Sonja Kinner1,2, Tilman B Schubert1,3, Christopher Brace1, Elisabetta A Nocerino1, Timothy J Colgan1, Shannon Hynes1, Scott B Reeder1,4,5,6,7, and Emily R Winslow8
1Department of Radiology, University of Wisconsin School of Medicine and Public Health, Madison, WI, United States, 2Department of Diagnostic and Interventional Radiology and Neuroradiology, University Hospital Essen, Essen, Germany, 3Clinic for Radiology and Nuclear Medicine, Basel University Hospital, Basel, Switzerland, 4Department of Medical Physics, University of Wisconsin School of Medicine and Public Health, Madison, WI, United States, 5Department of Emergency Medicine, University of Wisconsin School of Medicine and Public Health, Madison, WI, United States, 6Department of Medicine, University of Wisconsin School of Medicine and Public Health, Madison, WI, United States, 7Department of Biomedical Engineering, University of Wisconsin School of Medicine and Public Health, Madison, WI, United States, 8Department of Surgery, University of Wisconsin School of Medicine and Public Health, Madison, WI, United States
Synopsis
In this work we describe the design and
preliminary validation of a novel MR-compatible localization device aimed at
facilitating lesion detection in ex vivo tissue, such as resected liver. Experiments
in two swine liver specimens with small microwave ablations to simulate lesions
demonstrated the successful feasibility and ease of use of this localization
device. Further evaluation in human liver tissue specimens containing
metastatic lesions will be needed to demonstrate the performance of the device
for ex vivo radiographic-pathologic correlation.Introduction
Rapid
analysis of pathologic specimen to assess tumor resection margins and tumor
burden during surgery remains a challenging task [1]. Ex vivo identification of
lesions found at in vivo imaging can be particularly demanding due to
deformation and altered orientation of the resected specimen. In order to
facilitate lesion detection and improve direct comparison of pathologic
specimens with prior in vivo imaging, we developed an MRI compatible localization
device, with MR-visible fiducial grid markings to facilitate ex vivo imaging and
radiographic-pathologic co-localization and correlation. The purpose of the
present study was to test the feasibility of this device and to evaluate the
sensitivity and specificity in liver specimens.
Methods
Imaging device description
The device is
entirely constructed from Plexiglas and designed to consist of two stationary MR
visible grids and one removable grid that together create a 3D matrix (Figure
1). Total size of the device is 27x14x14cm3. Along the width of the box, laser-etched
grid lines and labels were filled with silicone gel (Dow-Corning, Midland,
Michigan, US) to facilitate MR-visible fiducial grid marks. Etched numbers
(1-11) were oriented along the length, upper case letters (A-W) along the width,
lower case letters (a-l) along the length. All grid lines were spaced 1cm apart
(Figure 2).
When
examining an ex vivo tissue specimen, alginate can be used as fixation for the
histopathologic specimen to stabilize tissue (eg. liver) during imaging and
cutting. Once the alginate hardens to a soft solid, MR examination of the device,
containing the tissue specimen is performed. Accurate localization of lesions in
three dimensions is determined by referencing the lesions in space relative to
the three dimensions of the MR-visible grid.
After lesion
identification in MRI, the removable vertically oriented grid along the length
of the box is removed and the specimen can be sliced in a guided fashion along
the vertical grid lines (Figure 3) using prefabricated grooves spaced 1cm apart
to facilitate slicing at the identical locations corresponding to transaxial
slices aligned with one dimension of the device grid.
Imaged specimen
To test the
device, two specimens of swine liver were used for radiographic-pathologic
correlation. In each specimen, lesions were created using a random number
(0-15) of ultrasound-guided microwave ablation zones. The number and location
of the ablations were blinded to the MR-readers. Ablations were carried out in
three different sizes (5, 10 and 15 mm), based on previous calibration of the
time-energy deposition relationship.
MR-protocol
MR imaging
was performed on a clinical 3T scanner (GE Healthcare Discovery MR750,
Waukesha, WI) using an 8-channel head coil. Acquisition parameters of a 3D
spoiled gradient echo T1 weighted acquisition included: TR/TE=9.8/4.1ms, flip
angle = 20deg, BW=+/-162 kHz, FOV=280x252x180mm3, 450x420x225 matrix, for true
spatial resolution of 0.6x0.6x0.8mm3 interpolated to 0.5x0.5x0.4mm3
through zero-filling. Total scan time was 21:16min.
Imaging analysis
Two
radiologists blinded to number and localization of lesions independently
analyzed the MR images. The total numbers and coordinates of the lesions as
well as lesion diameter (measured on MR images) were reported on a spreadsheet.
Histopathological analysis and correlation
After
analysis of MR data, the specimens were sliced according to the MR-analysis
spreadsheet and lesions were measured within each slice using a ruler with 1cm
spaced etching. If lesions were identified within the coordinate-slice and not
on the cutplanes between the coordinate slices, the cut section was cut again
longitudinally. Histological findings were reported on the same spreadsheet as
the radiographic data.
Discussion
In this work we described the design, and preliminary feasibility testing and evaluation of an MR-visible device intended for radiographic-pathologic co-localization and correlation for ex vivo tissue specimens with internal lesions. Use of the device proved feasible with 100% specificity and 100% sensitivity for both readers on a per specimen and per lesion basis. 5 and 6 lesions respectively were under- or overestimated in size (+/-3mm) according to pathological evaluation.
Conclusion
The MR-visible
device was feasible for the detection and three-dimensional localization of
liver lesions, and has potential to play an important role in lesion
identification and in radiologic-pathologic evaluation. Future evaluation of
the device will include human liver tissue specimens containing metastatic
lesions.
Acknowledgements
The authors wish to acknowledge support from the NIH (UL1TR00427, K24 DK102595), the Societt of Surgeons (The Alimentary Tract Young Investigator Grant) as well GE Healthcare who provides research support to the University of Wisconsin.References
1. Juettner et al., Eur J Cardiothorac Surg. 1990