Development of a Hybrid Ultrasound Biopsy Phantom

1
Development of a Hybrid Ultrasound Biopsy Phantom
to Study The Effect of Tissue Compressibility on
Target Identification and Localization Gaurav
Shukla, Bing Wu PhD, Roberty Klatzky PhD, Jules
Sumkin DO, George Stetten MD,PhD Department of
Bioengineering, University of Pittsburgh/Carnegie
Mellon University Medical Scientist Training
Program
Motivation
Preliminary Results and Prior Work
Research Design and Methods

• Breast cancer is the most common cancer in women.
• gt200,000 new diagnoses in 2006
• gt40,000 deaths in 2006
• Diagnosis typically involves screening
  (mammography), ultrasound of suspicious lesions
  (Figure 1a), and subsequent biopsy for
  pathological evaluation (Figure 1b).
• A number of issues arise in breast biopsy that
  require special attention, including
• tissue compressibility
• increased difficulty in identifying and keeping a
  target in the ultrasound image
• greater reliance on the fine detail in the image
• the extra requirement of keeping the needle in
  the plane of the ultrasound scan

• We will develop the Hybrid Biopsy Phantom (HBP),
  in which a blank gel ultrasound phantom is
  augmented with virtual targets (Figure 6)
• The physical gel will provide realistic
  interaction including physical compressibility
  and actual image speckle
• An optically tracked SF and biopsy needle will
  permit graphically simulated cysts and tumors to
  be overlaid on the ultrasound image and targeted
  at predetermined 3D locations
• These virtual targets will be deformable by
  compression based on information from the tracked
  implements and tracking of the distortion of the
  image speckle
• Modeling of the interaction between real
  ultrasound images and virtual targets
• It will be necessary to incorporate differences
  in the compressibility of the simulated lesion
  vs. that of the surrounding tissue. Such
  differences are important in diagnosing benign
  vs. malignant lesions, since malignant lesions
  tend to be less deformable.,
• When the simulated lesion is modeled to have a
  deformability different from the surrounding gel,
  we will explore methods of deforming the
  displayed speckle around the lesion to
  accommodate its motion relative to the gel.

Figure 1a ultrasound image of a fluid-filled
cyst in breast tissue. Figure 1b ultrasound
image of cyst aspiration (needle visible)
Hypothesis (Aim)
We will construct a hybrid biopsy phantom,
comprised of a realistic breast phantom and a set
of simulated breast lesions which will be
integrated with one another, creating a tool
with which we can study the psychophysical issues
above.
Figure 3a (above) Sonic flashlight used in
clinical trials placing catheters in deep veins
of the arm. Custom plastic housing attaches
half-silvered mirror and flat-panel monitor to
ultrasound transducer.
Figure 3b (above) Sonic flashlight used to guide
a needle into the jugular vein of a cadaver. The
needle tip is visible in the virtual image of the
jugular vein, which floats at its correct
location.
Background
Our laboratory has developed an innovative
technology to guide interventional procedures
which merges ultrasound (US) images with a direct
view of the patient. The device, called the
Sonic Flashlight (SF), consists of a US
transducer, on which is mounted a small
flat-panel display and a semitransparent mirror
(Figure 2). Looking through the mirror, the
operator sees a reflection of the US image
floating at the actual location within the
patient. By aiming for targets in this virtual
image, the operator takes advantage of natural
hand-eye coordination to guide percutaneous
procedures such as the placement of a venous
catheter, or potentially, the biopsy a suspected
tumor. We have recently conducted NIH-funded
clinical trials using the SF to place catheters
in the deep veins of the arm, with considerable
success. We propose to extend the SF to another
important application the guidance of biopsy in
the diagnosis of breast cancer.
Figure 6 Example images which demonstrate the
concept of the hybrid biopsy phantom. A real
ultrasound image of a blank phantom (left) is
augmented with a computer-simulated cyst
(center). When the phantom is compressed by the
ultrasound probe, the simulation determines the
amount of compression and alters the morphology
of the virtual lesion appropriately.
Future Directions
Through-plane (Figure 4a, above, left) and
in-plane (Figure 4b, above, right) insertion of
needle into a cyst in a breast phantom
(BluePhantomTM). In-plane insertion (right) shows
the shaft of the needle in the scan. For our
purposes, however, these physical phantoms have
limitations, including an inability to easily
redefine target shape and location.
Figure 2 (below) schematic of Sonic Flashlight.

• When completed, the HBP will demonstrate a number
  of features that, to our knowledge, do not yet
  exist in an ultrasound phantom
• We will investigate the psychophysics of
  ultrasound guided breast biopsy
• e.g., Comparison studies between performance in
  guidance tasks between conventional ultrasound
  and the sonic flashlight (Figure 7)
• We will compare learning with the SF vs. CUS and
  develop training procedures based on the HBP.
• HBP as clinical training tool for novices
  learning to perform breast biopsies
• Develop and test a sonic flashlight for breast
  biopsy in Phase I clinical trials

The conception (Figure 5a, near right) and a
prior implementation (Figure 5b, far right) of a
virtual tomographic reflection (VTR) system, in
which a mock sonic flashlight is optically
tracked to provide continuous location and
orientation information. Using the tracker, it
is possible to simulate virtual targets in a
phantom and visualize them on the SF display.
Prior work in our lab utilized this VTR system to
study the psychophysics of the SF as compared
with conventional ultrasound (CUS), and we will
adapt this system in the development of the
hybrid biopsy phantom.
Figure 7 Performance of guidance tasks using
conventional US vs. SF in a fully-real system
using a water tank and real targets
Figures 5a (left) and 5b (right) Virtual
Tomographic Reflection