As a result of the increased number and complexity of interventional angiography procedures, x-ray fluoroscopy is a major source of diagnostic x-ray dose to patients and staff. Fluoroscopy images are quantum limited and simply reducing dose leads to unacceptably noisy images. Two techniques for reducing dose will be investigated: image processing to reduce noise without blurring and pulsed fluoroscopy. An interdisciplinary team consisting of image processing scientists, a perceptual psychologist, and angiographers will tackle the problem using tools of image processing and human perceptual testing. Preliminary perception studies, conducted on a unique device called the video tachistoscope, indicate that the dose of 15 frames/sec pulsed can be reduced by approximately 25% of that in conventional continuous fluoroscopy while maintaining the same ability to detect stationary, low- contrast objects. this result is well-described by an extension to the ideal observer model. Several refinements are proposed that include introducing a forced-choice experimental paradigm, using more realistic, moving objects, and measuring reaction times. A down-side to pulsed is the choppy display. To make a smoother image, we intend to investigate image interpolation based on motion-estimation (see below) and a novel method for adjusting the gray-scale of objects in the gap-filling frame. Regarding noise reduction, good preliminary results have been obtained with motion-compensated, temporal filtering using a motion estimation scheme that well preserves motion boundaries in the optic-flow field. Several extensions to this basic method will be investigated. An alternative, heuristic approach to image sequence smoothing uses object detection rather than conventional motion-detection filtering. Briefly, we will roughly segment the long, thin objects of interest using morphological image processing techniques and perform less temporal filtering in these areas. In this way, we can smooth the image while not blurring the objects of interest. Filtering schemes will be evaluated both with objective numerical techniques and with perception studies.

Agency
National Institute of Health (NIH)
Institute
National Heart, Lung, and Blood Institute (NHLBI)
Type
First Independent Research Support & Transition (FIRST) Awards (R29)
Project #
5R29HL048918-02
Application #
2225023
Study Section
Diagnostic Radiology Study Section (RNM)
Project Start
1993-05-17
Project End
1997-04-30
Budget Start
1994-05-01
Budget End
1995-04-30
Support Year
2
Fiscal Year
1994
Total Cost
Indirect Cost
Name
Case Western Reserve University
Department
Biomedical Engineering
Type
Schools of Engineering
DUNS #
077758407
City
Cleveland
State
OH
Country
United States
Zip Code
44106
Xue, P; Wilson, D L (1996) Pulsed fluoroscopy detectability from interspersed adaptive forced-choice measurements. Med Phys 23:1833-43
Wilson, D L; Xue, P; Aufrichtig, R (1994) Perception of fluoroscopy last-image hold. Med Phys 21:1875-83
Aufrichtig, R; Thomas, C W; Xue, P et al. (1994) Model for perception of pulsed fluoroscopy image sequences. J Opt Soc Am A Opt Image Sci Vis 11:3167-76
Aufrichtig, R; Xue, P; Thomas, C W et al. (1994) Perceptual comparison of pulsed and continuous fluoroscopy. Med Phys 21:245-56
Labbe, M S; Chiu, M Y; Rzeszotarski, M S et al. (1994) The x-ray fovea, a device for reducing x-ray dose in fluoroscopy. Med Phys 21:471-81