Three male patients underwent 1.5 Tesla liver MRI.
Real-time image fusion of US and MRI data modalities was performed by an US system (MyLab Twice,
Esaote S.p.A.
Italy),
equipped with a dedicated Virtual Navigator Fusion Imaging tool and QElaXto pSWE technology (Esaote S.p.A.,
Genova,
Italy).
An abdominal multimodality phantom (Model 057 - CIRS - Computerized Imaging Reference Systems Inc.,
Norfolk,
Virginia,
USA),
was used regarding real-time fusion imaging and virtual biopsy.
Fusion Imaging
MRI data were transferred in DICOM format to the US system query/retrieving hospital PACS system (IMPAX 6,
Agfa Healthcare NV,
Mortsel,
Belgium).
An US convex Array Probe (Operating Bandwidth: 1- 8 MHz; CFM-PW Frequencies: 1.9-2.1-2.3–2.6–3.3 MHz,
CA541,
Esaote) and a reusable tracking bracket with sensor mounted (639-041,
Civco Medical Solutions,
Kalona,
Iowa,
USA),
were used.
Virtual Navigator fusion procedures were allowed by an electromagnetic tracking system,
composed by a transmitter and a small receiver,
mounted on the US probe.
The transmitter’s position,
which is the origin of the reference system,
was fixed by a support and the receiver provided the position and the orientation of the US probe in relation to the transmitter in the created 3D space.
The electromagnetic field source tip was oriented to point the phantom and/or patient’s interested area,
in order to address the highest intensity of the created field to the US scanning area.
A non-metallic table was used to reduce interferences as much as possible.
Before starting,
a check of the accuracy of the electromagnetic field was performed: the same point coordinates were measured twice by a dedicated registration pen with the electromagnetic sensor mounted in two different spatial orientations.
An accuracy of 0.2 cm or less was considered acceptable.
One plane registration procedure
After importing the MRI data on the US system,
the system was ready to start the fusion procedure between MRI and real-time US data.
One plane registration was performed selecting the same plane in axial view both on US scan and on MRI dataset.
After this selection,
the system roughly registered the two imaging modalities.
Therefore,
moving the US probe,
real-time US scans of the phantom and simultaneous navigation within its MRI volume were achieved.
This procedure is mandatory in order to give the US system the information about the examined phantom orientation and position within the electromagnetic field and respecting the second modality volume dataset (MRI).
In vivo registration procedure for fusion of MRI and US dataset was carried out similarly.
In case of real patient this operation had to be performed in the same patient’s respiratory phase of the MRI acquisition,
in order to reduce the possible sum of errors.
One point registration procedure
After the one plane registration,
the one point registration procedure is performed: it consisted in real-time selection of the same point on US scan and on MRI volume dataset.
During in vitro tests,
the region of interest was selected moving the probe on the phantom and setting the same point on US scan and MRI visible lesion-like target.
In vivo tests were carried out similarly,
selecting the same anatomical point on the US scan and on the MRI dataset,
i.e.
vessels,
trying to pick a target where the hepatic vascular pattern presented a bifurcation.
One point registration corrected the spatial error in the three coordinates (X,
Y,
Z),
even if the operator’s error,
related to the point identification/selection accuracy,
and the initial angular error,
related to the probe position (always considering the same patient’s respiratory phase – expiration),
were still present.
Automatic Registration
An Automatic Registration Algorithm was used for in-vivo tests.
The Autoregistration algorithm works simultaneously on the US and the second imaging modality volume dataset.
It is based on automatically matching the vessels visible in both modalities.
In order to do this,
both volumes have to be translated in a common coordinated system.
After the one point registration,
the Automatic registration algorithm was used,
in order to fine tune the remaining spatial and angular error.
Automatic registration with hepatic anatomical markers (vascular tree) was carried out acquiring a US three-dimensional Power Doppler volume of the hepatic vascular tree,
that is in the same selected region of interest of the one point registration procedure.
Then the data were matched automatically by the US system and finally it showed the US Power Doppler volume together with the MRI segmentation.
Spatial error between US and MRI scans was evaluated both for in-vivo and in-vitro tests.
Visual control of the correspondence of anatomical structures on US and MRI in axial,
coronal and sagittal views during fusion navigation and measurement of the same anatomical point,
i.e.
portal vein bifurcation,
were used as assessment of the error of the registration procedure.
The distance of the same anatomical point visualized with the two modalities was measured in the sagittal,
axial and coronal views and the Euclidean distance was computed.
Point Shear Wave Elastography
Point Shear Wave Elastography (pSWE) technology is characterized by the generation of a shear wave originated by an Ultrasound focused beam,
which creates a localized perturbation in a small region around a focused shock point.
This perturbative phase is followed by a reading phase in a Region of Interest (ROI) providing a quantitative estimate of liver stiffness inside the ROI.
The ROI is therefore the graphical representation of the considered reading area where the shear wave propagates and where the quantitative analysis is performed.
QElaXto is one-dimensional with small ROI and some special features to facilitate the data analysis,
in particular the 3D eWave tool, Shear Wave Quality Graph for immediate feedback about measurement quality: it’s a 3D histogram qualitative display of the propagation of the perturbation Space–Time-Displacement (blue: low,
red: high).
Virtual Biopsy
A phantom test was also performed for the assessment of Virtual Biopsy feasibility,
using a triple modality 3D abdominal phantom.
Proper target signs were automatically saved for pSWE measurement spatial indication.
The biopsy was planned and performed using the Virtual Biopsy tool on both the 2D scan and the MRI fusion.
Needle insertions were performed in plane and out of plane: proper graphical indications of the correct or incorrect trajectory of the needle were given to the operator in real-time.
In our tests,
once the planning path and insertion point were decided,
the operator switched off the US reference image inserting the needle,
following only the Virtual Biopsy graphical indications (the system gives to the operator also the distance measure in cm from the target of the needle tip).
The US reference image was switched-on again only when the target was hit,
according to the Virtual Biopsy graphical indications.