UZH-Logo

Maintenance Infos

Comparison of different reconstruction algorithms for three-dimensional ultrasound imaging in a neurosurgical setting


Miller, D; Lippert, C; Vollmar, V; Bozinov, O; Benes, L; Schulte, D M; Sure, U (2012). Comparison of different reconstruction algorithms for three-dimensional ultrasound imaging in a neurosurgical setting. International Journal of Medical Robotics and Computer Assisted Surgery, 8(3):348-359.

Abstract

Background Freehand three-dimensional ultrasound imaging (3D-US) is increasingly used in image-guided surgery. During image acquisition, a set of B-scans is acquired that is distributed in a non-parallel manner over the area of interest. Reconstructing these images into a regular array allows 3D visualization. However, the reconstruction process may introduce artefacts and may therefore reduce image quality. The aim of the study is to compare different algorithms with respect to image quality and diagnostic value for image guidance in neurosurgery. Methods 3D-US data sets were acquired during surgery of various intracerebral lesions using an integrated ultrasound-navigation device. They were stored for post-hoc evaluation. Five different reconstruction algorithms, a standard multiplanar reconstruction with interpolation (MPR), a pixel nearest neighbour method (PNN), a voxel nearest neighbour method (VNN) and two voxel based distance-weighted algorithms (VNN2 and DW) were tested with respect to image quality and artefact formation. The capability of the algorithm to fill gaps within the sample volume was investigated and a clinical evaluation with respect to the diagnostic value of the reconstructed images was performed. Results MPR was significantly worse than the other algorithms in filling gaps. In an image subtraction test, VNN2 and DW reliably reconstructed images even if large amounts of data were missing. However, the quality of the reconstruction improved, if data acquisition was performed in a structured manner. When evaluating the diagnostic value of reconstructed axial, sagittal and coronal views, VNN2 and DW were judged to be significantly better than MPR and VNN. Conclusion VNN2 and DW could be identified as robust algorithms that generate reconstructed US images with a high diagnostic value. These algorithms improve the utility and reliability of 3D-US imaging during intraoperative navigation.

Background Freehand three-dimensional ultrasound imaging (3D-US) is increasingly used in image-guided surgery. During image acquisition, a set of B-scans is acquired that is distributed in a non-parallel manner over the area of interest. Reconstructing these images into a regular array allows 3D visualization. However, the reconstruction process may introduce artefacts and may therefore reduce image quality. The aim of the study is to compare different algorithms with respect to image quality and diagnostic value for image guidance in neurosurgery. Methods 3D-US data sets were acquired during surgery of various intracerebral lesions using an integrated ultrasound-navigation device. They were stored for post-hoc evaluation. Five different reconstruction algorithms, a standard multiplanar reconstruction with interpolation (MPR), a pixel nearest neighbour method (PNN), a voxel nearest neighbour method (VNN) and two voxel based distance-weighted algorithms (VNN2 and DW) were tested with respect to image quality and artefact formation. The capability of the algorithm to fill gaps within the sample volume was investigated and a clinical evaluation with respect to the diagnostic value of the reconstructed images was performed. Results MPR was significantly worse than the other algorithms in filling gaps. In an image subtraction test, VNN2 and DW reliably reconstructed images even if large amounts of data were missing. However, the quality of the reconstruction improved, if data acquisition was performed in a structured manner. When evaluating the diagnostic value of reconstructed axial, sagittal and coronal views, VNN2 and DW were judged to be significantly better than MPR and VNN. Conclusion VNN2 and DW could be identified as robust algorithms that generate reconstructed US images with a high diagnostic value. These algorithms improve the utility and reliability of 3D-US imaging during intraoperative navigation.

Citations

2 citations in Web of Science®
3 citations in Scopus®
Google Scholar™

Altmetrics

Downloads

172 downloads since deposited on 04 Sep 2012
43 downloads since 12 months
Detailed statistics

Additional indexing

Item Type:Journal Article, refereed, original work
Communities & Collections:04 Faculty of Medicine > University Hospital Zurich > Clinic for Neurosurgery
Dewey Decimal Classification:610 Medicine & health
Language:English
Date:29 February 2012
Deposited On:04 Sep 2012 12:32
Last Modified:05 Apr 2016 15:57
Publisher:Wiley-Blackwell
ISSN:1478-5951
Publisher DOI:https://doi.org/10.1002/rcs.1420
PubMed ID:22374854
Permanent URL: https://doi.org/10.5167/uzh-64516

Download

[img]
Preview
Filetype: PDF
Size: 1MB
View at publisher
[img]
Preview
Filetype: PDF
Size: 130kB

TrendTerms

TrendTerms displays relevant terms of the abstract of this publication and related documents on a map. The terms and their relations were extracted from ZORA using word statistics. Their timelines are taken from ZORA as well. The bubble size of a term is proportional to the number of documents where the term occurs. Red, orange, yellow and green colors are used for terms that occur in the current document; red indicates high interlinkedness of a term with other terms, orange, yellow and green decreasing interlinkedness. Blue is used for terms that have a relation with the terms in this document, but occur in other documents.
You can navigate and zoom the map. Mouse-hovering a term displays its timeline, clicking it yields the associated documents.

Author Collaborations