Discovering quality in radiotherapy through transmission ... · Is delivery QA beneficial? 1WHO...

Discovering quality in radiotherapy through transmission detection

Mark Geurts, Researcher and Adjunct ProfessorUniversity of Wisconsin, Madison, USA

January 20, 2018

Presentation outline

1. Introduction to University of Wisconsin

2. Review goals of patient-specific Quality Assurance

3. Explore role of transmission detection in QA

4. Evaluation of the ScandiDos Delta4 Discover

Author Disclosure

• ScandiDos AB and HGPT funded travel to this meeting

and provided an honorarium

• The University of Wisconsin is a Center of Excellence

for ScandiDos products

• Images in this presentation are from delta4family.com

University of Wisconsin, Madison

UW medical physics history

John Cameron

Thermo-Luminescent Dosimetry (TLD)

Bone densitometry

Frank Attix

Spencer-Attix cavity theory

UW medical physics history

Charles Mistretta

Digital Subtraction Angiography (DSA)

Vascular imaging

Thomas “Rock" Mackie

Pinnacle treatment planning system

TomoTherapy treatment system

UW Radiation Oncology by the Numbers

Delivery

• 4 Varian

• 3 TomoTherapy

• 2 Elekta HDR

• 1 ViewRay

Planning

• RayStation

• Pinnacle

• Eclipse

• TomoTherapy

• Oncentra

Quality Assurance

• 5 Delta4

phantoms

• Mobius3D

Delta4 history at the UW

• First unit commissioned in 2008

• Established as primary tool in 2012

• Upgraded to Delta4+ in 2016

• All UW centers now use Delta4

UW IMRT QA policy

• Performed prior to treatment start, or

before 3 fractions/10% on transfers

• >95% data points must pass 3%/3 mm

global Gamma, using 20% threshold

• Median absolute dose < 3%

Delta4 performance

• 2135 clinical plans since 2012

• Pass rate (3%/3mm): 96.3%

• Mean pass rate: 98.9%

• Absolute dose: 0.2% ± 1.3%

Delta4+ upgrade

• Pass rate increased from

95.8% to 98.8%

• Standard deviation

decreased 1.4% to 1.0%

• No service in 2 years

Goals of Patient Specific QA

How safe is radiotherapy?

• WHO survey from 1976-2007

identified 3125 incidents1

• Patients died in 38 of them

• Other studies2 report error rates

between 0.1% to 4.7%

1WHO Radiotherapy Risk Profile (2008). www.who.int2Ford (2009). IJROBP 74:852-8

Radiation therapy complexity

Huq (2016). Med Phys 43: 4209-62

What is patient specific QA?

Quality Management

Quality Assurance

Machine QA Patient Specific QA

Pre-treatment Review

Delivery QAOn-

treatment Review

Quality Control

• Quality controls force the

desired level of quality

• Quality assurance

demonstrates a level of

quality for a process

American Society for Quality. asq.org

Is delivery QA beneficial?In house ILS found delivery QA only detected 1.4% of all incidents:

Ford (2012). IJROBP 84: e263-9

Is delivery QA beneficial?

1WHO Radiotherapy Risk Profile (2008). www.who.int2Gopan (2016). Med Phys 43: 5181-7

• WHO1 found 10% of all errors

occurred during treatment

delivery, 9% in data transfer

• In a study of SAFRON

events2, 23 of 81 (28%) had

incorrect field parameters

(angle, MU, wedge, etc.)

PlanningCommissioning

Delivery

Data Transfer

Source of Incident

Plan quality sensitivity to delivery error

Fredh (2013). Med Phys 40: 031716

Plan quality sensitivity to delivery error

Arumugam (2016). Physica Medica 32: 1238-44

Recent surveys of medical physicists

Yes

No

Have you ever changed a plan based on IMRT QA?

On every

patient

Routinely

Not at all

Is absolute dose measurement necessary?

www.surveymonkey.com

Transmission Detection QA

What Is Transmission QA?

• Transmission QA systems place

an array of detectors between

the collimated beam and patient

• They allow intra-fraction

measurement of machine

parameters during treatment

Strengths and Weaknesses

Strength Weakness

Real time QA measurement Difficult to compare to TPS

Independent of treatment system Low energy discrimination

Output, gantry angle, and collimation Surface dose, beam quality affected

Accurate MLC position verification*

*some systems

Transmission vs. exit dosimetry

• 21st Century Oncology

evaluated 6.4 million EPID

exit doses across 84 centers

• 13% to 26% of all images

failed using 3% or 5% criteria50%

60%

70%

80%

90%

100%

H&NPelv

isBrea

stLu

ng

Prostat

e

Average Weekly Portal Pass Rate

Olivera et al. (2016) www.standardimaging.com

What about log file analysis?

Several groups have discovered errors not caught by log file:

Agnew (2014). Phys Med Bio 59: N49-63

What about log file analysis?

Neal (2016). Med Phys 43: 2933-5

Several groups have discovered errors not caught by log file:

What about log file analysis?Several groups have discovered errors not caught by log file:

Rowshanfarzad (2015). Br J Radiol 88: 20140581

Transmission detector comparison

*6 MV, 10x10

Li (2016). JACMP 17: 235-44

ScandiDos Delta4 Discover

Delta4 design

• 4040 p-type diodes with 1 mm active diameter

• Active area 25 x 20 cm at isocenter

• Diode spacing 1.6 (2.5) mm along MLC, 3.2 (5) mm against

• Overall system thickness 23 mm

Li (2016). JACMP 17: 235-44

Delta4 Discover evaluation

Parameters

• Transmission

• Surface dose

• Beam quality

• Beam profile

• Output sensitivity

• IMRT reproducibility

• MLC error sensitivity

References

• Riley (2011). Med Phys 38: 3534

• Li (2016). JACMP 17: 235-44

• Saenz (2016). Med Phys 43: 3539

• Hoffman (2016). Med Phys 43: 3538

• Gonod (2017). Phys Med 44: 30

• Cheung (2017). PRO 7: e559-67

Transmission

6 MV 10 MV 15 MV 18 MV Reference

0.989 0.993 Li (2016). JACMP 17: 235-44

0.988 0.993 Saenz (2016). Med Phys 43: 3539

0.985 Gonod (2017). Phys Med 44: 30

0.983 0.986 0.988 Hoffman (2016). Med Phys 43: 3538

Surface dose

Author Cheung Li

Field (cm2) 10x10, 40x40 4x4, 10x10, 30x30

SSD (cm) 80, 100 90

Energy (MV) 6, 10 6, 15

Detector OSLD Diode

Result 10 x 10 40 x 40

80 +35% +63%

100 +4% +32%

Beam quality

Author Hoffman Cheung Li Saenz

Field 1x1, 30x30 3x3 to 40x40 4x4 to 30x30 10x10

SSD 80, 100 90

Energy 6, 10, 15 6, 10 6, 15 6, 18

PDD 1% (> 2.5 cm)1.2% (100)

3% (80)

PDD(10):

-0.5%

PDD(10):

-0.7% (6)

+0.6% (18)

Dmax

4 mm (100)

8 mm (80)1.2 mm

Beam profile

Author Hoffman Cheung Li

Field (cm2) 30x30 3x3 to 40x40 30x30

SSD (cm) 80, 90, 100 90

Energy (MV) 6, 10, 15 6, 10 6, 15

Result 1.25%2.6% (100)

6.3% (80)

Flatness -0.3%

Symmetry +0.5%

Output sensitivityAuthor Li Saenz

MU error +1 to +5% -5 to +5%

Modality Static MLC field Static MLC field

Result

IMRT reproducibility

Author Hoffman Gonod

Modality VMAT VMAT

Plan type Prostate Cervix, Head and Neck

Frequency Measurements over 9 days

Metric Gamma pass rate (3%/3mm)

StandardDeviation

0.5% 1.1-1.5%

MLC error sensitivity

Author Li Saenz

Modality Static MLC Prostate VMAT Conformal arc

Leaf error 1-5 mm 0.5-5 mm 0-4 mm

Sensitivity 0.7 mm 1.2 mm

Result

Combination with Delta4 Phantom

0

1

2

3

Discover Phantom+

Discover

Phantom+

Phantom+

Phantom+

Discover

Discover

Error sensitivity relative to Delta4 Phantom

Riley (2011). Med Phys 38: 3534

3DCRT evaluation

Without a Delta4 Phantom measurement, Discover can still measure

and compare to the DICOM RTPLAN:

• Monitor Units

• MLC position

• Gantry angle

• Collimator angle

Anatomic evaluation

• Delta4 results can be used to

reconstruct dose to the planning CT

• Several groups have investigated:

Hu (2013). Med Phys 40: 213

Stambaugh (2014). JACMP 15: 169-180

Hauri (2014). JACMP 15: 181-90

Crockett (2015): Med Phys 42: 3411-2

Bängtsson (2016). Med Phys 43: 3370-1

Thank you for your time