Informatics driven Quality Initiatives Using the Tools of Six Sigma
William Pavlicek PhD
AAPM, Charlotte, NC
Aug 2, 2012 2:50PM
©2011 MFMER | slide-1
Outline of Talk
• Six Sigma Concepts
• PET Patient Dose reduction
• CT Patient Dose reduction – positioning and Z
axis coverage
• Fluoroscopy – Lowering Patient Skin Dose
• Questions
©2011 MFMER | slide-2
W. Edwards Deming
• Taught engineering,
physics in the 1920s,
finished PhD in 1928
• Met Walter Shewhart at
Western Electric
• Long career in
government statistics,
USDA, Bureau of the
Census
W. Edwards Deming, 1900 – 1993
3
©2011 MFMER | slide-3
1
W. Edwards Deming
• Deming was asked by Japan to lecture on statistical quality control
to management
• Japanese adopted many aspects of Deming’s management
philosophy
• Deming stressed “continual never-ending improvement”
• “If I had to reduce my message for management to just a few
words, I’d say it all had to do with reducing variation”. (Deming,
as quoted in The Deming Dimension, Henry R. Neave (SPC
Press, 1990), p. 57)
• Deming lectured widely in North America during the 1980s
4
©2011 MFMER | slide-4
Deming energized “Statistical Thinking”
• Integral aspect of Six-Sigma (3.4 per million)
• Six Sigma is a ‘goal’ not often achieved!
• Fundamentally different from Statistical Methods
• The definition is broadened to include:
• Operations Research tools
• Discrete event “Simulation’
• Other elements of ‘lean process’
*ASQ Statistics Division, http://www.asqstatdiv.org/stats-everywhere.htm
5
©2011 MFMER | slide-5
Defects Per Million Opportunities
©2011 MFMER | slide-6
2
How Variation Hurts …
* From Montgomery, D. C. (2009), Introduction to Statistical Quality Control 6th edition, Wiley, New York
7
©2011 MFMER | slide-7
The Motorola Six-Sigma Concept
• The Motorola Six-Sigma concept is to reduce the variability in the process so that the specification
limits are six standard deviations from the mean, or only about 2 parts per billion defective (Fig 111a).
• When the Six-Sigma concept was initially developed, an assumption was made that when the process
reached the Six-Sigma quality level, the process mean was still subject to disturbances that could
cause it to shift by as much as 1.5 standard deviations off target (Fig 1-11b).
• There is an apparent inconsistency in this. Process performance isn’t predictable unless the process
behavior is stable.
• No process or system is ever truly stable, and even in the best of situations, disturbances occur. The
concept of a Six-Sigma process is one way to model this behavior. Like all models, it’s probably not
exactly right, but it has proven to be a useful way to think about process performance, and has direct
application to manufacturing.* * From Montgomery, D. C. (2009), Introduction to Statistical Quality8Control 6th
edition, Wiley, New York. Figure
1-12.
©2011 MFMER | slide-8
The Generations of Six Sigma
• Generation I
• Focus on defect elimination
• Motorola, 1987-1993
• Generation II
• Focus on cost reduction
• GE, Allied Signal/Honeywell, 1994-1999
• Generation III
• Focus on value creation
• DuPont, 2000-present
9
©2011 MFMER | slide-9
3
Process Variation
• Variation is present in all processes and every aspect of the workplace
• Excess variation reduces process performance, decreases customer
satisfaction, and has a negative impact on the bottom line
• Applies to all aspects of a business
• Statistical methods, an integral component of six sigma, are useful in
shifting the process target to the desired level and reducing the
variation around this target
10
©2011 MFMER | slide-10
As has been said, “the mean never happens,” — a heroic 4-day delivery time on one
order, with an awful 20-day delay on another, and no real consistency. The customers
in this chart feel nothing. Their life experience hasn’t changed; one bit. These
customers hear the sounds of celebration coming from within GE and ask, “What’s the
big event; what did we miss?” The customer only feels the variance that we have not
yet removed. … Variation is evil in any customer-touching process. *
*Jack Welch, General Electric Company 1998 Annual Report
11
©2011 MFMER | slide-11
Patients feel the variation!
• Too long Waiting for their exam
• Lack of proper prep means waiting, poss
rescheduled exam
• Repeated exam due to xyz
• Multiple x-ray exams due to lack of available
priors
• Poss need of follow up exam due to confidence
– due to IQ?
©2011 MFMER | slide-12
4
Why “Quality Improvement” is Important:
• Patient goes for Out Patient Radiology exam: order, transport, waiting
room, exam prep, tech/device performance, x-ray room availability, image
quality, interpretation, communication to clinician.
• The patient experience has 10 components - is 99% okay?
©2011 MFMER | slide-13
Propagation of Error
-repeatability of the exam
-reproduced with diff techs/equipment/rads
©2011 MFMER | slide-14
Repeatability (i.e. back to back Ca Scores,
bone density, kVp 4X, etc)
©2011 MFMER | slide-15
5
Reproducibility: Different techs, devices,
patients…
Different Quality?
©2011 MFMER | slide-16
CT fluoro biopsy dose, breast compression,
fluoroscopy dose by physician,
etc.
©2011 MFMER | slide-17
Focus of Six Sigma is on Process
Improvement with an Emphasis on
Achieving Significant Impact
• A process is an organized sequence of activities that produces an
output that adds value to the organization
• All work is performed in (interconnected) processes
• Easy to see in some situations (manufacturing)
• Harder in others
• Any process can be improved
• An organized approach to improvement is necessary
• The process focus is essential to applying Six-Sigma methods
18
©2011 MFMER | slide-18
6
Organized Approach:
DMAIC (duy–may–ick)
©2011 MFMER | slide-19
‘D’efine the Goal – What is important!
(you are what you measure!)
Project Charter – Admin Directive (like an internal contract with yourself)
Problem Statement
Goal Statement
In/Out of Scope
Team & Time Commitments
Timeline / Milestone
Estimate Financial Benefits
Risks, Constraints & Compliance Issues Identified
20
©2011 MFMER | slide-20
The Primary Six Sigma Tools
• Process map
• Cause and effect analysis
• Measurement systems analysis*
• Capability study*
• Failure mode and effects analysis
• Observational study (regression)*
• Designed experiments*
• Control charts and out-of-control-action-plans*
* Statistical Tools
21
©2011 MFMER | slide-21
7
Use of Reference Values
Measure, Analyze and Improve!
©2011 MFMER | slide-22
General Comments
• One doesn’t really achieve Six Sigma
• Informatics with control charts are enormously
helpful
• Shining a light by sharing data is very useful!
• Change can be difficult but fun!
• Leadership is key: drive out fear!
©2011 MFMER | slide-23
©2011 MFMER | slide-24
8
Joint Commission SE #47
©2011 MFMER | slide-25
How to determine if doses are ‘out of
range’ (reproducible)?
©2011 MFMER | slide-26
Lots of Software
Solutions!
©2011 MFMER | slide-27
9
RIMS
System Architecture
DICOM
DICOM, HL7 Gateway
DICOM SR
HL7
Exam
Information Database
Dose
Tracking
System
WAN
Knowledge Base
Dosimetry Analyzer
Reporting &
Alerting Module
Emails
Texts
©2011 MFMER | slide-28
#1. PET-CT Six Sigma driven lower dose
©2011 MFMER | slide-29
PET Scanner ‘Blocks’ BGO - PMT
©2011 MFMER | slide-30
10
PET F-18
γ
Coincidence ?
γ
Register
Yes Event
©2011 MFMER | slide-31
Types of Coincident Events
True
Scatter
Randoms
Gamma ray (Singles)
Line of Response (LOR)
Prompts = True + Scatter + Randoms
©2011 MFMER | slide-32
2D PET – SEPTA!
All events used to reconstruct a slice
are detected within that slice.
©2011 MFMER | slide-33
11
3D Volume Acquisition
Events which cross several transaxial
slices are collected.
These events are used in the
reconstruction of each of the slices
crossed.
©2011 MFMER | slide-34
NECR kCounts per second
100000
Practical Clinical
Range for
Oncology
2D
80000
60000
40000
3D
20000
kBq/cc
0
10
10
20
20
30
40
30
50
60
40
70
50
80
60
70
80
mCi
©2011 MFMER | slide-35
Why wouldn’t one lower dose from 15-20
to 10 mCi (or lower) for 3D?
• What if patient is late?
• What if patient is early?
• What if patient has low uptake?
• What if patient is non-compliant?
Variation of time/dose has greater effect
with smaller injected dose!
©2011 MFMER | slide-36
12
Solution: Bulk F-18 Infusion System
No patient to patient variability of dose!
©2011 MFMER | slide-37
Problem: Uptake Room Restroom
Tech gets patient to scanner ‘on time’?
After 45min
©2011 MFMER | slide-38
Solution: Monitor Uptake Restroom
After 45min
©2011 MFMER | slide-39
13
Automated Report Summary by Room/Day
©2011 MFMER | slide-40
Problem: Muscle uptake when walking
Tool: Spaghetti Diagram
©2011 MFMER | slide-41
Solution: Minimum distance!
©2011 MFMER | slide-42
14
Six Sigma Results
• Variation in administered dose removed!
• Variation in ‘uptake time’ removed!
• Variation in transport uptake removed!
• These are ‘monitored’!
• Dose is LOWERED!
• Patients still provide unpredictable variation
©2011 MFMER | slide-43
Variability still present!
PET Muscle uptake
(variability of counts in lesions)
©2011 MFMER | slide-44
#2. CT Dose Index
Reduction
Define
Measure
Analyze
Improve
Control
©2011 MFMER | slide-45
15
Goal: AAPM Notification Values
©2011 MFMER | slide-46
©2011 MFMER | slide-47
DMAI “C”: Control Chart
©2011 MFMER | slide-48
16
Variation:
Non-Size-Adjusted DLP vs #CT Exams for MCA Scanners
January 2011
(0018,1030) protocol name: 6300_AbdPel_uro_w_fur
(0008,103e) series description: 10m AbdPel a300
Standardized:
Mean - 2SD
288 mGy-cm
Mean
705 mGy-cm
Mean + 2SD
1122 mGy-cm
N um be r o f C T Ex am s
10
8
8
6
5
4
4
5
3 3
3
1
2
1
1
00
0
50 0-54 9
55 0-59 9
0
???????
2
2
2
5
4 4
4
3
2
1
22
2
1 1 11
0
1
1
0
0
0 0 0 0 0 0 0 0
1 1
0 0 0 0
0
≥ 1300
12 50-1 299
12 00-1 249
11 50-1 199
11 00-1 149
10 50-1 099
10 00-1 049
95 0-99 9
90 0-94 9
85 0-89 9
80 0-84 9
75 0-79 9
70 0-74 9
65 0-69 9
60 0-64 9
45 0-49 9
40 0-44 9
35 0-39 9
30 0-34 9
25 0-29 9
DLP w/o patient size adjustment
(mGy-cm)
©2011 MFMER | slide-49
Analyze - Why Variation of DLP?
• Really Large Patient?
• Really Tall Patient?
• Elevated CTDIvol prescription?
• Other cause(s)?
©2011 MFMER | slide-50
©2011 MFMER | slide-51
17
Same patient 30 days apart
Can you see the mAs/centering?
©2011 MFMER | slide-52
Not Centered
Centered
23 mSv
13 mSv
40 cm Projected SCOUT
36 cm Projected SCOUT
©2011 MFMER | slide-53
Variation in DLP seen with variation in
centering
©2011 MFMER | slide-54
18
ANALYZE: Variation due to Scout Order
(180 mAs)
8.09 mGy
9.45 mGy
9.1 mGy
8.09 mGy
©2011 MFMER | slide-55
ANALYZE new example
CT of Abdomen/Pelvis 1/16/2011
120kVp
180 ref. mAs
CT of Abdomen/Pelvis 4/15/2011
120kVp
180 ref. mAs
14.9mGy
18.4mGy
23% increase in dose between these scans on the same patient.
©2011 MFMER | slide-56
CT Center FOV Positioning
©2011 MFMER | slide-57
19
% Dose Increase Due to Position
%mGy Increase Isocenter:Displaced
0.8%
-9.0%
0%
-1.0%
2.5%
8.9%
4.5%
Position changes mAs lookup table.
©2011 MFMER | slide-58
Same Patient
CT of Abdomen/Pelvis 1/16/2011
120kVp
CT of Abdomen/Pelvis 4/15/2011
120kVp
14.9mGy
18.4mGy
23% increase
on
the
same patient.
180in
ref.dose
mAs between these scans
180
ref.
mAs
Dose Affecting Parameters:
1. Sliderboard
2. X-Y Position
3. Body Width
4. Z-Axis Attenuation
5. Metal Monitor
©2011 MFMER | slide-59
Pareto Graph: Rank Change in CTDIvol
60%
50%
40%
30%
20%
10%
0%
Board Attenuation
Patient Too Low
Patient Wider
Z Axis Coverage Different
©2011 MFMER | slide-60
20
Variation: Z axis: Abdomen & Pelvis
©2011 MFMER | slide-61
New ‘standard’ Z axis coverage
©2011 MFMER | slide-62
©2011 MFMER | slide-63
21
Height/Weight with AP and Lateral Size
Use Random Forest Model estimates
potentially ‘excessive’ Z axis coverage
©2011 MFMER | slide-64
Six Sigma
Routine AbdomenPelvis
N= 300+ pts
Overscan
superior
1 cm
inferior
.4 cm
Top 10 Overscans
Superior (cm)
6.3
4.5
4.2
4.2
4.2
4.2
3.9
3.9
3.6
3.6
Inferior (cm)
11
7.4
6.5
5.3
4.2
3.8
3.5
3.2
3.1
2.9
©2011 MFMER | slide-65
CT Colonography
OK to exclude lung bases, just include colon
Good scan coverage
©2011 MFMER | slide-66
22
CT Colonography
N= 20 pts
superior
inferior
Overscan
6.4 cm
2.3 cm
Top 10 Overscans
Superior
(cm)
22.12
13.6
9.1
8.3
8.3
8
7.9
7.5
5.9
5.6
Inferior
(cm)
9.5
8
6.5
4.2
4.6
2.6
2.6
2
2
1
©2011 MFMER | slide-67
Six Sigma Results
• Standardize Protocol Names
• Standardize CT’s Techniques
• Standardize CT Radiograph
• Standardize Patient Centering
• Standardize Z Axis coverage
• Control Charts
©2011 MFMER | slide-68
#3 Six Sigma DMAIC
with Fluoroscopic Procedures
©2011 MFMER | slide-69
23
NEW NORMAL!
• New arterial and endoscopic procedures save
open surgery!
• Patients are having multiple episodes of
fluorsocopy
• 18% of currently scheduled patients have had
more than 1 fluoroscopy!
• Patients are bigger
• Most fluoroscopy is outside of Radiology
©2011 MFMER | slide-70
Technical vs Behaviorial
©2011 MFMER | slide-71
5 ALARA Technical Tasks
for reproducibility
• Standardize Protocol Names!
• Standardize Fluoro/Acquistions!
• Default with Low(est) Fluoro pulse rate?
• Default with Low(est) DSA/Cine frame rate?
• Default with Low(est) dose per pulse?
• Always use Copper (0.1mm is 40%!)
©2011 MFMER | slide-72
24
15
cine
OldFPS
Dose
7.5
FPS ½
cine
Current
Dose
(70-75 bpm, II)
(70-75 bpm, flat panel)
©2011 MFMER | slide-73
Use copper filter!
vary thickness Lucite 15 cm to 35 cm
20% Iodine solution for CNR/mGy measures
©2011 MFMER | slide-74
~ 10% Lower Contrast to Noise for 0.1 mm
Copper
CNR vs mm Cu Acquisition
160
140
120
15cm Lucite
CNR
100
20cm Lucite
25cm Lucite
80
27.5cm Lucite
30cm Lucite
60
35cm Lucite
40
20
0
0
0.2
0.4
0.6
0.8
1
mm Cu
©2011 MFMER | slide-75
25
40% Lower Dose with 0.1mm Copper!
Differential Dose Acquisition
P e rc e n t C h a n g e in D o s e fo r
A d d itio n o f C o p p e r
0%
-10% 0.0
0.2
0.4
0.6
0.8
1.0
-20%
15cm Lucite
20cm Lucite
25cm Lucite
-30%
-40%
27.5cm Lucite
30cm Lucite
35cm Lucite
-50%
-60%
-70%
-80%
-90%
mm Cu
©2011 MFMER | slide-76
Use at least 0.1 mm Copper!
• Every fixed room without exception
• Every protocol!!!! (especially obese pt
technique!)
• Portable C-arms (check with vendor)
• ALARA -100% exams with copper.
©2011 MFMER | slide-77
8 ALARA Behaviors
for reproducibility
• Proper collimation
• Minimize use of Hi Mag
• Return to Normal FOV Fluoro prior to DSA
• Use of Fluoro Save (with LIH)
• Optimize patient positioning
• Table distant from tube
• DSA and Tap Lightly
• Announce cumulative Gy levels
©2011 MFMER | slide-78
26
Collimation Is Key: improves IQ, lowers PSD
< 1 Gy
1 -2 Gy
2 - 3 Gy
3 - 4 Gy
4 - 5 Gy
5 - 6 Gy
6- 7 Gy
7-8 Gy
8-9 Gy
9-10 Gy
79
©2011 MFMER | slide-79
Skin Dose Rate (mR/min)
Behavior: choosing small FOV
Result: High dose fluoro AND high DSA
6000
5000
Low
Normal
4000
High
3000
2000
1000
0
19
17
13
10.5
8
7
6
Diagonal Size (in)
©2011 MFMER | slide-80
Single frame of II fluoro – fluoro save
0.7R/m
8in Fluoroscopy
1.2 R/m
7in Fluoroscopy
2 R/m
6in Fluoroscopy
©2011 MFMER | slide-81
27
II Image Quality and small FOV Spot
Normal
8in Single Shot
Normal
7in Single Shot
Normal
6in Single Shot
©2011 MFMER | slide-82
Control: Are Fluoro Saved images in PACS?
Informatic tools can monitor
XA radiation events
©2011 MFMER | slide-83
©2011 MFMER | slide-84
28
-
+
5/6/2011
5/13/2011
5/20/2011
5/27/2011
6/3/2011
6/10/2011
6/17/2011
6/24/2011
7/1/2011
7/8/2011
7/15/2011
7/22/2011
7/29/2011
8/5/2011
8/12/2011
8/19/2011
8/26/2011
9/2/2011
9/9/2011
9/16/2011
9/23/2011
9/30/2011
10/7/2011
10/14/2011
10/21/2011
10/28/2011
11/4/2011
11/11/2011
11/18/2011
11/25/2011
12/2/2011
12/9/2011
12/16/2011
12/23/2011
12/30/2011
1/6/2012
1/13/2012
1/20/2012
1/27/2012
2/3/2012
2/10/2012
2/17/2012
2/24/2012
3/2/2012
3/9/2012
3/16/2012
3/23/2012
3/30/2012
4/6/2012
4/13/2012
4/20/2012
4/27/2012
5/4/2012
5/11/2012
5/18/2012
5/25/2012
SET ALARA I and II Levels (AK is ok)
16000
IR Radiation Dose in mGy - EVAR
14000
12000
10000
Sentinel Event (15,000 mGy)
8000
6000
4000
2000
0
Date of IR Procedure
Radiation Dose in mGy
Too much!
Yellow
Red
Sentinel Event
©2011 MFMER | slide-85
Fluoroscopy Positioning Review
Excellent
Skin below IRP
with Fluoro!
©2011 MFMER | slide-86
Peak Skin Dose Calculator
©2011 MFMER | slide-87
29
©2011 MFMER | slide-88
Fluoroscopy
Service
OLD
Exposure Rate
CURRENT
Exposure Rate
GI Endoscopy/ERCP
3 ½ Pulses per second
3 ½ Pulses per second
Urology
15 Pules per second
3 ½ Pulses per second *
IR Room 1 Radiology
IR Room 1 Vascular
IR Room 1 DSA Aorta, C-E
10 pulses per second
10 pulses per second
4 Frames per second
7 ½ Frames per second *
7 ½ Frames per second *
2 Frames per second **
IR Room 2 Radiology
IR Room 2 Vascular
IR Room 2 DSA Aorta, C-E
15 Frames per second
15 Frames per second
4 Frames per second
7 ½ Frames per second *
7 ½ Frames per second *
2 Frames per second **
Cardiac Cath Rm 2 and 3
15 Pulses persecond FLUORO
15 Fr/second CINE
7 ½ Pulses per second FLUORO
7 ½ Fr/second CINE **
Electrophysiology Lab
Venograms
15 Fr/second FLUORO
30 Fr/second CINE
4 Fr/second FLUORO *
4 Fr/second CINE ***
©2011 MFMER | slide-89
Thank You!
• Six Sigma is a way of problem solving
• DMAIC
• Drive out variation – measure reproducibility
• Processes are the problem, not people!
• Team effort
©2011 MFMER | slide-90
30