; Acoustic Radiation Force Impulse _ARFI_ Imaging of the
Documents
Resources
Learning Center
Upload
Plans & pricing Sign in
Sign Out

Acoustic Radiation Force Impulse _ARFI_ Imaging of the

VIEWS: 34 PAGES: 4

  • pg 1
									 Acoustic Radiation Force Impulse (ARFI) Imaging
           of the Gastrointestinal Tract
                        Mark Palmeri, Kristin Frinkley, Liang Zhai, Rex Bentley, Kirk Ludwig,
                                     Marcia Gottfried, and Kathryn Nightingale

                                            Duke University, Durham, NC 27708
                                                  mark.palmeri@duke.edu


   Abstract— Currently, the evaluation of lesions in the gastroin-   trointestinal (GI) tract are most commonly imaged using
testinal (GI) tract using ultrasound suffers from poor contrast      endoscopic ultrasound (EUS) [3], [8]; however, EUS can suffer
between healthy and diseased tissue. Acoustic Radiation Force        from poor contrast between healthy and diseased tissues in
Impulse (ARFI) imaging provides information about the me-
chanical properties of tissue using brief, high-intensity, focused   the GI tract. The organs of the GI tract are layered; from
ultrasound to generate radiation force, and conventional, ultra-     the lumen outward, these layers consist of the mucosa (1-2
sonic, correlation-based methods to track tissue displacement.       mm thick), submucosa (∼2 mm thick), muscularis propria (2-
Using conventional linear arrays, ARFI imaging has shown             3 mm thick), and the adventitia. GI cancers are staged by
improved contrast over B-mode images when applied to solid           degree of tumor invasion (T-staging) into the layers of the
masses in the breast and liver. The purpose of this work is
to (1) demonstrate that ARFI imaging can be performed with           bowel wall, along with determining the potential spread of
an endocavity probe, and (2) demonstrate that ARFI imaging           malignant cells to local lymph nodes (N-staging). The most
can provide improvements over conventional B-mode imaging of         critical staging distinction must be made between stage uT2
GI lesions. An EC94, 6.2 MHz, endocavity probe was modified           and uT3 tumors, which indicates whether or not a tumor has
to perform ARFI imaging in tissue-mimicking phantoms using           penetrated through the muscularis propria. For rectal cancers,
a Siemens SONOLINE AntaresTM scanner. ARFI imaging was
                                                                     this distinction determines whether a patient will undergo local
performed on fresh, surgically-excised, GI lesions using a 75L40,
7.2 MHz, linear array on a modified Siemens SONOLINE                  tumor excision, or neoadjuvant chemotherapy and radiation,
ElegraTM scanner. The endocavity probe created ARFI images           followed by radical resection [1]. The latter treatment option
to a depth of over 2 cm in tissue-mimicking phantoms, with           is associated with greater risks to the patient, including infec-
maximum displacements of 5 µm. The endocavity probe did not          tion, bleeding, decreased colonic motility, and incontinence.
heat appreciably during ARFI imaging, demonstrating that the         Treating a stage uT2 tumor as a uT3 tumor unnecessarily
probe’s small size will not limit in vivo ARFI imaging. ARFI im-
ages of an adenocarcinoma of the gastroesophageal (GE) junction,     subjects a patient to these risks; however, staging uT2 cancers
status-post chemotherapy and radiation treatment, demonstrate        as uT3 occurs in 10-35% of EUS scans [2], [5], leading
better contrast between healthy and fibrotic/malignant tissue         to overly-aggressive medical treatments and surgeries. The
than standard B-mode images. ARFI images of healthy gastric,         National Cancer Institute has established a research priority to
esophageal, and colonic tissue specimens differentiate normal        improve the staging of colorectal tumors to more accurately
anatomic tissue planes (i.e., mucosal, muscularis, and adventitial
layers), as confirmed by histologic evaluation. ARFI imaging of an    guide treatment decisions [6]. Applying ARFI imaging to the
ex vivo colon cancer portrays interesting contrast and structure     GI tract may provide such an improvement by more clearly
not present in B-mode images. These findings support the clinical     delineating healthy tissue layers and the penetration of tumors
feasibility of endoscopic ARFI imaging to guide diagnosis and        into those tissue layers.
staging of disease processes in the GI tract.
                                                                                             II. M ETHODS
            I. I NTRODUCTION & BACKGROUND                            A. Experimental
   Acoustic Radiation Force Impulse (ARFI) imaging is one of            ARFI imaging of freshly-excised, surgical specimens was
many acoustic radiation force-based imaging modalities being         performed using a Siemens 75L40 linear array (7.2 MHz,
studied to generate images of the mechanical properties of           F/1.3 focal configuration) on a Siemens SONOLINE ElegraTM
tissue [9]. ARFI imaging uses short-duration (< 100 µs), high-       Scanner (Siemens Medical Solutions USA, Inc., Ultrasound
intensity, acoustic pulses to generate localized, micron-scale       Division, Issaquah, WA). The system was modified to control
displacements in tissue, and these displacements are tracked         acoustic beam sequences and intensities, and raw Radio Fre-
using ultrasonic, correlation-based methods [10]. Images of          quency (RF) data was stored and processed off-line. Data was
two-dimensional regions of interest are generated by sequen-         acquired sequentially, as is done in conventional ultrasound
tially interrogating multiple lateral locations, as is done in       imaging. At each lateral location, a reference tracking beam
color Doppler imaging.                                               was fired, followed by a high-intensity pushing beam that
   Potentially malignant lesions and lymph nodes of the gas-         displaced the tissue. A series of tracking beams followed the
pushing beam to track the tissue recovery for 4 ms with a
                                                                                                                                             Mucosa −>
Pulse Repetition Frequency (PRF) of 5632 Hz.
                                                                                                                                     Submucosa −>
   The transducer was held in a fixed ring-stand, while the sur-                                                                                             MP

gical specimens were placed on top of a water bag. Ultrasonic
gel was used to couple the transducer to the specimens. When                                                             Fibrosis         Muscularis Propria (MP)

possible, surgical specimens were marked with dissection ink
so that histologic sections could be made in the imaging plane.                                                                              Adventitia

   ARFI imaging was also performed using a Siemens EC94
endocavity probe (6.2 MHz, F/1.5 focal configuration, 30
mm elevation focus) on a Siemens SONOLINE AntaresTM                                    5




                                                                         Depth (mm)
Scanner. Imaging was performed on a gelatin-based, tissue-                            10

mimicking phantom containing a stiffer, spherical inclusion.                          15

The fabrication of such phantoms is outlined by Hall et al. [4].                      20
                                                                                        0       10   20      30      40      50     60      70
                                                                                                          Lateral Position (mm)
B. Finite Element Modeling
                                                                                                                                                           20
   Finite element models were developed to evaluate ARFI                               5    MP




                                                                         Depth (mm)
                                                                                            MP
imaging performance for different transducers, system param-                          10                                                                   10
                                                                                            A
eters, and tissue mechanical and acoustic properties [11]. Field                      15

II (http://www.es.oersted.dtu.dk/staff/jaj/field/) [7] was used                        20
                                                                                                                                                           0
                                                                                        0       10   20      30      40      50     60      70
to solve for the three-dimensional, high-intensity, focused,                                              Lateral Position (mm)
acoustic fields generated during ARFI imaging. A rectilinear
                                                                                                                                                           1.2
mesh with uniform 0.2 mm node spacing was generated using                              5                                                                   1

                                                                         Depth (mm)
LS-PREPOST (Livermore Software Technology Corporation,                                10
                                                                                                                                                           0.8
                                                                                                                                                           0.6
Livermore, CA). The mesh material properties were defined                              15                                                                   0.4
                                                                                                                                                           0.2
to simulate a three-layer, elastic structure, as specified in                          20
                                                                                        0       10   20      30      40      50     60      70
Table I, with the Young’s moduli being rough estimates of                                                 Lateral Position (mm)
the stiffnesses of mucosal, muscularis, and adventitial layers.
No slippage was allowed between the material layers in this                            5                                                                   1.5
                                                                         Depth (mm)




preliminary model.                                                                    10                                                                   1
                                                                                      15
                                                                                                                                                           0.5
                             TABLE I                                                  20
                                                                                        0       10   20      30      40      50     60      70
          F INITE E LEMENT M ODEL M ATERIAL P ROPERTIES                                                   Lateral Position (mm)
          Depth from   Young’s Modulus    Poisson’s Ratio
         Lumen (mm)         (kPa)                                  Fig. 1. From top to bottom: Photo of gross tissue specimen (top left); H&E
            0-1               1.0             0.499                stain of esophageal side of the ulceration dissected in the imaging plane,
            1-5              16.0             0.499                as indicated by the break in the dashed line in the gross photo (top right);
           5 - 25             4.0             0.499                Matched B-mode image; ARFI maximum displacement image (µm); ARFI
                                                                   time-to-peak displacement image (ms); and ARFI recovery time image (ms)
                                                                   of an ex vivo, status-post chemotherapy and radiation, adenocarcinoma in the
                                                                   gastroesophageal junction. The histology image demonstrates disruption of the
                III. R ESULTS & D ISCUSSION                        mucosal, submucosal and muscular layers by stiffer, fibrotic scar tissue. Note
                                                                   the boundary definition in the three visualized layers of the stomach (left) in
   Fig. 1 shows an ex vivo gastroesophageal (GE) junction          the ARFI displacement image. Note also the thin black line separating the
with an ulcerated adenocarcinoma, status-post chemotherapy         peri-adventitial tissue from the bottom layer of the muscularis propria. The
                                                                   top gel bridge and bottom gauze have been masked in black.
and radiation treatment. ARFI images were created down the
dashed, black line in the upper-left image (Fig. 1), with the
stomach on the left-hand side of the ARFI images, and the
esophagus on the right-hand side of the images. Histologic         improved delineation of the tissue layers in the ARFI image.
correlation was provided in the imaging plane along the break      The bottom row shows an enlarged region of the upper right
in the dashed line. Notice that in the healthy stomach, the        ARFI displacement image (left), and the normalized axial
anatomical layers of the gastric tissue can be delineated (left    displacements averaged across the region indicated by the
side, gray layers). These layers become interrupted in the         vertical dashed lines (right). The regions of decreased displace-
ulcerated region by stiffer, fibrotic tissue, which then gives      ment in the submucosal layer (15 - 17 mm) may correspond
way to healthy, esophageal tissue on the right. These layers       to stiffer lymphoid aggregates. Better layer distinction with
can be visualized in the ARFI images, as confirmed by the           ARFI imaging may allow for more accurate staging of tumor
histology image.                                                   invasion.
   The top row of Fig. 2 shows a comparison of B-mode                 Fig. 3 shows B-mode and ARFI images for an ex vivo cecal
and ARFI images for normal, proximal colon tissue, with            colon with a uT3/N- tumor, for three different locations in the
                              B−mode              ARFI Displacement (0.7 ms)       Fig. 4 shows matched B-mode and ARFI displacement
      Depth (mm)   15                             15                             images of a breast lymph node. Note the detailed structure
                   20                             20                             of the hilum and efferent duct present in the ARFI image
                                                                                 (white arrow). The ability to image structural and material
                    −10         0         10       −10         0         10
                      Lateral Position (mm)          Lateral Position (mm)       properties of lymph nodes may allow ARFI imaging to help
                   ARFI Displacement (t=0.7 ms)
                                                                                 with N-staging of lymph nodes.
                   14                             14                                                                                                                        Max Disp
                                                                                                                          B−mode                                            (~5 µm)
                   15                             15                                                        4                                                  4
                                                                                                            6                                                  6
                   16                             16
                                                                                                            8                                                  8
      Depth (mm)




                   17                             17                                                  10                                                      10




                                                                                      depth, (mm)
                                                                                      Depth (mm)
                                                                                                      12                                                      12
                   18                             18                                                  14                                                      14
                                                                                                      16                                                      16
                   19                             19
                                                                                                      18                                                      18
                   20                             20                                                  20                                                      20
                                                                                                      22                                                      22
                   21                             21                                                    −10          −5       0       5           10             −10   −5      0           5   10
                                                                                                                                   lateral position, (mm)
                   22                             22                                                                                           Lateral Position (mm)
                        −5        0        5        0         50        100
                        Lateral Position (mm)       ARFI Displacement (µm)
                                                                                 Fig. 4. B-mode (left) and matched ARFI maximum displacement image
Fig. 2. The top row shows matched B-mode (left) and ARFI displacement            (right) of an in vivo, biopsy-proven reactive lymph node in the breast. The
(right) images of a normal section of proximal colon. The ARFI image was         lymph node is the darker oval region in the B-mode image appearing from 12-
0.7 ms after excitation. The lower left image shows an enlarged region of        22 mm in depth, with a central, more echoic region. The echogenic center in
the upper right ARFI displacement image. The lower right image shows a           the B-mode image is a sign of normal hilar structure. The hilum corresponds
normalized axial displacement profile, averaged across the region indicated       to a region of decreased displacement in the ARFI image (darker region). The
by the vertical dashed lines. Notice the greater structural contrast between     node itself appears slightly stiffer (darker) than the surrounding tissue, with
layers in the ARFI images compared with the B-mode image. The mucosa             a ductal, stiff structure (arrow) extending from the dark core to the edge of
spans from 14 - 15 mm, the submucosa from 15 - 17 mm, and the muscularis         the node.
propria from 17 - 19 mm, followed by deeper adventitial tissue. Also notice
regions of increased stiffness in the submucosal layer that may correspond to       Fig. 5 shows ARFI images generated with the Siemens
stiffer lymphoid aggregates.                                                     EC94 endocavity probe in a gelatin-based, tissue-mimicking
                                                                                 phantom. These images demonstrate that smaller, endocavity
                                                                                 arrays can deliver enough energy to generate ARFI images.
tumor. Notice that the ARFI images show much better contrast
                                                                                 There was not appreciable heating of the endocavity probe
between structures than the B-mode images.
                                                                                 during these experiments, further supporting the feasibility of
                                                                                 in vivo ARFI imaging.
                              B−mode               ARFI Disp (t=0.7ms,~5µm)
                    5                              5
                                                                                                                                              Displacement                     Recovery
                   10                             10                                                                 B−mode                 (~5 µm, t=0.7ms)                  (t = 1.3 ms)
                   15                             15
                                                                                         Axial Depth (mm)




                                                                                                            15
                   20                             20
                    −10          0         10      −10         0         10
                    5                              5
      Depth (mm)




                                                                                                            20
                   10                             10
                   15                             15
                                                                                                            25
                   20                             20
                    −10          0         10      −10         0         10
                    5                              5                                                            −5        0        5     −5       0       5 −5                         0       5
                                                                                                                                        Lateral Position (mm)
                   10                             10
                   15                             15                             Fig. 5.     Matched B-mode and ARFI images of a gelatin-based, tissue-
                                                                                 mimicking phantom with a 5 mm spherical inclusion that is approximately
                   20                             20                             three times stiffer than the background material. These images were generated
                    −10         0         10       −10         0         10
                      Lateral Position (mm)          Lateral Position (mm)       using the Antares scanner and an EC94, 6.2 MHz, end-looking, endocavity
                                                                                 transducer.
Fig. 3. B-mode (left) and ARFI (right) images of an excised uT3/N- colon
tumor near the cecal pouch. The top row of images show stiff tumor tissue           Fig. 6 shows simulated displacement profiles along the
surrounded by softer tumor tissue that extends down to a very soft layer         axial-lateral plane, centered in elevation in homogeneous (E =
of presumed adventitial fat. These images were acquired over a region of
palpably stiff and tethered tumor. Notice the surrounding tumor has replaced
                                                                                 4.0 kPa, left) and layered materials (Table I, right). In contrast
normal tissue layers of the colon, shown in Fig. 2. The middle images show       to 2-D ARFI images, these images show the 2-D displace-
tumor that disrupts a normal dark band of tissue (possibly the distal boundary   ment field generated by a single ARFI push with subsequent
of the muscularis propria) at a depth of ∼14 mm. Again, the normal layers of
the colon have been replaced by tumor tissue. The bottom images demonstrate
                                                                                 shear wave propagation at later times. Notice the decreased
structural features in the ARFI image not present in the B-mode image in         displacement confined to the stiffer, middle layer (right),
regions where tumor and healthy tissue may be interspersed.                      indicating that ARFI imaging could provide information about
the location of layers with different material properties. Also                                                  R EFERENCES
notice that through time, shear waves propagate away from the                       [1] D. Blumberg and R. Ramanathan. Treatment of colon and rectal cancer.
region of excitation (ROE) with a faster velocity in the stiffer                        J Clin Gastroenterol, 34(1):15–26, 2002.
(middle layer) material than the softer surrounding material.                       [2] J. Garcia-Aguilar, J. Pollack, S. Lee, E. Hernandez de Anda, A. Mell-
                                                                                        gren, W. Wong, C. Finne, D. Rothenberger, and R. Madoff. Accuracy
This highlights another metric that ARFI imaging may use to                             of endorectal ultrasonography in preoperative staging of rectal tumors.
generate images delineating the tissue layers of the GI tract.                          Dis. Colon Rectum, 45(1):10–15, 2002.
                                                                                    [3] P. Gibbs, M. Chao, and J. Tjandra. Optimizing the outcome for patients
                                                                                        with rectal cancer. Dis. Colon Rectum, 46(3):389–402, 2003.
                                                                                    [4] Hall T.J., Bilgen M., Insana M.F., and Krouskop T.A. Phantom materials
                    5                               5                                   for elastography. IEEE Transactions on Ultrasonics, Ferroelectrics and
                   10                              10                                   Frequency Control, 44(6):1355–65, 1997.
                                                                                    [5] F. Hulsmans, T. Tio, P. Fockens, A. Bosma, and G. Tytgat. Assessment
                   15                              15                                   of tumor infiltration depth in rectal cancer with transrectal sonography:
                                                                                        Caution is necessary. Radiology, 190(3):715:720, 1994.
      Depth (mm)




                                                                                    [6] N. C. Institute, O. of Science Planning, and Assess-
                    5                               5                                   ment.         Colorectal cancer, progress review groups, 2004.
                                                                                        http://prg.nci.nih.gov/colorectal/default.html.
                   10                              10
                                                                                    [7] J. Jensen and N. Svendsen. Calculation of pressure fields from arbitrarily
                   15                              15                                   shaped, apodized, and excited ultrasound transducers. IEEE Trans.
                                                                                        Ultrason., Ferroelec., Freq. Contr., 39(2):262–267, 1992.
                                                                                    [8] D. Malfair, A. Jacqueline, and P. Phang. Preoperative rectal cancer
                    5                               5                                   imaging. BC Medical Journal, 45(6):259–261, 2003.
                                                                                    [9] K. Nightingale, M. Palmeri, R. Nightingale, and G. Trahey. On the
                   10                              10
                                                                                        feasibility of remote palpation using acoustic radiation force. J. Acoust.
                   15                              15                                   Soc. Am., 110(1):625–634, 2001.
                    −5             0           5    −5             0           5   [10] K. Nightingale, M. Soo, R. Nightingale, and G. Trahey. Acoustic
                         Lateral Position (mm)           Lateral Position (mm)
                                                                                        radiation force impulse imaging: In vivo demonstration of clinical
                                                                                        feasibility. Ultrasound Med. Biol., 28(2):227–235, 2002.
Fig. 6.     Simulated, normalized axial displacement fields generated from          [11] M. Palmeri and K. Nightingale. A finite element method model of
a single ARFI excitation on the axial-lateral plane, centered in elevation,             soft tissue response to impulsive acoustic radiation force. IEEE Trans.
comparing a homogeneous medium (4 kPa, left side) and the layered model                 Ultrason., Ferroelec., Freq. Contr., in review.
(right side, layer interfaces indicated by the dashed lines). The rows indicate
displacements at 5 ms, 10 ms, and 15 ms post-excitation in the top, middle,
and bottom rows, respectively. Notice the decreased displacement within
the middle layer, in the right-hand column, as compared with the softer,
homogeneous model (left column). Also notice the faster shear wave velocity
in the stiffer, middle layer as compared with the softer surrounding layers.



                                      IV. C ONCLUSION

   ARFI imaging can be used to delineate the normal tissue
layers of the GI tract in esophageal, gastric, and colonic
tissues. ARFI images correlate well with histology in imaging
spatial extent of stiff, fibrotic scar tissue in the GE junction.
Furthermore, ARFI images show the boundaries of colonic
tumors, providing a potential mechanical metric to improve
ultrasonic tumor staging (T-staging). Compared with B-mode
imaging, ARFI imaging more clearly illustrates internal struc-
tural information in lymph nodes, with the clinical motivation
of improving N-staging. ARFI imaging has been implemented
on an endocavity probe. Finally, FEM models demonstrate that
ARFI imaging is capable of distinguishing layers of varying
material properties through both displacement maps and shear
wave velocities. Overall, ARFI imaging appears promising for
improving the staging of GI tumors and lymph nodes.

                                 V. ACKNOWLEDGMENTS

   This work was supported by NIH grant 8 R01 EB002132,
the Whitaker Foundation, and the Medical Scientist Training
Program grant T 32 GM-07171. We thank Dr. Gregg Trahey
for his valuable insight, and Siemens Medical Solutions USA,
Inc., Ultrasound Division for their technical assistance.

								
To top