Download The Potential Use of Ultrasound-Magnetic Resonance Imaging

TECHNICAL INNOVATION
The Potential Use of UltrasoundMagnetic Resonance Imaging
Fusion Applications in
Musculoskeletal Intervention
Christopher J. Burke, MBChB, Jenny Bencardino, MD, Ronald Adler, MD, PhD
Videos online at
http://wileyonlinelibrary.com/journal/jum
We sought to assess the potential use of an application allowing real-time ultrasound
spatial registration with previously acquired magnetic resonance imaging in musculoskeletal procedures. The ultrasound fusion application was used to perform a
range of outpatient procedures including piriformis, sacroiliac joint, pudendal and
intercostal nerve perineurial injections, hamstring-origin calcific tendonopathy barbotage, and 2 soft tissue biopsies at our institution in 2015. The application was
used in a total of 7 procedures in 7 patients, all of which were technically successful.
The ages of patients ranged from 19 to 86 years. Particular use of the fusion application compared to sonography alone was noted in the biopsy of certain soft tissue
lesions and in perineurial therapeutic injections.
Key Words—fiducial markers; fusion; MRI; registration; ultrasound
C
Received February 9, 2016, from the New York
University Langone Medical Center, Hospital
for Joint Diseases, New York, New York USA.
Revised manuscript accepted for publication
April 3, 2016.
The authors have no disclosures. Institutional review board approval was obtained for
this study.
Address correspondence to Christopher J.
Burke, MBChB, New York University Langone
Medical Center, Hospital for Joint Diseases, 301
E 17th St, New York, NY 10003 USA.
E-mail: [email protected]
Abbreviations
MR, magnetic resonance; 3D, threedimensional
doi:10.7863/ultra.16.02024
onventional two-dimensional ultrasound systems
equipped with position sensors can acquire threedimensional (3D) spatial data allowing registration with
previously acquired computed tomography and magnetic resonance (MR) for fused real-time ultrasound imaging.1–3 The fusion
software requires identification of internal fiducial points that act
as localization markers, corresponding to targeted anatomic points
in the pre-acquired imaging, the accuracy of which is important to
the subsequent real-time registration. Image fusion alignment
involves sequential algorithmic transformations minimizing the
error between the output and target image. An electromagnetic
field generator is used to track transducer orientation to simultaneously map real-time ultrasound with corresponding anatomy on
pre-acquired MR. This technology allows improved navigation
with hand-swept imaging and has demonstrated use in a range of
specialties including liver,4,5 prostate,6,7 and breast8 procedures.
There has also been recent interest with regard to musculoskeletal
applications9–13 with real-time fusion being used to guide sacroiliac injections in vitro and vivo9,10 and guide biopsy of musculoskeletal tumors.13 Despite this, a clear role for the technology
remains yet to be established in musculoskeletal intervention. We
describe our experience and the potential use in musculoskeletal
procedures.
C 2016 by the American Institute of Ultrasound in Medicine | J Ultrasound Med 2017; 36:217–224 | 0278-4297 | www.aium.org
V
Burke et al—Ultrasound-MRI Fusion Applications in Musculoskeletal Intervention
Materials and Methods
Informed verbal consent was obtained from each patient
for the use of the application in addition to routine written consent, and the study was approved by our institutional review board. The eSieFusion imaging application
on an ACUSON S3000 ultrasound system (Siemens
AG, Munich, Germany) was used to automatically integrate real-time ultrasound images using a position sensor
with previously acquired MR images. The corresponding
appropriate MR sequence for fusion was imported to
the Ultrasound machine S3000 from the picture archiving and communication system prior to beginning the
case. All registrations were performed using either an L9
linear or C6 curved transducer with an attached probe
used in conjunction with an electromagnetic field generator for transducer localization in space (Figure 1). The
procedures were performed by 2 musculoskeletal radiologists (C.B., R.A.) with 8 and 25 years of experience,
respectively, each with 1 year of experience with the
fusion application. The application was used in a total of
7 procedures in 7 patients between March and July
2015. The procedures were identified as being appropriate in advance. A minimum of 3 fiducial markers were
used for registration to fuse the 2 data sets as close to
the intended needle tip target as possible, which varied
with respect to the anatomy involved (Figure 2).
One percent lidocaine local anesthesia was used in
all procedures. The fusion imaging was utilized in
pudendal (Figure 2) and ninth intercostal perineurial
therapeutic injections, a piriformis muscle injection targeting the sciatic nerve, a sacroiliac joint injection for
sacroiliitis, and barbotage/therapeutic injection for the
treatment of hamstring tendon-origin calcific tendopathy. The application was also used in 2 soft tissue
biopsies, the first in a case of a suspected metal-onmetal hip arthroplasty debris-related periprosthetic soft
tissue mass and the second in a patient following resection of a forearm sarcoma with positive margins and
new enhancing soft tissue on follow up MR imaging
surveillance (Figure 3).
T1-weighted MR sequences were selected for fusion
in all cases except the ninth intercostal nerve perineurial
injection where the ultrasound was fused to a 3D
Figure 1. A, Ultrasound probe with position sensor. A bracket secures a sensor to the transducer that affirms the appropriate “place in space”
within the electro-magnetic field. All registrations were performed using either an L9 linear or C6 curved transducer. B, Electromagnetic field generator (transmitter). C, The application of the fusion application (eSieFusion imaging) for an ultrasound-guided barbotage of the rotator cuff. Note,
the electromagnetic field generator (transmitter or cube) seen on the right of the image is positioned at least 60 cm away from the electronics unit
(tracker box) and within a 20 to 30 cm radius from the patient, as recommended by the manufacturer.
218
J Ultrasound Med 2017; 36:217–224
Burke et al—Ultrasound-MRI Fusion Applications in Musculoskeletal Intervention
Figure 2. A 26-year-old male with pelvic pain. MR ultrasound fusion images have been acquired with the patient positioned prone. Fiducial
markers have been placed at the (A) ischial tuberosity margin labelled PtC01, (B) the obturator internus margin labelled PtC04, and on the (C)
pudendal nerve itself labelled PtC03 (white arrow) allowing for localization of the previously acquired MR to the ultrasound examination.The green
bar is an indicator of proximity of the transducer within the electromagnetic field. The red “needle unplugged” represents the fact that the needle
tracking function is not in use. D, The pudendal nerve is mildly thickened within Alcock’s canal (white arrow in both MR and ultrasound images)
consistent with the patient’s symptoms. E, MR ultrasound registration image demonstrates the needle being positioned from the left side of the
screen (white arrow in ultrasound image) and perineurial injection solution (*) with the black arrow on the corresponding MR image demonstrating the position of the nerve. IT, ischial tuberosity; OI, obturator internus. See also the video available online at wileyonlinelibrary.com/journal/jum.
(continued)
J Ultrasound Med 2017; 36:217–224
219
Burke et al—Ultrasound-MRI Fusion Applications in Musculoskeletal Intervention
SPACE sequence and the right forearm biopsy where
fusion was performed with a contrast-enhanced fat-suppressed T1 sequence. All fused MR sequences used had
been acquired in the axial plane. In the current work,
only the split screen mode was used due to operator
preference, although the images can also be displayed in
an overlay mode with variable transparency of the ultrasound image. In all cases, axial images were displayed
within and on either side of the target region of interest.
The fused sequence was aligned in the same orientation as the real-time ultrasound. In all cases, this was in
the axial plane. For the pudendal nerve, hamstring-origin
Figure 2. (Continued)
220
J Ultrasound Med 2017; 36:217–224
Burke et al—Ultrasound-MRI Fusion Applications in Musculoskeletal Intervention
Figure 2. (Continued)
Figure 3. An 86-year-old male postforearm sarcoma resection. Ultrasound fusion to axial fat suppressed T1 contrast-enhanced MR image allows
co-registration to the new enhancing soft tissue region (*) at the resection margins adjacent to the proximal radius. R, radius; U, ulna.
J Ultrasound Med 2017; 36:217–224
221
Burke et al—Ultrasound-MRI Fusion Applications in Musculoskeletal Intervention
barbotage, and piriformis and sacroiliac joint injections,
the patient was positioned prone, therefore the imported
sequence was inverted in the vertical and horizontal
planes. All other cases were performed in the supine
position, therefore no change in orientation of the axial
image was required.
Results
The application was used in 7 different procedures (see
Table 1). The mean time for the procedures ranged from
6 minutes to 33 minutes with a mean of 19.7 minutes.
The application was able to register to different MR scanners with varying sequence parameters and interspaces.
The pudendal nerve, piriformis/sciatic nerve, and
intercostal nerve therapeutic injections were all technically
successful and produced relief of symptoms. The sacroiliac joint therapeutic injection and hamstring-origin calcific
tendinosis barbotage were both successful demonstrating
symptomatic improvement. Both of the biopsies yielded
diagnostic tissue. The periprosthetic lesion was confirmed
as a non-malignant, particle-related mass, and the biopsy
of the suspicious region of enhancement adjacent to the
forearm resection bed yielded inflammatory soft tissue
without evidence of local neoplastic recurrence.
Discussion
This application allows the operator to align 3D computed tomography or MR data with real-time ultrasound
imaging and has been successfully used in a range of
applications to improve navigation and ultrasound-based
guidance for interventional procedures. The software
requires identification of internal fiducial points that act
as localization markers, corresponding to targeted anatomic points in the pre-acquired imaging, the accuracy
of which is important to the subsequent real-time registration. Image fusion alignment involves multiple steps,
the first being to generate synthetic data, to which a
computed transformation is applied to establish fiduciary
and target registration error. An input image with several
fiduciary points identified undergoes a series of translations and rotations and is compared to a target image
from which an error can be calculated, followed by
sequential algorithmic transformations minimizing the
error between the output and target image. Using this
technique point-based registration of both fiducial and
222
target points has been reported to yield an average target
registration error as low as 5.4 mm.3
With respect to musculoskeletal interventions, realtime fusion software has been described in sacroiliac
injections9,10 and guiding biopsy of musculoskeletal
tumors.13 We similarly found the fusion application
effective for sacroiliac joint injection and in guiding soft
tissue biopsies yielding diagnostic tissue. We also used
the application for ninth intercostal and pudendal nerve
perineurial injections, a piriformis injection, and a hamstring tendon-origin calcific barbotage.
Sequence selection varied depending on the most
favorable anatomic depiction. T1-weighted images were
used in the majority of cases, however, in the case of
intercostal neuritis, fusion was performed to the 3D
SPACE sequence and in the postoperative sarcoma
resection patient with enhancement around the resection bed, the biopsy was fused to a fat-suppressed contrast-enhanced T1 sequence (Figure 3). In all cases,
registration was performed using axial acquisitions and
at least 3 fiduciary markers.
It is acknowledged that practitioners are generally
able to plan and directly target needle placement on the
ultrasound images without additional benefit of real-time
fusion. Furthermore, in cases targeting the gadoliniumenhancing viable portion of a mass, power Doppler
alone may be sufficient for the same purpose. The target
anatomy was generally well depicted with ultrasound
alone, however, there were certain scenarios where the
use of dynamic real-time MR ultrasound fusion allowed
for more confident needle placement. The position of
the pudendal nerve within Alcock’s canal may be difficult
to identify clearly on ultrasound, and the addition of MR
fusion improved accurate localization for needle placement. Similarly, in the case of the ninth intercostal nerve
injection, localization allowed for accurate targeting of
the affected nerve in a case made technically more
demanding by the patient’s body habitus. In this case,
symptomatic relief was crucial after many years of misdiagnosis and failed therapy. In the case of previous forearm sarcoma resection and radiation therapy, the use of
fusion allowed for accurate sampling of the new enhancing tissue identified on follow-up MR with a greater
degree of confidence than simply using ultrasound alone,
the histopathology of which yielded inflammatory tissue
with no evidence of local recurrence.
Piriformis injections are usually technically straightforward using ultrasound alone, although real-time
J Ultrasound Med 2017; 36:217–224
J Ultrasound Med 2017; 36:217–224
Left piriformis
syndrome
Right-sided
rib pain
Right-sided pelvic
pain
Postsarcoma
resection and
radiation therapy
right forearm
27-year-old female
19-year-old male
27-year-old female
86-year-old male
58-year-old female
Right hamstrings
origin calcific
tendinopathy
27-year-old female
Right-sided sacroiliitis
Right ninth intercostal nerve
thickening and increased
3D SPACE signal on MR
neurogram
No significant findings on
the prior MR; normal
course of the sciatic
nerve
Right hamstrings calcific
tendinopathy
Mild thickening of the
pudendal nerve
Imaging Findings
New enhancing region
around the proximal
aspect of resection bed
adjacent to the radius on
follow up MR (Figure 3)
Status post revision Large mixed cystic and
left total hip
solid periprosthetic
arthroplasty and
mass on MR
pain
Pelvic pain
Presentation
26-year-old male
Participant
Table 1. Applications of the Ultrasound MR Fusion Registration Software
Supine
Supine
Prone
Supine
Prone
Prone
Prone
Patient
Positioning
Targeted aspiration of the periprosthetic mass; 3 markers
placed on the prosthesis
components
Right ninth intercostal nerve
therapeutic injection; 1
marker was placed on the
nerve and 2 markers placed
upon the adjacent rib
margins
Right sacroiliac joint injection;
3 fiduciary markers were
placed around the sacroiliac
joint margins
Targeted biopsy of new
enhancing region; 3 markers
were placed on the radial
neck
Pudendal nerve therapeutic
injection; 3 markers in total
were placed on the ischial
tuberosity, obturator internus
and nerve itself (Figure 2)
Barbotage and therapeutic
injection; 3 markers in total
were placed on the hamstrings origin and the ischial
tuberosity
Piriformis/perisciatic nerve therapeutic injection; 2 markers
placed on the ischial tuberosity and 1 on the sciatic
nerve
Procedure and
Fiduciary Marker
Placement
19
14
6
33
20
27
19
Total Procedure
Time (mins)
Fusion allowed accurate
needle placement
Fusion allowed accurate
targeting of new enhancing region identified on
MR
Fusion allowed convenient
access point to the sacroiliac joint
Fusion allowed accurate
placement of the needle
within the perineural
space in a location where
the nerve may be poorly
visualized using ultrasound alone
Fusion provided greater
level of confidence for
needle placement based
on neurogram
Fusion helped localize the
calcific deposit
Fusion helped visualize the
nerve in Alcock’s canal
Authors Comments
Burke et al—Ultrasound-MRI Fusion Applications in Musculoskeletal Intervention
223
Burke et al—Ultrasound-MRI Fusion Applications in Musculoskeletal Intervention
correlation with fused MR imaging added confidence to
the procedure and may possibly identify a variable course
of the sciatic nerve, which in up to 15% of patients, all or
part of the sciatic nerve may pierce the piriformis muscle
or emerge superior to it.14–16 In the case of periprosthetic
soft tissue mass, targeted biopsy could have been performed easily with ultrasound alone, although real-time
fusion enabled affirmation that both ultrasound and MR
were, in fact, the same abnormal soft tissue. Furthermore,
the hamstring tendon-origin calcific deposit was well seen
on ultrasound, although the ability to perform fused imaging confirmed the abnormality seen on MR as being calcification before barbatoge.
There are several apparent limitations to this application. There is an additional delay in the process of importing the appropriate MR sequence, alignment, and fusion,
however, this will likely reduce with increasing experience.
Ultrasound interventions frequently require positioning
the target anatomy for optimal acoustic access and needle
placement, as a result probe alignment is not in the true
axial, sagittal, or coronal planes with possible resultant difficulty correlating with standard MR cross-sectional imaging. True fiduciary points may not be practical. The best
fiducial points should be closely placed to the relevant
anatomic target for optimal registration and ideally be
placed on fixed anatomic structures (ie, not across a joint
where motion may lead to malalignment). An additional
limitation is the requirement of using an electromagnetic
field generator near the area of interest and once the registration occurs, its position must remain fixed to maintain
the fusion. Image acquisition is, as with all sonographic
procedures, user-dependent and the quality of the imaging
is affected by the physical characteristics of the patient
and overlying structures.
Our cohort size is small and prospective work is
required to confirm increased accuracy and improved
operator performance compared to standard practice in
musculoskeletal interventions. It appears that the greatest
use of this application may be in the localization of small
or enhancing soft tissue structures, readily visible on MR
but difficult to appreciate on ultrasound such as deep
biopsies or perineurial injections, particularly where factors
such as body habitus may increase technical difficulty.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
References
1. Roche A, Pennec X, Malandain G, Ayache N. Rigid registration of
3D ultrasound with MR images: a new approach combining intensity
224
16.
and gradient information. IEEE Trans Med Imaging 2001; 20:1038–
1049.
Rohlfing T, Maurer JCR. Modeling liver motion and deformation
during the respiratory cycle using intensity-based nonrigid registration
of gated MR images. Med Phys 2004; 31:427–432.
Wein W, Brunke S, Khamene A, Callstrom MR, Navab N. Automatic
CT-ultrasound registration for diagnostic imaging and image-guided
intervention. Med Image Anal 2008; 12:577–585.
Crocetti L, Lencioni R. DeBeni S, See T, Pina, C, Bartolozzi, C. 2008.
Targeting liver lesions for radiofrequency ablation: an experimental
feasibility study using a CT-US fusion imaging system. Invest Radiol
2008; 43:33–39.
Lee MW. Fusion imaging of real-time ultrasonography with CT or
MRI for hepatic intervention. Ultrasonography 2014; 33:227–239.
Costa DN, Pedrosa I, Donato F Jr, Roehrborn CG, Rofsky NM. MR
imaging-transrectal US fusion for targeted prostate biopsies: implications for diagnosis and clinical management. Radiographics 2015; 35:
696–708.
Siddiqui MM, Rais-Bahrami S, Turkbey B, et al. Comparison of MR/
ultrasound fusion-guided biopsy with ultrasound-guided biopsy for
the diagnosis of prostate cancer. JAMA 2015; 313:390–397.
Pons EP, Azcon FM, Casas MC, Meca SM, Espona JL. Real-time
MRI navigated US: role in diagnosis and guided biopsy of incidental
breast lesions and axillary lymph nodes detected on breast MRI but
not on second look US. Eur J Radiol 2014; 83:942–950.
Klauser AS, De Zordo T, Feuchtner GM, et al. Fusion of real-time
US with CT images to guide sacroiliac joint injection in vitro and in
vivo. Radiology 2010; 256:547–553.
Zacchino M, Almolla J, Canepari E, Merico V, Calliada F. Use of
ultrasound-magnetic resonance image fusion to guide sacroiliac joint
injections: a preliminary assessment. J Ultrasound 2013; 16:111–118.
Hu Z, Zhu J, Liu F, Wang N, Xue Q. Feasibility of US-CT image fusion
to identify the sources of abnormal vascularization in posterior sacroiliac
joints of ankylosing spondylitis patients. Sci Rep 2015; 5:18356.
Wong-On M, Til-Perez L, Balius R. Evaluation of MRI-US fusion
technology in sports-related musculoskeletal injuries. Adv Ther 2015;
32:580–594.
Khalil JG, Mott MP, Parsons TW 3rd, Banka TR, van Holsbeeck M.
2011 Mid-America Orthopaedic Association Dallas B. Phemister
Physician in Training Award: Can musculoskeletal tumors be diagnosed with ultrasound fusion-guided biopsy? Clin Orthop Relat Res.
2012; 470:2280–2287.
Smith J, Hurdle MF, Locketz AJ, Wisniewski SJ. Ultrasound-guided
piriformis injection: technique description and verification. Arch Phys
Med Rehabil. 2006; 87:1664–1667.
Fishman L, Dombi G, Michaelsen C, et al. Piriformis syndrome: diagnosis, treatment, and outcome—a 10-year study. Arch Phys Med Rehabil 2002; 83:295–301.
Pecina M. Contribution to the etiological explanation of the piriformis
syndrome. Acta Anat 1979; 105:181–187.
J Ultrasound Med 2017; 36:217–224