Artikel
Experimental evaluation of accuracy of 3D-reconstructed navigated ultrasound compared to intraoperative MRI
Suche in Medline nach
Autoren
Veröffentlicht: | 8. Juni 2016 |
---|
Gliederung
Text
Objective: Currently, intraoperative MRI(iMRI) is the “gold standard” of intraoperative imaging in glioma surgery. In addition, intraoperative navigated 3D-reconstructed ultrasound (iUS) is another fast viable, inexpensive and easily available possibility of intraoperative imaging with potentially similar results. Aim of the study was to evaluate accuracy of a linear (liUS) and a sector array ultrasound (siUS) probe and their 3D-reconstruction in comparison to iMRI based on measurements of a gel phantom.
Method: We developed an agarose gel phantom containing 3D-objects, visible in iMRI and US. All of the objects had the same measures and were arranged in a defined angle towards each other to detect 2-point-discrimination. We performed 56 navigated ultrasound sweeps in a predefined depth in perpendicular (per) and parallel (par) direction referred to the orientation of the objects and iMRI imaging of the phantom. Differences of reconstructed 3D-objects among the three imaging methods were evaluated in the navigation software and also compared to 2D-US resolution.
Results: Imaging error in 3D reconstruction of iUS shows a correlation between sweep direction and results in siUS and liUS. 2D-images have a significantly (p < 0.01) worse resolution than 3D-reconstructed images with its image plane perpendicular to the fissure, whereas there is no significant deviation from results with 3D-reconstructed parallel US images. Reconstructed image sets of siUS showed significantly worse resolution (siUS per: 0,6mm; siUS par: 1.2mm; siUS 2D: 1.1mm) than iMRI and liUS in all sweep directions. While liUS has a significantly (p < 0.01), better resolution than iMRI in reconstructed objects with an image plane perpendicular to sweep (iMRI: 0.2mm; liUS per: 0.1mm). In parallel sweep direction, iMRI is significantly (all p < 0.01) better for length, width and volume, but there is no significant difference in resolution (liUS par: 0.3mm, iMRI 0.2mm).
Conclusions: In an experimental setting, image quality and resolution of liUS shows similar results as iMRI for 3D-reconstructed images, depending on sweep directions, while simple 2D-images have a significant lower resolution than iMRI and 3D-reconstructed liUS. siUS transducers have a significant lower resolution than both, iMRI and liUS in 2D and 3D reconstruction. Therefore, 3D liUS might be an easily available alternative to modern iMRI imaging in intracranial neurosurgery.