powerful tool due to the lack of ionizing radiation and because of its fast, inexpensive and easy data acquisition. We propose a surface volume matching algorithm for the registration of bone surfaces and ultrasound volume data. The bone surface is estimated from the preoperative CT data by taking into account that ultrasound only shows parts of the bone surface. By our method reliable matching results are obtained. They are shown with data of the lumbar spine.

Bone registration with 3D CT and ultrasound data sets

Abstract For many navigated surgical procedures, the precise registration of preoperative data sets with bones of the patient is an important requisite. Conventional navigation systems use paired point registration based on anatomical landmarks or fiducial markers. This approach increases the invasiveness, since landmarks must be exposed and fiducial markers must be connected to the bone. Intraoperative imaging modalities can overcome this disadvantage. Ultrasound seems to be ideal because of the easy data acquisition. The problem, however, is the low imaging quality regarding bones. The proposed algorithm for the registration of CT and ultrasound data sets considers the ultrasound imaging properties. That part of the bone surface, which should be visible in the ultrasound data is estimated from the CT data. The ultrasound data is preprocessed to emphasize bone surfaces. Thus, the ultrasound data contains a bright shape that is formed like the surface estimated from the CT data. A surface–volume registration tries to correlate the estimated surface with this bright shape. The algorithm was validated using an ex vivo preparation of a human lumbar spine. The algorithm was shown to cope with initial misalignments of about 30 mm and 15°. Successful registration of in vivo data of lumbar spine, tibia and shoulder indicate the feasibility of the approach.

In vivo Evaluierung und in vitro Genauigkeitsmessung für einen Algorithmus zur Registrierung von Ultraschall- und CT-Datensätzen

Ein Algorithmus zur Registrierung von Knochen in dreidimensionalen CT- und Ultraschalldatensatzen, der auf einem Oberflachen-Volumen-Matching basiert, wurde bezuglich seiner Genauigkeit in vitro getestet. Als Vergleichsverfahren diente eine Punktregistrierung anhand kunstlicher Mar-ken. Der Vergleich ergab, dass fur einen Wirbel Abweichungen bis 1.7 mm zwischen den Registrierungen auftraten, wobei der systematische Anteil der Abweichung vernachlassigbar klein war. Weiterhin wurde die Ultraschallregistrierung an in vivo Daten evaluiert, wobei sich ergab, dass der Algorithmus fur unterschiedliche Knochenstrukturen geeignet ist.

Registration of Intraoperative 3D Ultrasound with MR Data for the Navigated Computer Based Surgery

Computer based navigated surgery assists the spatial orientation of the surgeon. Our system registers preoperative data like CT or MR with intraoperative ultrasound data to get the coordinate transformation between the preoperative and the intraoperative data. With a surface volume registration we avoid a difficult surface segmentation in the ultrasound data. To prevent radial exposure and to get more details in the soft tissue the use of MR data for the operation planning is common. Extracting the bone surface in MR data is more difficult than in CT data because MR data has no normalized gray values. To register the ultrasound with the MR data at the knee we detected distinctive anatomic regions in the ultrasound data. We selected an adequate MR sequence in which we could segment the bone surface at the specific region. We evaluate the registration with 1000 random starting positions. 99.2% of the 1000 trails reached the optimum with an error less than 1 mm.

Modellfunktion zur Approximation von Ultraschallkontrastmittelkonzentration zur semi-quantitativen Gewebeperfusionsbestimmung

Kontrastmittelultraschall wird zur Diagnose von Tumoren der Leber oder Schlaganfallen eingesetzt. Die Eignung von Kontrastmittelultraschall zur Darstellung von Hirntumoren wurde ebenfalls bereits nachgewiesen. Eine Moglichkeit zur Auswertung ist die Approximation von Modellfunktionen, welche insbesondere den Hauptanstieg der Kontrastmittelkonzentration abbilden. In dieser Arbeit wird ein Modell zur Approximation von Kontrastmittelverlaufen in Ultraschalldaten vorgestellt, welches in der Lage ist zusatzlich zu diesem Hauptanstieg weitere Eigenschaften im Zeitverlauf, wie z.B. einen zweiten Anstieg durch Rezirkulation, abzubilden. Das Modell erreichte eine hohere Genauigkeit der Approximation als die zum Vergleich herangezogenen Modelle.

Registrierung von Knochen in 3D-Ultraschall- und CT-Daten: Vergleich verschiedener Optimierungsverfahren

Ein zentrales Problem in der bildgestutzten, navigierten Chirurgie ist die exakte Registrierung praoperativer CT/MR-Daten im Koordinatensystem des Operationssaales. Um die Nachteile einer landmarkenbasierten Registrierung zu umgehen, dient bei unserem Ansatz 3-dimensionaler, intraoperativer Ultraschall zur Registrierung. Ein automatischer Registrieralgorithmus leistet die Abbildung der Knochenstrukturen aus einem praoperativen CT in den Ultraschalldatensatz. Gegenstand unserer Untersuchungen ist die Umsetzung der Registrierung, die Robustheit des Registrieralgorithmus und die intraoperativ benotigte Rechenzeit. Das Zielgebiet ist exemplarisch die Lendenwirbelsaule; die Datensatze stammen von einem Phantom mit 3 Plastikwirbeln und von einem Patienten.

Parametrisierung evolutionärer Strategien für die Registrierung von Wirbelknochen in Ultraschall und CT-Daten

Ein zentrales Problem der bildgestutzten, navigierten Chirurgie stellt die Registrierung praoperativer Bilddaten mit dem Koordinatensystem des Operationssaals dar. Bei der rigiden Registrierung werden sechs Parameter benotigt, um die korrekte Transformation eines Koordinatensystems in ein anderes zu bestimmen. Diese Parameter werden mit einem Optimierungsverfahren ermittelt. Das Ergebnis einer Optimierung hangt von der Parametrisierung der Optimierungsstrategie ab und die Wahl dieser Hyperparameter muss in Abhangigkeit von dem zu losenden Problem getroffen werden. Anhand zweier reprasentativer realer Datensatze (jeweils 3-dimensionale CT- und Ultraschalldatensatze) wurden verschiedene Einstellungen der Hyperparameter getestet.