Radionuclides in Nephrourology, Mikulov 2010

Quantitative Methods in
Renography

Richard Lawson
Central Manchester Nuclear Medicine Centre

[email protected]
 Overview
Problems of quantifying the renogram
Complex shape of the curve
Unwanted background
Background subtraction
Recognising correct subtraction
Tissue and vascular background components
Compare two solutions
Rutland plot
Deconvolution
Quantitative parameters
Relative function
Absolute function
Elimination
ISCORN Consensus Report: Sem.Nucl.Med. 1999, 29:146-159
Richard Lawson Central Manchester Nuclear Medicine Centre
 Renography
Renography is a dynamic study of kidney function
Time is the important dimension
It is the renogram curves that show transit of tracer
through the kidneys
So curves are more important than the images
Upslope of curves demonstrate kidney uptake
Relative function
one kidney compared with the other
Absolute function
each kidney compared with normal
Downslope of curves demonstrate elimination

It is important to produce the correct curves

Richard Lawson Central Manchester Nuclear Medicine Centre
 The Problems
The kidney activity-time curve is a combination of
three factors:
Uptake into the kidney
Transit through the kidney
Elimination from the kidney
Uptake depends on blood activity, which varies with
Speed of injection
Kidney function
Function of the other kidney
Recirculation of tracer
Renogram curve is a superimposition of the desired
kidney activity and unwanted background activity

Renogram quantification must overcome these problems

Richard Lawson Central Manchester Nuclear Medicine Centre
Background Subtraction
 Renogram Model

Concentration
Extravascular
Tissue
Time

Concentration
Blood

Time

Concentration Concentration
Kidney
Tubules + pelvis

Time

Bladder
Richard Lawson Central Manchester Nuclear Medicine Centre
 Renogram Model

Concentration
Extravascular
Tissue
Time

Concentration
Blood

Time
This is the

Concentration Concentration
curve that Kidney
we want Tubules + pelvis

Time

Bladder
Richard Lawson Central Manchester Nuclear Medicine Centre
 Regions of Interest
Each kidney ROI includes
Blood background
}
Renal blood vessels
Renal tubules Renogram
Renal pelvis
Overlying tissues Tissue background

Background ROI includes

Some blood vessels Blood background
Some tissues Tissue background
Kidney minus background
Should leave desired renogram

The optimum background ROI must include the

correct mixture of both blood and tissue background
Richard Lawson Central Manchester Nuclear Medicine Centre
 Background Subtraction
Kidney ROI curve Background subtracted
cps + Renal tubules cps
& Renal pelvis
Renal tubules
& Renal pelvis
+ Blood

Tissue
Time Time
Background ROI curve
cps If background
subtraction is correct,
+ Blood renogram curve should
rise smoothly from zero
Tissue
Time
Richard Lawson Central Manchester Nuclear Medicine Centre
 Mixing
During the first few seconds after injection
Tracer has not had enough time to mix uniformly
throughout all of the blood
Therefore blood activity in the background region may
differ from blood activity in the kidney region
So background subtraction may be wrong
Therefore ignore first few seconds of the curve
Until mixing is complete
Varies between patients
Often about 30 seconds
Possibly up to 1 minute

Richard Lawson Central Manchester Nuclear Medicine Centre

 Recognising Correct Subtraction

Curve is not
smooth during
first minute

Extrapolate to
overcome
mixing phase

0 1 2 3 4 5 min
Question 1: Is this renogram curve correctly subtracted ?
1 = Correct subtraction, 2 = Under-subtracted, 3 = Over-subtracted
Answer: Under-subtracted
Richard Lawson Central Manchester Nuclear Medicine Centre
 Recognising Correct Subtraction

0 1 2 3 4 5 min
Question 2: Is this renogram curve correctly subtracted ?
1 = Correct subtraction, 2 = Under-subtracted, 3 = Over-subtracted
Answer: Over-subtracted
Richard Lawson Central Manchester Nuclear Medicine Centre
 Recognising Correct Subtraction

0 1 2 3 4 5 min
Question 3: Is this renogram curve correctly subtracted ?
1 = Correct subtraction, 2 = Under-subtracted, 3 = Over-subtracted
Answer: Correct subtraction after extrapolation
Richard Lawson Central Manchester Nuclear Medicine Centre
 Recognising Correct Subtraction

Under-subtracted
If curve has a
kink during first
Correctly subtracted
minute then
extrapolate to
overcome Over-subtracted
mixing phase

0 1 2 3 4 5 min

Richard Lawson Central Manchester Nuclear Medicine Centre

 Three Phases of the Renogram
Phase 2
(Uptake) Textbooks
Phase 3 3 phases
(Elimination)
Renogram model
Phase 1 No vascular phase
(Vascular)

Originally renograms were acquired using probes

Without any means of background subtraction
The vascular phase was always present
The vascular phase is not part of the true renogram
With modern gamma camera techniques it should be
removed by proper computer processing
Richard Lawson Central Manchester Nuclear Medicine Centre
 Teaching Point
Background subtraction is correct when the
renogram curve rises smoothly from zero
After extrapolating to overcome incomplete mixing
during the first minute
Assuming that the bolus appears in kidneys during
the first frame of the study time zero

0 1 min
Richard Lawson Central Manchester Nuclear Medicine Centre
Background Subtraction
Examples
 Example 1

50 MBq 99mTc MAG3

Good uptake in both kidneys
Both kidneys equal function

Richard Lawson Central Manchester Nuclear Medicine Centre

 Example 1 Background below kidney

Relative function Right Kidney

Left kidney: 50%

Left Kidney
Right kidney: 50%

Question 4: Are these curves correctly subtracted ?

1 = Correct subtraction, 2 = Under-subtracted, 3 = Over-subtracted
Answer: Slight under-subtraction
Richard Lawson Central Manchester Nuclear Medicine Centre
 Example 1 Peri-renal background

Left Kidney
Relative function
Left kidney: 52% Right Kidney
Right kidney: 48%

Question 5: Are these curves correctly subtracted ?

1 = Correct subtraction, 2 = Under-subtracted, 3 = Over-subtracted
Answer: Very slight under-subtraction
Richard Lawson Central Manchester Nuclear Medicine Centre
 Example 1 Rutland method

L R

Right Kidney
Relative function
Left kidney: 52% Left Kidney
Right kidney: 48%
Question 6: Are these curves correctly subtracted ?
1 = Correct subtraction, 2 = Under-subtracted, 3 = Over-subtracted
Answer: Correct subtraction
Richard Lawson Central Manchester Nuclear Medicine Centre
 Teaching Point

If using MAG3 and kidney function is good

Using different background regions only makes a
small difference to background subtraction

If both kidneys have equal function

Then under-subtracting doesnt alter relative
function significantly

Richard Lawson Central Manchester Nuclear Medicine Centre

 Example 2

50MBq 99mTc MAG3

Both kidneys poor function
Right worse than left

Richard Lawson Central Manchester Nuclear Medicine Centre

 Example 2 Background below kidney

L R

Right Kidney
Relative function Left Kidney
Left kidney: 56%
Right kidney: 44%
Question 7: Are these curves correctly subtracted ?
1 = Correct subtraction, 2 = Under-subtracted, 3 = Over-subtracted
Answer: Significantly under-subtracted after extrapolation
Richard Lawson Central Manchester Nuclear Medicine Centre
 Example 2 Peri-renal background

L R

Right Kidney
Relative function
Left Kidney
Left kidney: 90%
Right kidney: 10%
Question 8: Are these curves correctly subtracted ?
1 = Correct subtraction, 2 = Under-subtracted, 3 = Over-subtracted
Answer: Over-subtracted
Richard Lawson Central Manchester Nuclear Medicine Centre
 Example 2 Rutland method

L R

Right Kidney
Relative function
Left kidney: 66% Left Kidney
Right kidney: 34%
Question 9: Are these curves correctly subtracted ?
1 = Correct subtraction, 2 = Under-subtracted, 3 = Over-subtracted
Answer: Correct subtraction after extrapolation
Richard Lawson Central Manchester Nuclear Medicine Centre
 Teaching Point
If using MAG3 and function is poor
Then different background regions can have a
big effect
If using DTPA this happens even with good
function
Because of lower extraction efficiency
If function is asymmetric
Then incorrect background subtraction can
affect relative function measurement

Richard Lawson Central Manchester Nuclear Medicine Centre

 Example 3

50 MBq 99mTc MAG3

Left kidney good function
Right kidney hydronephrotic

Richard Lawson Central Manchester Nuclear Medicine Centre

 Example 3 - Background below kidney

L R

Right Kidney

Relative function Left Kidney

Left kidney: 76%
Right kidney: 24%
Question 10: Is the right kidney curve correctly subtracted ?
1 = Correct subtraction, 2 = Under-subtracted, 3 = Over-subtracted
Answer: Under-subtracted Left kidney little under-subtracted
Richard Lawson Central Manchester Nuclear Medicine Centre
 Example 3 Peri-renal background

L R

Right Kidney

Relative function
Left Kidney
Left kidney: 85%
Right kidney: 15%

Question 11: Is the right kidney curve correctly subtracted ?

1 = Correct subtraction, 2 = Under-subtracted, 3 = Over-subtracted
Answer: Over-subtracted But left kidney is OK
Richard Lawson Central Manchester Nuclear Medicine Centre
 Example 3 Rutland method

L R

Right Kidney

Relative function
Left Kidney
Left kidney: 81%
Right kidney: 19%

Question 12: Is the right kidney curve correctly subtracted ?

1 = Correct subtraction, 2 = Under-subtracted, 3 = Over-subtracted
Answer: Correct subtraction For both kidneys
Richard Lawson Central Manchester Nuclear Medicine Centre
 Teaching Point

If the vascularity of each kidney is different

Then it is difficult to find a single background
region that works for both kidneys

The Rutland method overcomes this by

automatically adjusting the amount of vascular
background to suit each kidney

Richard Lawson Central Manchester Nuclear Medicine Centre

The Rutland Method
 The Rutland Method *
The real renogram is the response of the kidney
to a single injection
Resulting in a blood activity that is continually
falling
Imagine what the renogram would look like if we
gave a constant infusion
The blood activity could be kept constant
The Rutland method predicts what the constant
infusion renogram would look like
Based on the real renogram
And the real blood curve
* Rutland MD Nuc.Med.Comm 6: p11-20 (1985)

Richard Lawson Central Manchester Nuclear Medicine Centre

 Constant Infusion Model

Concentration
Extravascular
Tissue
Time

Concentration
Blood
Time

Concentration Concentration
Kidney
Time

Bladder
Richard Lawson Central Manchester Nuclear Medicine Centre
 Rutland Theory
Define:
-
Tissue ROI counts
Blood activity B = Vascular ROI counts (scaled for ROI size)

Kidney activity K = Kidney ROI counts - (scaled

Tissue ROI counts
for ROI size)
Assumptions:
1) Input rate to kidney is proportional to B True kidney counts = UC x Bdt
2) Nothing leaves during first few minutes (where UC is an uptake constant)
3) Vascular background = a x B (where a is a background subtraction factor)

Now: K = True kidney counts + Vascular background

K = UC Bdt + a B Rutland time

K
= UC
+ a
Bdt
Rutland Plot

B B
(Patlak plot)
This is the equation of a straight line with slope UC and intercept a
Richard Lawson Central Manchester Nuclear Medicine Centre
 Rutland Practice
Draw regions of interest L R
Kidneys
Vascular background
heart or spleen
Tissue background
below kidney
Bladder
Generate activity-time curves Post

Subtract tissue background

from kidneys, bladder and vascular
(scaling only for region size)
Construct Rutland plot for each kidney
Select range of points to fit straight line
Richard Lawson Central Manchester Nuclear Medicine Centre
 Choosing the fit range

Typical Rutland Plot

Fit should include

just points in
straight section

*
*
Late points will be below the line
*
**
(kidney is emptying)

Early points may be below the line

(incomplete mixing)
Richard Lawson Central Manchester Nuclear Medicine Centre
 Rutland Practice (continued)
Calculate relative function
Using ratio of slopes from Rutland fit
Subtract vascular background
Using factor from Rutland fit intercept
Display background subtracted curves as usual
Guarantees that curves start at zero

Relative function
Left kidney: 81%
Right kidney: 19%

Richard Lawson Central Manchester Nuclear Medicine Centre

 Summary - Rutland
The Rutland plot is a mathematical manipulation
of the renogram that simulates what would
happen if blood activity was constant
Kidney Counts / Blood Counts

ke
t a Slope is a
U p measure of the
kidney function

Intercept tells us
Blood background how much blood
background to
subtract

Rutland Time
Rutland Plot
Richard Lawson Central Manchester Nuclear Medicine Centre
Deconvolution
 Simple Kidney Model
Renal Renal Blood Activity
Artery Vein
Bolus Input
Glomerulus
Renal
Tubules time
Kidney Activity
Transit

Impulse

Elim
Uptake
Renal
Retention

ina
Pelvis

tion
Function
Ureter time
Richard Lawson Central Manchester Nuclear Medicine Centre
 Mean Transit Time
Impulse Retention Function
MTT Mean transit time is
the average time for
a molecule to pass
through the system
Initial Height

Two areas are equal

Area
H(t) MTT = Initial height

time
Richard Lawson Central Manchester Nuclear Medicine Centre
 Idealised Renogram
Blood Activity Perfect bolus Input
But not practicable
Perfect injection
Direct into renal artery
No recirculation
time
Kidney Activity
Impulse Response Transit

But easy to interpret

Elim
Uptake, transit and
Uptake

ina
elimination are separated

tion
time
Richard Lawson Central Manchester Nuclear Medicine Centre
 But what about the real renogram ?
Blood Activity IV injection
Slow recirculation
Blood activity persists

time

Kidney Activity
Real Renogram
What shape is the
kidney curve ? ?

time
Richard Lawson Central Manchester Nuclear Medicine Centre
 Convolution
100 Must be linear 25

Must be stationary
70 50% 35
50%
50
50%

50

Input time

Represent blood curve as

series of bolus inputs time
Response
Richard Lawson Central Manchester Nuclear Medicine Centre
 Deconvolution
100 50 50 50 40
100 100 100 100 18
50%
70 25
40% 25 12
50
18
35 35
36
24 35 12
25

18
50 50 50
40 28 12
Input time 20
14

Response time
Richard Lawson Central Manchester Nuclear Medicine Centre
 Convolution
Bolus R4
R3
I(t)
R(t)
R2

I1
I2 R1
I3
I4
H(t)
Input time
H1 H2 H3 H4

Response time
Richard Lawson Central Manchester Nuclear Medicine Centre
 Deconvolution
Bolus R4
R3
I(t)
R(t)
R2

I1
I2 R1
I3
I4
H(t)
Input time
H1 H2 H3 H4

Response time
Richard Lawson Central Manchester Nuclear Medicine Centre
 Teaching Point
Given the input to a system I(t) and the response
to that input R(t), you can use deconvolution to
calculate the expected response to an ideal
impulse input
This is the impulse retention function , H(t)
The impulse retention function is easy to interpret
because
The initial height represents uptake
The duration represents transit
The downslope represents elimination

Richard Lawson Central Manchester Nuclear Medicine Centre

 How to do it
I
i
R (t ) = I (t ) * H (t ) Ri = j H i j +1 t
j =1

Matrix method 1 Ri i 1
As previous illustration H i = I i j +1 H j
I1 t j =1
Fourier transform
FT of a convolution is
H (t ) = F -1F {R(t )}

product of FTs F {I (t )}
Constrained optimisation
Find smooth solution consistent with noise

1. ISCORN Consensus report. Durand E, et al Semin.Nucl.Med. 2008, 38:82-102

2. Application of mathematical methods in dynamic nuclear medicine studies
Lawson RS. Phys. Med. Biol. 1999, 44: R57-98
Richard Lawson Central Manchester Nuclear Medicine Centre
 Renogram Deconvolution

Blood Activity Renogram

Deconvolution

Bolus Input Impulse Response Function

Richard Lawson Central Manchester Nuclear Medicine Centre
 Effect of Vascular Background
Before Deconvolution

Kidney Activity
=
Blood Activity + Renogram

Not easy to
remove
Vascular Background background
Richard Lawson Central Manchester Nuclear Medicine Centre
 Effect of Vascular Background
After Deconvolution

Kidney Activity
=
Blood Activity + Renogram
Easy to
remove
background
Vascular Background
Richard Lawson Central Manchester Nuclear Medicine Centre
 Renogram Example - Raw Curves
Activity-time curves

After smoothing

Richard Lawson Central Manchester Nuclear Medicine Centre

 After Deconvolution
Impulse Response Functions
Relative MTT
Function
LEFT 41% 10.5 min
RIGHT 59% 4.9 min

After trimming

LEFT KIDNEY
RIGHT KIDNEY

Richard Lawson Central Manchester Nuclear Medicine Centre

 Practical Considerations
System must be linear
OK
System must be stationary
No furosemide
No large pelvic contractions
Suitable ROI for blood input
Aorta or heart
With tissue background subtracted
Suitable ROI for kidney contents
Whole kidney
Renal parenchyma (whole kidney - pelvis)
With tissue background subtracted
Richard Lawson Central Manchester Nuclear Medicine Centre
 More Practical Considerations
Correct start time
Problems if kidney activity appears later than heart
Need good statistics
Use higher administered activity
Must smooth curves
But not too much
Identify plateau of retention function
Difficult if curves are too noisy
Height is used to measure relative function
Extrapolate to remove vascular background
Produce background subtracted renogram
By reconvolving subtracted retention function with
blood input curve

Richard Lawson Central Manchester Nuclear Medicine Centre

 Summary - Deconvolution
Deconvolution gives the renogram that would be
obtained if an idealised bolus injection
could be given straight into the renal artery
with no recirculation

Vascular
Background
Impulse
transit
Retention
Function

El
im
in
Uptake

at
io
Mean Transit Time

n
Time
Richard Lawson Central Manchester Nuclear Medicine Centre
Quantitative Parameters
 Renogram Components
Uptake
Elimination
Elimination
Difference
Activity

starts
Transit = kidney
contents
Uptake
only Renogram peaks when
rate of uptake = rate of elimination
Downslope when
rate of elimination is
greater than rate of uptake
Renogram

Time
Richard Lawson Central Manchester Nuclear Medicine Centre
 Quantifying Relative Function
During first 2 or 3 minutes there is no elimination
So we can calculate relative uptake from:
Relative counts in summed image
eg 1 to 3 min
Difficult to get correct background subtraction
Relative area under uptake phase of curves
eg from 1 to 3 min
Provided background subtraction is correct
Relative height of impulse retention function
After deconvolution
Provided plateau can be identified
Relative slope of Rutland plot
Using linear fit
Guarantees correct background subtraction
Richard Lawson Central Manchester Nuclear Medicine Centre
 Quantifying Absolute Function
Absolute function is much harder than relative
function
Compare kidney activity with administered activity
Using known camera sensitivity
eg Manchester method
By imaging dose syringe first
eg Gates method
(Gates GF. Am.J.Roentgenol,1982, 138: 565-70)
Compare kidney activity with blood curve
Calibrate by taking one blood sample
(eg Piepsz A et.al. Eur.J.Nucl.Med. 1977, 2:173-77)
Proper measurement requires formal blood clearance
Multiple blood samples
(ISCORN Report. Blaufox et.al. J.Nucl.Med 1997, 37: 1883-1890)

Richard Lawson Central Manchester Nuclear Medicine Centre

 Quantifying Elimination
Integrate blood input and fit to uptake phase (Rutland plot)
Extrapolate to later times = Zero output curve

Difference between
Uptake phase
Activity

zero output and

Renogram
= Urine output

Urine Output / Zero Output

= Renal Output Efficiency *

Renogram
Blood (input)

Time
* Chaiwatanarat et. al. J.Nucl.Med. 1993, 34: 845-848
Richard Lawson Central Manchester Nuclear Medicine Centre
 We have looked at two Methods
The problem
Variable input function
Complex curve shape & superimposed background
The solution
Standardise input function
Simpler curve shape with separable background
Delta function input Calculate using:
Perfect bolus injection
Impossible in practice Deconvolution
Bolus spreading and recirculation

Constant input
Rutland plot
Continuous infusion
Possible but complicated
Richard Lawson Central Manchester Nuclear Medicine Centre
 Comparison of Methods
Renogram Rutland Plot
Deconvolution
Quantifies uptake Quantifies uptake
Quantifies MTT ?
Facilitates vascular Facilitates vascular
background removal background removal
Very sensitive to noise Insensitive to noise
Sensitive to timing Robust against timing
errors errors
Spoiled by furosemide Still OK with furosemide

Richard Lawson Central Manchester Nuclear Medicine Centre

 Summary
Renogram
Real Blood Activity
Kidney
renogram
Background

Impulse
Background Retention
Decon- Bolus input Function

Integrate
volution MTT
Kidney

Rutland plot

Rutland Constant infusion MTT

Kidney

Background
Richard Lawson Central Manchester Nuclear Medicine Centre
 ISCORN Reports
Quantification of the renogram
Prigent A, Cosgriff PS, Gates GF, et al.
Consensus report on quality control of quantitative measurements of
renal function obtained from the renogram.
Semin.Nucl.Med. 1999 29:146-159.

Renal transit times

Durand E, Blaufox MD, Britton KE, et al
ISCORN Consensus on renal transit time measurements.
Semin.Nucl.Med. 2008, 38:82-102

Renal clearance
Blaufox MD, Aurell M, Bubeck B et al.
Report of the Radionuclides in Nephrourology Committee on renal
clearance.
J Nucl Med 1997; 37:1883-1890.
Richard Lawson Central Manchester Nuclear Medicine Centre