Segmentierung des Femurs aus MRT-Daten mit Shape-Based Level-Sets

Dekomien, Claudia; Busch, M.; Teske, Wolfram; Winter, Susanne

doi:10.1007/978-3-540-93860-6_65

Cited by 1 publication

(3 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In this work, we evaluated only one dataset; however, we expect the results to be reproducible on different data. Previous work on segmentation of vertebrae in CT data [7] and knee in MRI data [9] showed that our approach generalizes well for multiple datasets.…”

Section: Discussionmentioning

confidence: 71%

“…In contrast to the approach of Tsai et al [20], the covariance matrix adaptation evolutionary strategy (CMA-ES) [9] was used to optimize the parameters w and p in the functions F MRI and F CT . The CMA-ES is an adaptive algorithm that is less prone to get stuck in local optima than gradient-based optimization algorithms [12].…”

Section: Preoperative Data Processingmentioning

confidence: 99%

“…Additionally, the manual marking of all the landmarks is a time-consuming step. To avoid these drawbacks, we developed a method that uses intraoperative ultrasound to visualize the bone surface in real time, and uses this information to perform a precise registration of preoperative computed tomography (CT) or magnetic resonance imaging (MRI) data [4,9]. It was already demonstrated, both in phantoms and in vivo, that the algorithm achieves high registration success rates with acceptable computation times using realistic starting deviations [23].…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

System architecture for intraoperative ultrasound registration in image-based medical navigation

Dekomien¹,

Roeschies²,

Winter³

2012

Biomedizinische Technik/Biomedical Engineering

Self Cite

View full text Add to dashboard Cite

Medical navigation systems for orthopedic surgery are becoming more and more important with the increasing proportion of older people in the population, and hence the increasing incidence of diseases of the musculoskeletal system. The central problem for such systems is the exact transformation of the preoperatively acquired datasets to the coordinate system of the patient's body, which is crucial for the accuracy of navigation. Our approach, based on the use of intraoperative ultrasound for image registration, is capable of robustly registering bone structures for different applications, e.g., at the spine or the knee. Nevertheless, this new procedure demands additional steps of preparation of preoperative data. To increase the clinical acceptance of this procedure, it is useful to automate most of the data processing steps. In this article, we present the architecture of our system with focus on the automation of the data processing steps. In terms of accuracy, a mean target registration error of 0.68 mm was achieved for automatically segmented and registered phantom data where the reference transformation was obtained by performing point-based registration using artificial structures. As the overall accuracy for subject data cannot be determined non-invasively, automatic segmentation and registration were judged by visual inspection and precision, which showed a promising result of 1.76 mm standard deviation for 100 registration trials based on automatic segmentation of magnetic resonance imaging data of the spine.

show abstract

Section: Discussionmentioning

confidence: 71%