A mechanism for simplified scanner control with application to MRI-guided interventions
Abstract
Magnetic Resonance Image (MRI)-guided interventions involving percutaneous biopsies of
lesions, or trajectory alignment with prospective stereotaxy are conducted in real time using rapid
image acquisition. A mechanism of passively localizing a device and calculating its orientation
is desired to improve interventional outcomes in these situations. In this work, we propose and
evaluate an image-based technique to determine the position and alignment of a linearly shaped
interventional device within an ex-vivo tissue specimen. Low resolution 3D orientation scan data
is processed to produce a virtual line tting using principal component analysis. The line tting
algorithm was incorporated into a biopsy needle tracking system implemented with an MRscanner
operated using a footswitch. A GUI application was written to collect foot pedal input
and display automated visualization of device placement inside the scanner room. Placement time
trials (N=3) conducted with this system using porcine muscle and phantom samples suspended
in rigid frames with inserted gadolinium-enhanced targets. The mean targeting error across
all directions was 3:6 mm and 5:1 mm for the phantom trials and ex-vivo trials respectively.
The average entry-to-target time was 247 sec. Device localization during trials was adequate to
contain a 11-gauge titanium biopsy needle within a visualization slice volume of 10 mm after
93:8% of alignments over insertion lengths between 30 mm to 110 mm at insertion angles between
1:4 to 20 from the static magnetic eld and frequency encoding axes. Practical considerations
were identi ed and occupational exposure measurements were collected as part of determining
the system's overall feasibility.