Elastographic Measurement of the Area and Volume of Thermal Lesions Resulting from Radiofrequency Ablation: Pathologic Correlation
Tomy Varghese1,
Udomchai Techavipoo1,2,
Wu Liu1,
James A. Zagzebski1,
Quan Chen1,
Gary Frank1 and
Fred T. Lee, Jr.3
1 Department of Medical Physics, The University of Wisconsin-Madison, 1530
Medical Sciences Center, 1300 University Ave., Madison, WI 53706.
2 Department of Electrical Engineering, The University of Wisconsin-Madison,
Madison, WI 53706.
3 Department of Radiology, The University of Wisconsin-Madison, Madison, WI
53706.

View larger version (21K):
[in a new window]
|
Fig. 1. Schematic diagram of electrosurgical system (model 1500, RITA
Medical Systems, Mountain View, CA) used for radiofrequency ablation
elastography. Tips of heating elements in radiofrequency ablation electrode
are equipped with thermosensors. Electrode is inserted perpendicular to
scanning plane of sonography transducer.
|
|

View larger version (69K):
[in a new window]
|
Fig. 2A. Images of radiofrequency ablation thermal lesion in section
of liver tissue encased in gelatin phantom. Sonographic images were obtained
with 3.5-MHz sector transducer. Sector B-mode gray-scale sonogram of area
treated with radiofrequency ablation. Lesion is difficult to see.
|
|

View larger version (31K):
[in a new window]
|
Fig. 2B. Images of radiofrequency ablation thermal lesion in section
of liver tissue encased in gelatin phantom. Sonographic images were obtained
with 3.5-MHz sector transducer. Two-dimensional elastogram of lesion shows
contour of ablated area (yellow outline).
|
|

View larger version (85K):
[in a new window]
|
Fig. 2C. Images of radiofrequency ablation thermal lesion in section
of liver tissue encased in gelatin phantom. Sonographic images were obtained
with 3.5-MHz sector transducer. In digitized photograph of gross pathology
specimen, contour of lesion is outlined in green.
|
|

View larger version (128K):
[in a new window]
|
Fig. 3A. Radiofrequency ablation thermal lesion in lobe of liver
tissue encased in gelatin phantom. Sonographic images were obtained with 5-MHz
linear transducer. Linear B-mode gray-scale sonogram shows increased
echogenicity near apparent ablation site and some shadowing below region of
high echogenicity. However, distinguishing boundaries (size and position) of
thermal lesion is still difficult on conventional sonogram.
|
|

View larger version (71K):
[in a new window]
|
Fig. 3B. Radiofrequency ablation thermal lesion in lobe of liver
tissue encased in gelatin phantom. Sonographic images were obtained with 5-MHz
linear transducer. On axial two-dimensional (2D) elastogram, thermal lesion
(green outline) is clearly delineated from surrounding tissue. Yellow
contour is overlay of lesion area as outlined in C.
|
|

View larger version (107K):
[in a new window]
|
Fig. 3C. Radiofrequency ablation thermal lesion in lobe of liver
tissue encased in gelatin phantom. Sonographic images were obtained with 5-MHz
linear transducer. Axial digitized photograph of lesion area (yellow
contour) in gross pathology specimen from which elastogram (B) was
generated compares well with area of lesion (green contour, B)
in elastogram. Areas on top (1.5 cm) and bottom (0.5 cm) of elastogram
correspond to gelatin, which is slightly stiffer (low-strain regions appear
darker) than normal liver tissue. Note close correspondence between lesion as
depicted in pathologic specimen photograph and on elastogram, including notch
in 11-o'clock position visible on both images.
|
|

View larger version (41K):
[in a new window]
|
Fig. 3D. Radiofrequency ablation thermal lesion in lobe of liver
tissue encased in gelatin phantom. Sonographic images were obtained with 5-MHz
linear transducer. Three-dimensional (3D) elastogram of thermal lesion was
produced by reconstructing lesion from multiple 1-mm 2D elastograms obtained
along parallel scanning planes. Surface and volume rendering of lesion is
possible because of marked contrast in stiffness between lesion and
surrounding normal tissue. To render lesion surface, we segmented 2D
elastograms by setting a single threshold (strain value 90% below maximum
strain surrounding thermal lesion) to separate background strains from lower
strains in thermal lesion. With 3D representation, lesion volume can be
calculated.
|
|

View larger version (12K):
[in a new window]
|
Fig. 4. Scatterplot of lesion widths at center planes between
pathology and elastography and their linear fits. Mean square error is 0.0515.
Dotted line at 45° denotes perfect fit between elastographically and
pathologically derived areas of lesions, whereas solid line denotes best
linear fit of data. Note that lesion widths obtained using elastography
closely correspond to widths of thermal lesions observed on pathology
specimen, with r value of 0.8693.
|
|

View larger version (11K):
[in a new window]
|
Fig. 5. Scatterplot of lesion heights at center planes between
pathology and elastography and their linear fits. Mean square error is 0.0420.
Dotted line at 45° denotes perfect fit between elastographically and
pathologically derived heights of lesions, whereas solid line denotes best
linear fit of data. Pathologic and elastrographic heights show slightly better
correlation (r = 0.8742) than was found in lesion widths, implying
better fit between data sets for heights than those for widths
(Fig. 4).
|
|

View larger version (12K):
[in a new window]
|
Fig. 6. Scatterplot of bounding ellipse areas of lesions at center
planes between pathology and elastography. Dotted line at 45° denotes
perfect fit between elastographically and pathologically derived bounding
ellipse areas of lesions, whereas solid line denotes best linear fit of data.
Areas were computed using width and depth data; correlation was r =
0.9126. Note that areas obtained using width and depth data do not take into
account errors that may be caused by shift in angle of thermal lesion along
principal axes.
|
|

View larger version (12K):
[in a new window]
|
Fig. 7. Scatterplot compares lesion areas obtained with elastography
with areas obtained from gross pathology specimens along center planes. Dotted
line at 45° denotes perfect fit between elastographically and
pathologically derived areas of lesions, whereas solid line denotes best
linear fit of data. Scatterplot and linear fit were obtained over 40
independent data sets. Mean square error is 0.1961.
|
|

View larger version (13K):
[in a new window]
|
Fig. 8. Scatterplot compares elastographic measurement of lesion
volumes with gross pathologic measurement. Dotted line at 45° denotes
perfect fit between elastographically and pathologically derived volumes of
lesions, whereas solid line indicates linear fit of data. Scatterplot and
linear fit were obtained over 40 independent data sets. Mean square error is
0.11.
|
|

CiteULike
Complore
Connotea
Del.icio.us
Digg
Reddit
Technorati What's this?
Copyright © 2003 by the American Roentgen Ray Society.