novel computational algorithm to quantify blood flow and ... · novel computational algorithm to...
TRANSCRIPT
Methods: Experimental Set-up :
Novel Computational Algorithm to Quantify Blood Flow and Vascular Resistance from Contrast Angiography with High Accuracy
Introduction: Contrast angiography is routinely used to diagnose and treat arterial occlusive disease, but is limited in its ability to provide hemodynamic data.
• Measurement Error >10% (All Phase of Cardiac Cycle1) •Theoretical Flow Maximum based on technique1,2 • Algorithms difficult to use in the clinical setting
Hypothesis: Blood flow can be calculated from a contrast angiogram with <10% error through the application of a novel computational algorithm.
Doran Mix, B.S.1, Daniel Phillips, Ph.D.2, Steven Day, Ph.D.2, Nicole Varble, M.S.1, Karl Schwartz, M.D.1, David Gillespie M.D.1, Ankur Chandra, M.D.1 1University of Rochester, Rochester, NY, USA, 2Rochester Institute of Technology, Rochester, NY, USA
Conclusion: We conclude that using this approach, blood flow can be angiographically measured with increased accuracy.
Limitations: • Pixel Resolution of ROI •Frame Rate of Angiographic Sequence
Future Applications: •Diagnostics •Therapeutic Interventions
Experimental Design: Utilizing the angiographic simulator (Figure #1) the resistance of a in-flow stenosis was systematically varied at the AVF (Figure#2A). The custom computational algorithm was applied under two experimental conditions:
• Experiment #1: For a fixed ~9000 pixel ROI with the proximal edge of the ROI 11.5cm from pigtail catheter (Figure #3)
• Experiment #2: For a progressively increasing user selected ROI, with the proximal edge of the ROI 11.5cm from the pigtail catheter (Table #1 and Figure #4)
Figure#2: Fistula Model and Matlab Algorithm Showing Vessel Segmentation Algorithm
A. Upper Arm AVF model showing location of stenosis and contrast injection using 6F pigtail catheter. B. Light based simulated angiogram showing DICOM image of contrast injection at injection site (Figure A) C. Matlab algorithm showing user selected region of interest from DICOM image (Red Circle Figure B) D. Automatic vessel detection in ROI (Red Box) using Otsu method and 4x4 search algorithm (Green Box) E. Automatic vessel segmentation, perpendicular to flow using results of vessel detection (Figure D)
A.
Figure #1.A:Radiation-free Angiographic Simulation System Conceptualization: Computer Controlled Pneumatic Compression Chamber Blood Mimicking Fluid Transonic Transit Time Flow Meter Hewlett Packard Multi-Parameter Pressure Monitor 5cc Optically Opaque Dye dissolved in 250cc Optiray 350 1024x1024 Digital CCD X-ray Camera X-ray Generator Frame Timing Simulator DICOM Network USB Network Figure #1.B&C: Resulting Simulator Pressure and Flow Waveforms
PlatinumOne™ Digital Angiography System
In-Vitro Hemodyanmic Flow Circuit Simulator
©
MEDRAD Mark V Power Injector
Pressure/Flow Meters
Data Acquisition System PACS Sever
Adjustable Green Light
Table
X-ray Generator Frame Timing
Simulator
CCD Digital X-ray Camera
Dye Diversion Chamber
Hydraulic Pump
Results:
Angiographic Frame Number at 15fps
Segment Location (0.3mm/Segment)
Degree of Fistula Stenosis Flow Restriction (%) Resistance (mmHg*min/L)
100% Flow Restriction
Resistance: ∞ mmHg*min/L
88% Flow Restriction
Resistance: 485 mmHg*min/L
78% Flow Restriction
Resistance: 245 mmHg*min/L
65% Flow Restriction
Resistance: 138 mmHg*min/L
54% Flow Restriction
Resistance: 94 mmHg*min/L
40% Flow Restriction
Resistance: 57 mmHg*min/L
30% Flow Restriction
Resistance: 39 mmHg*min/L
13% Flow Restriction
Resistance: 19 mmHg*min/L
0% Flow Restriction
Resistance: 8 mmHg*min/L
Axillary Flow 180 cc/min
Distal Flow 180 cc/min
Post- Stenosis MAP 29 mmHg
Pre- Stenosis MAP 87 mmHg
Axillary Flow 280 cc/min
Distal Flow 173 cc/min
Post- Stenosis MAP 31 mmHg
Pre- Stenosis MAP 83 mmHg
Axillary Flow 380 cc/min
Distal Flow 180 cc/min
Post- Stenosis MAP 32 mmHg
Pre- Stenosis MAP 81 mmHg
Axillary Flow 470 cc/min
Distal Flow 152 cc/min
Post- Stenosis MAP 35 mmHg
Pre- Stenosis MAP 79 mmHg
Axillary Flow 570 cc/min
Distal Flow 155 cc/min
Post- Stenosis MAP 37 mmHg
Pre- Stenosis MAP 76 mmHg
Axillary Flow 700 cc/min
Distal Flow 153 cc/min
Post- Stenosis MAP 40 mmHg
Pre- Stenosis MAP 71mmHg
Axillary Flow 780 cc/min
Distal Flow 138 cc/min
Post- Stenosis MAP 43 mmHg
Pre- Stenosis MAP 68 mmHg
Axillary Flow 920 cc/min
Distal Flow 125 cc/min
Post- Stenosis MAP 47 mmHg
Pre- Stenosis MAP 62 mmHg
Axillary Flow 1030 cc/min
Distal Flow 118 cc/min
Post- Stenosis MAP 50 mmHg
Pre- Stenosis MAP 57 mmHg
192 cc/min
294 cc/min
387 cc/min
504 cc/min
525 cc/min
658 cc/min
785cc/min
923 cc/min
1039 cc/min
Histogram for User Specified Axillary Artery Region of Interest
Segmental Tracking of Maximum Gradient Frame Number
Contrast Derived Flow (cc/min)
Schematic of Hemodynamic Circuit Under Test
Frame #34
Abso
lute
Pix
el
Dens
ity in
RO
I
Fram
e w
ith M
axim
um
Dens
ity G
radi
ent
Fram
e w
ith M
axim
um
Dens
ity G
radi
ent
Fram
e w
ith M
axim
um
Dens
ity G
radi
ent
Fram
e w
ith M
axim
um
Dens
ity G
radi
ent
Fram
e w
ith M
axim
um
Dens
ity G
radi
ent
Fram
e w
ith M
axim
um
Dens
ity G
radi
ent
Fram
e w
ith M
axim
um
Dens
ity G
radi
ent
Fram
e w
ith M
axim
um
Dens
ity G
radi
ent
Fram
e w
ith M
axim
um
Dens
ity G
radi
ent
Number of segments automatically determined based on user ROI size
Frame# Maximum Gradient for ROI
Frame #38 Ab
solu
te P
ixel
De
nsity
in R
OI
Frame #39
Abso
lute
Pix
el
Dens
ity in
RO
I
Frame #44
Abso
lute
Pix
el
Dens
ity in
RO
I
Frame #53
Abso
lute
Pix
el
Dens
ity in
RO
I
Frame #71
Abso
lute
Pix
el
Dens
ity in
RO
I Ab
solu
te P
ixel
De
nsity
in R
OI
Frame #59
Frame #37
Abso
lute
Pix
el
Dens
ity in
RO
I Ab
solu
te P
ixel
De
nsity
in R
OI
Frame #35
Angiographic contrast 5cc light opaque dye mixed in 200cc Optiray 350 injected at 6cc/sec for a total of 10cc of contrast
Table#1: Experiment #2 Results of Angiographic Derived Flow Measurements
AVF Model and Angiographic Algorithm: B.
Stenosis
A.
C.
D.
E.
Figure #4: Linear regression of r2=.9921 between Transonic flow meter with 4% error (y-axis) and computationally derivate flow from angiographic images (x-axis) from experimental derived flow with increasing ROI Table#1.
References: 1. Shpilfoygel SD, Jahan R, Close RA, Duckwiler GR, Valentino DJ. Comparison of methods for
instantaneous angiographic blood flow measurement. Med Phys. 1999;26(6):862-871. 2. Shpilfoygel SD, Close RA, Valentino DJ, Duckwiler GR. X-ray videodensitometric methods for
blood flow and velocity measurement: A critical review of literature. Med Phys. 2000;27(9):2008-2023.
Figure #3: Absolute Pixel Density in fixed ~9000 pixel ROI, Experiment #1, with contrast injection controlled by simulated x-ray generator at 15th frame. Point of maximum gradient reflects change in distal fistula resistance.
Time (Frame Number at 15 frame/sec)
Abs
olut
e Pi
xel D
ensit
y in
RO
I
Flow
(LPM
) Pr
essu
re (m
mHg
)
B.
C.