Penn/Ohio Fall chapter MeetingSeptember 21, 2012

Jennifer Sessions, ME, DABR

• Headquarters in Melbourne, FL

• Privately held since 1984

• Aprox.140 employees- 14 in customer service team - 12 physicists

• Products available in over 60 countries

• ISO 13485 & 9001 certified/ FDA compliant

• Over 4000 installed users

About Sun Nuclear Corporation

• Quality assurance instruments using both diode and ionization chambers

• SunPoint™ Diode Detectors• Numerous OEM cooperative partnerships

The Right Detector for the Right Application

QA as a Function of Patient Safety

� QA devices that are easy to use are more likely to be used� Leads to error discovery and continuous improvement� How?

� Ease of Use� Standardization� Automation

� Products that increase functionality: � ArcCHECK™ VMAT Array� 3DVH™ QA Software� DoseLab™ TG-142 and Image QA Software� 3D SCANNER™ Water Phantom� ATLAS™ QA Management Software

Introduction

ArcCHECK™ VMAT Array with3DVH™ QA Software

&

3D SCANNER™ Water Phantom

Today’s Discussions

QA Challenge for Rotational Beams

• Modern technologies are becoming rotational� (RapidArc®, VMAT, TomoTherapy®, IMAT)

• Rotational beam delivery creates a challenge for patient specific QA

• An ideal QA tool would be a volume detector, such as Bang Gel, but this is not always practical. SNC developed the MapCHECK and MapPHAN combination as an effective solution. However, SNC developed the ArcCHECK as the best QA solution

What is 3D?

Is this 3D?� No� A 2D detector

plane designed for IMRT QA

Is this 3D?� No� Two orthogonal 2D

detector planes designed for IMRT QA

The Future in 3D

Make the detector plane “coherent” to the beam

Is this 3D?

�Yes!3D

An isotropic 3D array is defined by the detector geometry, not just by the phantom shape around the detectors!

ArcCHECK Introduction

• 1386 diodes in a helical geometry� Patented detector orientation

• 21cm diameter, 21cm length� Combine feature extends to 36cm length

• 1cm spacing, 3.3 cm effective depth� Merge feature reduces to 0.5cm spacing

• Virtual Inclinometer� Independently calculate gantry angle

• 4th Dimension = Time� Multi-arc control point analysis� 50ms update frequency

• Weight: 16kg U.S.Patent No. 8,044,359

Detector Geometry

• Entrance and exit dose are measured� Effectively doubling the detector density� Provides measurement capability at multiple depths – more stringent� Measure beam angles (Virtual Inclinometer)

• Central 10x10 contains ~221 detectors� Same as MapCHECK 10x10

• Detectors are arranged on a HeliGrid™� Increases sampling rate and reduces detector overlap from the BEV� Effective detector spacing could be as small as sub-mm

Why Cylindrical?

• Place detectors in best geometry possible to measure rotational beam• Forces TPS to account for topography and heterogeneities• Phantoms are ideally shaped like a patient

� ArcCHECK emulates patient geometry� Solid inserts are available to provide homogeneous density

• ArcCHECK features a versatile central cavity� May be used to accommodate different detectors and inserts� MultiPlug (left) also accommodates EBT film cassette and multiple

detector locations

Central Cavity Options

SNC Patient Software

• Software enables composite and control point analysis• DICOM RT Dose is imported and ArcCHECK software then extracts 3D

dose corresponding to detector locations, and performs a comparison• All data files from ArcCHECK are open format for easy export

� Raw data also available for export

Control Point Analysis

Virtual Inclinometer

3DVH Option

• The most advanced 3D patient dose and DVH tools available� Uses existing ArcCHECK measurements� Supports VMAT and RapidArc deliveries� No secondary dose calculation algorithm� 3D dose and DVH analysis on patient geometry (not phantom geometry)

What is 3DVH?

• Next Generation System for Dose QA� Dose-to-Patient estimated (no more guessing based on dose-to-phantom)

• Unrivaled Analysis Tools for 3D Dose & DVH� Designed for the Physicist/Dosimetrist/Physician who “wants it all” in one easy-to-

use system.

• Multiple and Distinct Uses:1. IMRT QA with Clinically-Relevant Analysis

⁻ Use conventional Dose QA tools to accurately predict dose in the patient using patented “PDP” method

2. VMAT/RapidArc QA with Clinically-Relevant Analysis⁻ Use conventional Dose QA tools to accurately predict dose in the patient using

patented “PDP” method3. Universal Plan Comparison

⁻ Compare and analyze Dose/DVH from any DICOM RT datasets (all TPS, all delivery modalities, all calculation settings, …)

?

Dose QA Today

What do these errors mean?

Are they clinically significant?

What passing criteria are best?

How Should We Approve Dose QA for a Plan?

• We learn a lot from diligent Dose (IMRT) QA…� We detect delivery errors.� We detect TPS errors and imperfections.� We fix the problems.

• But we still do not know which Dose (IMRT) QA Criteria are good predictors of the impact to patient dose…� Thought questions:

– How did you choose your % difference/DTA/Gamma criteria? Do you know what your “%” is really a % of (i.e. what is the absolute dose error for X%)?

– How do 3% / 3 mm DTA results correlate to 3% / 3 mm DTA results in the patient?

• Let us return to a simple question:� How was the plan dose approved in the first place?

Do Current IMRT QA Metrics Predict Impact toPatient Dose?

• "Per-beam, planar IMRT QA passing rates do not pred ict clinically relevant patient dose errors"Med. Phys. 38, 1037 (2011)

• Purpose: The purpose of this work is to determine the statis tical correlation between per-beam, planar IMRT QA passing rates and several clin ically relevant, anatomy-based dose errors for per-patient IMRT QA. The intent is to as sess the predictive power of a common conventional IMRT QA performance metric, the Gamma passing rate per beam.

• Conclusions: There is a lack of correlation between conventional IMRT QA performance metrics (Gamma passing rates) and dose errors in an atomic regions-of-interest. The most common acceptance criteria and published actions le vels therefore have insufficient, or at least unproven, predictive power for per-patient IM RT QA.

Do Current IMRT QA Metrics Predict Impact to Patient Dose?

How was the Treatment Plan Approved?

Do Current IMRT QA Metrics Predict Impact to Patient Dose?

3DVH Analysis

3DVH in Not Another Dose Algorithm

Replacing Dose/IMRT QA with an independent dose calculation algorithm introduces more potential questions…

� Is the QA dose calculation algorithm any better than your TPS?� Does the dose calculation algorithm introduce errors which were not

there to begin with?� How long will this independent calculation take?� What additional commissioning effort is needed?

?

Proven Accuracy with PDPTM

Planned Dose Perturbation (PDP)

U.S. Patent No. 7,945,022

Independent Research

• Maria Chan, Ph.D.� Memorial Sloan-Kettering Cancer Center

– “3DVH allows for a 3D patient specific comparison of the TPS generated plan with the treatment delivery…Most hidden errors could not be easily discovered in routine QA. Each potential source of errors found by 3DVH has been verified to be true”

• Michael Lawrence, PhD., Sha Chang, Ph.D.� University of North Carolina (UNC)

– Publishing research similar to UW showing minimal correlation between current QA passing rates and impact on patient. UW’s research on phantom, UNC’s is on patient data

• Genevieve Jarry, PhD� Maisonneuve-Rosemont Hospital

– Researched accuracy of 3DVH PDP algorithm. Modified their TPS model as a result of errors caught by 3DVH.

• Michelle Nielsen, MS, Miller MacPherson, Ph.D.� Princess Margaret Hospital

– Validation research on accuracy of PDP algorithm

PublicationsVMAT QA: Measurement-guided 4D dose reconstruction on a patient• Vladimir Feygelman Moffitt Cancer Center, MedPhys July 2012, AC-PDP

3D DVH-based metric analysis versus per-beam planar analysis in IMRT pretreatment verification.

• Carrasco P, Jornet N, Latorre A, Eudaldo T, Ruiz A, Ribas M., Med Phys. 2012 Aug;39(8):5040-5049.

Evaluation of software systems estimating patients planned dose errors based on planar IMRT QA measurementsM.Bakhtiari et al., Radamerica II LL

• AAPM 2012 Presentation - SU-E-T-37 Sunday 3:00:00 PM - 6:00:00 PM Room: Exhibit Hall

Per-beam, planar IMRT QA passing rates do not predi ct clinically relevant patient dose errors.

• Nelms BE, Zhen H, Tomé WA. Med Phys. 2011 Feb;38(2):1037-44.Moving from gamma passing rates to patient DVH-base d QA metrics in

pretreatment dose QA.

• Zhen H, Nelms BE, Tome WA., Med Phys. 2011 Oct;38(10):5477-89.

1 Min Summary of Vladimir’s PaperVMAT QA: Measurement-guided 4D dose reconstruction on a patient• Moffitt Cancer Center, MedPhys July 2012, AC-PDP• Ion Chamber point doses –

� 0.125cc chamber – 4 points per patient along lateral and vertical lines through the isocenter (5mm apart)

� Ion Chamber vs. TPS– 1.0+/-0.8% difference

� Ion Chamber vs. 3DVH– 0.1 +/- 1.0% difference

• Film - Gamma Absolute Dose analysis (Coronal & Sagittal)� Film vs. TPS

– Gamma passing rates

� Film vs. 3DVH– Gamma passing rates

• AC – PDP details • 4D Gating proof of concept

1%/2mm 2%/2mm 3%/3mm

83.0% 91.1% 98.4%

1%/2mm 2%/2mm 3%/3mm

88.6% 96.1% 99.5%

3DVH proved more accurate than Valdimir’s obviously well-

commissioned TPS!

Summary

• Supports IMRT, VMAT, RapidArc• Clinically relevant patient QA

� Patient DVH� Plan documentation matches QA documentation

• Eliminates dose to phantom guesswork• No commissioning required• No forward dose calculations required

� Does not introduce additional sources of error

• Cold and hot spots impacts to DVH� See actual dose fall off in target� See actual dose impact to critical structures� See why more stringent QA is required today

3D SCANNER

3D SCANNER Influencers

• AAPM Task Group 106: Accelerator beam data commissioning equipment and procedures, states� “Beam data commissioning should be independent of users and scanning systems if it

is performed with appropriate knowledge and proper tools”

• Experience during development of EDGE Detector� Sharp penumbra, small fields

• Experience as recipient and principal investigator in Linac beambenchmark data acquisition study, granted by the U.S. NIH/NCI

Challenges

• Challenge 1:� Eliminate setup subjectivity

– User dependent, based on experience level– Leveling of tank/scanning mechanism– Positioning at water surface, followed by proper chamber shift– Aligning chamber with Linac central axis

• Challenge 2:� Eliminate inconsistent detector orientation/resolution & scanning

– Smallest dimension of chamber may not always measure beam edge for sharpest penumbra

– Different mechanical motions for in-line and cross-line

• Challenge 3:� Increase efficiency

– Setup – can be very tedious and user dependent– Tank shifts disturb the original setup– Software workflow and post-scan processing can be slow

Challenge I: Setup Subjectivity I

• Setup quality is subjective and dependent on the user• Setup is manual and time intensive• Solution: AutoSetupTM of 3D SCANNER

� Motorized platform – 2 Vertical (Z) Motors, X- motor, Y-motor� 0.1mm/0.1o accuracy

� Less than 20 minutes

AutoSetup TM Process� Leveling the 3D SCANNER

– Water level sensing automates using 3 widely set points to determine if tank is level; repeats to confirm

� Detecting the Water Surface– Water level sensing automates detector positioning and offsets to detector effective depth if selected

� Align Ring Center to the Collimator Axis– Two profiles find the Diameter Drive’s center, dependent on detector in use

– Two additional profiles find Collimator Axis, and X,Y motors shift tank into alignment

� Align Ring Angular Orientation to the Collimator Axis– Verifying that the 0o and 90o of collimator align with the crossline and inline scan trajectories perfectly

� Measuring/Correcting Hysteresis– Two profiles find any hysteresis and correct - less than 0.1mm of hysteresis in the 3D SCANNER

Auto Leveling and Zero Point

Challenge I: Setup Subjectivity IITank Shifts disturb the original setupSolution: Full 65cm scan length removes the need for shifts

� 40x40cm Scans require no tank shift (100SSD, Depth 30cm, 5cm tail)

• Lift tables often rest on Linac floor ring, if ring is disturbed, setup needs to be re-verified and corrected

• Solution: 3D Lift TM and 3D MiniLift TM span all rings� Accidental setup disturbance avoided

Challenge I: Setup Subjectivity III

3D MiniLift TM 3D Lift TM

Example: Setup Subjectivity EliminatedTwo datasets, two users, same results

Day 1:

An experienced user positioned the tank under the Linac, filled it, used AutoSetup, acquired three profiles, then emptied the tank and put it to the side. The Linac was used clinically for the day.

Day 2:

A novice user (trained only once, a month before) put the tank back into position under the Linac, filled it, used AutoSetup, and acquired the same three profiles.

Results:

If you look closely you can see there are 2 profile sets, but the profiles overlap almost perfectly! An experienced user and a novice user get repeatable results with AutoSetup. (Note: no processing was done to these profiles).

Challenge II: Inconsistent Orientation I

Different orientations create different penumbrasSolution: Keep detector orientation consistentMore consistent, better scan quality

Challenge II: Inconsistent Orientation II

• Ideally scans should be performed with the detector axis perpendicular to the scan direction to minimize volume averaging

• Solution: Cylindrical design with ring drive mechanism

� Smallest dimension of chamber always measures beam edge – sharpest penumbra

� Consistent cable in beam minimizing Stem/Cable effects in scans

– Chamber always running along the Diameter Drive – Wrapping of cable around the outer ring edge

� Inline and Diagonal scans no longer have mechanical drive arm(s) traveling through water – minimizes ripples for shallow scans

� Ripples more quickly deconstructed with circular design� Full scan length eliminates tank shifts

Challenge III: Increase Efficiency• Using a 3D water tank has been a very time consuming process prior to

3D SCANNER

• Solution: Overall Design - AutoSetup, Cylindrical design, Efficient software, Lift Table� AutoSetup for fast and accurate setup of 3D SCANNER� Use the same cable for all SNC products – only 1 cable through the physics

port & no extension triax cables needed� No time consuming tank shifts

� No Gain reset in between scans due to large range of electrometer� No issues with disturbed setup due to Linac floor ring disturbance� Faster fill/drain time because no wasted space in the tank (less volume)

� Less time for water to settle (surface ripple is reduced with cylindrical design)� Scan speed can be increased due to reduced ripple because chamber is

always running along the diameter drive (instead of mechanical drive arms moving through water as in a square tank)

SNC Dosimetry Software• Next generation software…

� No resetting of gain, resulting in less time between scans� Queues are easily created and edited� TPS-specific queues for commissioning are pre-loaded in software� Ongoing scans can be visualized from queue-creation screen

• Database architecture� Database reduces time to search/organize data� Data processing layers may be performed in batch format and for multiple scans� Original data never lost – processing layers reversible at any time to restore raw data� File sharing feature allows easy sharing of scan data between users and computers� Open format files (.xml, .txt) allow easy export to spreadsheets for research/analysis

• Fixed window architecture� Offers familiarity and flexibility without complexity of customized views

• All TPS exports included

• Unlimited software seats per 3D SCANNER system

• Same software platform for all Sun Nuclear Dosimetry products� 3D SCANNER, 1D SCANNER, PC Electrometer

SNC Dosimetry Software

40 x 40cm field at SSD 100 & Depths 1.5, 5, 10, 20, 30cm� No Tank shift required

SNC Dosimetry SoftwareDiagonal & Large Field Scans

40 x 40 and Diagonal Scans

In a standard square tank, a tank shift would be necessary and partial or half scans would be acquired. In 3D SCANNER the full scan can be acquired without shifts.

On the Diagonal scan, note a primary collimator difference visible in the penumbra. In a square tank only one side of a half scan is generally visible. (Note –Diagonals must be done at 80SSD to get the full 5cm of tail at depth of 30cm.)

Thank you,Questions?

AutoSetup Info follows

How to Align a Polar Coordinate System to aCartesian Coordinate System?

LINAC

3D SCANNER

1st StepLevel

• Three point leveling• Leveling platform

� One fixed point Pivot point

� Two adjustable pointsZ-axis motors in leveling platform

• Leveling positions� Known (X,Y) positions of three points� Found (Z) positions

• AdjustmentBased on comparison of (Z) positions

2nd StepDetermine the Ring Center

• One of most import parts of setup routine

• A ring’s center defines its coordinate system

Polar coordinates

• Ring center is also the ring’s axis of rotation

• Can vary due to universal detector holder

x

y

1st RC

2nd RC

Effect of an Offset Ring Center

x

y

Well Defined Ringer Center

3rd StepCenter Tank

• Leveling platform� X-axis movement

� Y-axis movement

• Determine offset between ring center and collimator axis of rotation

• Need to use a 180°collimator rotation

x

y

Offset Beam Center

Collimator 0°

x

y

Offset Beam Center

Collimator 180°

x

y

4th StepDetermine Angle Offset

• Correct for difference between ring’s home position and the LINAC coordinate system

• Two scans� 10°movement from ring’s home position� 35°movement from previous positions

• Determine change in field size to measure angle offset

x

y

x

y

x

y

x

y

Final Result

x

y

x

y