Automatic Determination of the Needle Tip for
Stereo-Tactic Biopsy
Hong Liu, PhD
Smith Chair Professor in Bioengineering
University of Oklahoma
Liu@ou.edu
Ge Wang
Departments of Biomedical Engineering and Radiology
University of Iowa
ge-wang@uiowa.edu
Introduction
Stereo image guidance is an active area of investigation. For instance,
digital stereotactic breast biopsy has been developed and successfully
used in clinical practice. Similar techniques and applications include
stereo angiography, and stereo image guided neurosurgery, etc.
Different from the conventional planar radiography which projects a
three-dimensional object onto a two-dimension image plane, stereology
is the science of determining depth information of an object based on
multiple planar (two-dimensional) views. And also different from
tomographic technologies, such as CT or MRI, stereo technique provides
us with a low dose and high speed method for recovering limited, but
clinically necessary, 3D information during image guidance. This
feature leads to the development of a real time stereo fluoroscopic
image guidance system.
Problem Description
Stereo-tactic breast biopsy requires specially designed equipment to
calculate the locations of an abnormality and the needle tip. This is
accomplished by using two angled radiographs. Currently, an X-ray
stereo-tactic system incorporates CCD-based detectors for gray-scale
images of 1024 by 1024 pixels. Although it is adequate for
stereo-tactic biopsy of breast masses, it often suffers from the
imaging accuracy limitation in the case of micro-calcification biopsy.
To improve the imaging accuracy, it is desirable to automatically
localize the needle tip with a sub-pixel resolution.
Suggested Methods
In this project, an image processing algorithm will be developed to
separate the needle from the background, determine the tip of the
needle, and analyze the localization errors. First, the background may
be roughly estimated via some low-pass filtering or polynomial fitting.
Then, the needle may be detected using the Hough Transform. Finally,
the needle tip can be defined as the half maximum point along the
midline of the needle. The localization errors can be studied in both
numerical simulation and phantom experiments.
Expected Results
An image processing algorithm will be developed for accurate and
reliable localization of the biopsy needle tip. The relationships
between the localization error and various factors will be established.
A user-friendly interface will be developed for use in a research
prototype system.
References
Radiographic Image of Needles