IMAGE REGISTRATION ACCURACY WITH LOW-DOSE CT: HOW LOW CAN WE GO?
Omkar Dandekar
1,2
, Khan Siddiqui
3
, Vivek Walimbe
4,5
, Raj Shekhar
2
1
Department of Electrical and Computer Engineering, University of Maryland, College Park, MD 20740
2
Department of Diagnostic Radiology, University of Maryland, Baltimore, MD 21201
3
VA Maryland Health Care System, Baltimore, MD 21201
4
Biomedical Engineering Center, The Ohio State University, Columbus, OH 43210
5
Dept of Biomedical Engineering, The Cleveland Clinic Foundation, Cleveland, OH, 44195
ABSTRACT
Image-guided interventions are known to lead to
improved outcomes and significantly faster patient
recovery as compared with conventional open, invasive
procedures. Common intraoperative imaging techniques
such as endoscopy and fluoroscopy are two-dimensional
(2D), and provide a 2D representation of the 3D anatomy.
Use of recently emerged multislice computed tomography
(CT) can facilitate 3D visualization of anatomy during an
intervention. The speed and dimensionality of these CT
scanners make their use in image-guided interventions
technically feasible. For clinical acceptance, however, the
net radiation dose to the patient and the interventionist
must be minimized. This article suggests a strategy to
reduce radiation dose and describes an evaluation scheme
to identify the optimal dose which does not sacrifice the
specificity of the image-guided procedure. Our work
indicates at least a tenfold reduction in radiation dose.
1. INTRODUCTION
Minimal invasiveness is the defining feature and the
primary motivator for the adoption of image-guided
interventions. Other equally important advantages of
image-guided interventions are improved outcomes,
shorter procedures and a quick recovery [1]. The current
use of image guidance can be found in neurosurgery,
focal cancer ablative therapies, and radiosurgeries and
radiotherapies. In many clinical applications, image-
guided intervention has become the standard of care (e.g.,
laparoscopic cholecystectomy).
Image-guided procedures are usually planned on
preoperative CT/magnetic resonance (MR) images.
However, due to normal tissue motion, the anatomy at the
time of the surgical procedure generally differs from the
anatomy at the time of planning. Such misregistration
between pre- and intraoperative anatomy renders the
planned treatment and navigation, based solely on
preoperative data, inaccurate and unsafe.
To compensate for this misregistration, the anatomy
needs re-imaging immediately prior to and during the
intervention. Registering preoperative images then with
the intraoperatively collected images provides valuable
information for updating operative/surgical navigation
plan. An interesting aspect of this approach is a
possibility to augment the current anatomy, with
supplementary information (such as contrast) from the
preoperative scan, which will tremendously aid
navigation and localization.
With the advent of multislice CT (up to 64), intra-
operative imaging is now possible with computed
tomography. Comparatively, magnetic resonance imaging
(MRI) continues to lag in speed, while use of ultrasound
is limited due to poor image quality and, in image-guided
surgeries specifically, the difficulty of scanning across the
pneumoperitoneum with CO
2
insufflation. Radiation
exposure to the patient and the interventionist, however,
continues to be a concern with using CT, and must be
minimized.
Our primary radiation dose reduction strategy is to
acquire a standard-dose CT image preoperatively and
scan the dynamic operative field subsequently using low-
dose CT. Using high-speed nonrigid 3D image
registration techniques [2] our group has developed, the
preoperative CT image can be registered to low-dose
intraoperative CT images. Registered preoperative CT
will show the dynamic intraoperative anatomy and will
substitute the low-dose CT images. These diagnostic
quality images can be 3D rendered and used for
intraoperative guidance and navigation. Capability of
viewing hidden structures using CT together with the
additional capability of virtually inserting tracked tools in
the 3D renderings will add a new dimension to image-
guided interventions.
Our proposed dose reduction strategy necessitates the
determination of the lowest acceptable dose for
intraoperative CT, which permits accurate image
registration. This article describes an evaluation scheme
to judge the registration accuracy with low-dose CT. The
following sections describe this method and the results in
detail.
502 0-7803-9577-8/06/$20.00 ©2006 IEEE ISBI 2006