Measurement Guided Dose Reconstruction (MGDR)‐

Transitioning VMAT QA from phantom to patient geometry

Raj Varadhan, PhD, DABMPMinneapolis Radiation Oncology

Conflict of interest

• NoneAcknowledgement: Jim Ernsberger, SNC.

Transitioning from Phantom geometry to patient geometry for VMAT H & N QA:

Purpose:

The purpose of this study is to a) investigate the accuracy of patient DVH based QA using

volumetric measurement guided dose reconstruction (MGDR) for head & neck Volumetric

Modulated Arc Therapy (VMAT) delivery by benchmarking against RPC Head and Neck

Phantom results, b) investigate the correlation of 3D detector phantom based QA using

local gamma and global gamma results when compared against patient DVH based QA.

Methods:

The RPC head and neck phantom was planned and irradiated using 2 arc VMAT delivery

method. The results of point doses from RPC TLDs at eight locations and dose profiles from

radiochromic film through center of primary PTV inside the phantom were compared

against the treatment planning system (TPS) and the results from the MGDR analysis based

on 3D detector phantom QA. Also, the 3D gamma value was calculated using the MGDR

analysis which compares the TPS dose to the predicted dose distribution.

Five head and neck patients were planned using VMAT delivery method and the QA was

performed using both 3D detector phantom based QA and patient DVH based analysis

using MGDR. The correlation between gamma pass rates using both global and local gamma

criteria was compared against the gamma pass rates using MGDR analysis.

Results:

The ratio of MGDR calculated dose to primary and secondary planning target volume (PTV)

compared with both RPC TLD and Eclipse TPS doses was ≤ 1.05. The ratio of estimated

cord dose from MGDR analysis was ≤ 1.09 and 1.03 when compared to RPC TLD and TPS

doses respectively. The displacement between calculated dose gradient in the region

between primary PTV and organ at risk (OAR) in all 3 planes from MGDR was within 3mm

and 1mm when compared to RPC film and TPS profiles respectively.

The average 3D local gamma pass rate for the five clinical cases using MGDR was ≥ 97.5%

and 92% when using 3% 3mm and 2%2mm analysis criteria respectively. The average

gamma pass rate using global gamma 3%3mm criteria was 99.8%, compared to an average

local gamma pass rate of 87.1% using 3D detector phantom based QA with a maximum

increase of 18%.

Conclusion:

Benchmarking the accuracy of patient DVH based QA results to the RPC head and neck

phantom established a baseline accuracy and confidence in use of MGDR analysis for VMAT

delivery in a realistic patient geometry. For the 5 clinical test cases studied, the low pass

rates obtained using the local gamma evaluation criteria in phantom based QA had no

significant clinical impact for the patient when evaluated using DVH based QA.

Objectives

• Background• Status Quo adequate?• Review of Physics behind MGDR• Benchmarking accuracy of MGDR & comparison with traditional phantom QA

• Conclusions

Background:

• Traditional VMAT/RapidArc QA employs detector based geometry using the ubiquitous 3%/3mm dose and DTA threshold for gamma QA pass rates.

• The traditional QA detects errors (sensitivity)• However no information on the size and spatial location of errors in patient geometry. (specificity)

• Is status quo enough or accurate to detect clinically meaninful errors?

VMAT QA

• Every patient receives a patient specific QA.• ACR/ASTRO Practice guideline definition for IMRT/VMAT: …. “accuracy of dose delivery should be documented by irradiating a phantom containing a calibrated dosimetry system to verify dose delivered is same as dose planned”

• MGDR refers to a solution where errors in QA can be meaningfully correlated to patient specific geometry and structures

Gamma QAa)“Per‐beam, planar IMRT QA passing rates do not predict clinically relevant patient dose errors,” B. Nelms et al Med Phys, 38 January 2011: 1037‐44

Moving from gamma passing rates to patient DVH‐based QA metrics in pretreatment dose QA,” Zhen H. et al., Med. Phys. 2011 Oct;38(10):5477‐89

Is current QA/methodology adequate?

What does conventional QA results mean?

How to reconstruct 3Ddose? Commercially available

• ArcCheck ‐3DVH• Delta 4‐DVH‐anatomy• EPID‐Math Resolutions•etc.

ArcCHECK based MGDR

In BEV mode, the detector geometry is invariant of gantry angle

PDP or 3DVH 

3DVH

Physics of MGDR

• Basic premise is to use the errors measured in phantom geometry to “perturb” the TPS dose by a correction factor that faithfully represents the ACTUAL dose delivered to the patient.

6 step process going from phantom to patient in 3DVH

• Step 1 –AC Phantom measurement• MUST make the measurement with PMMA plug

Step 2‐Synchronize angle with time

• AC dose is updated every 50 ms ( in its own clock)

• RT Plan control points are a function of gantry angle NOT time

• In VMAT gantry angle is not linear with time (variable speed)

• Need to synchronize CP with Dose(t)• SNC does with virtual gantry angle inclinometer

Step 2‐Virtual Inclinometer

Step 3‐Calculate RELATIVE dose in uniform PMMA phantom

• Divide the dynamic RT plan(CP) into sub beams

• Process the time stamped sub‐beams to create the fluence

• Perform convolution “type” calculation for each sub beam to derive RELATIVE dose

Step 4‐ Convert sub beam dose to absolute dose

• Use entrance and exit measured absolute dose• Convert relative dose to  semi empirical

absolute dose• Called morphing because calibration factor 

changes along the ray• Larger weight given near the entrance diode

vs exit diode

Step 5‐ Sum the sub beams in phantom

Step 6‐Make the “jump” to patient geometry

PDP Model

• Is it accurate?• NOT interested in solid water and film

• Need validity in actual patient geometry • We used RPC H & N Phantom

VMAT Head & Neck Treatment Delivery‐Benchmarking Accuracy of 

Patient DVH based QA with Radiological Physics Center (RPC) Head and Neck Phantom Results. 

RPC HEAD & NECK PHANTOM

TLD Results comparison

Location Eclipse Mean RPC TLD(Mean) 3DVH(MEAN) (RPC/3DVH) (RPC/TPS)Primary PTV sup. 

ant. 671 682 672.9 1.01 1.02

Primary PTV inf. ant. 680 704 679.8 1.04 1.04Primary PTV sup. 

post. 692 707 671.7 1.05 1.02Primary PTV inf. 

post. 691 725 686.7 1.06 1.05

Secondary PTV sup. 552 567 547.5 1.04 1.03

Secondary PTV inf. 552 568 553.2 1.03 1.03

Organ at risk sup. 282 299 275.4 1.09 1.06

Organ at risk inf. 270 282 263.4 1.07 1.04

Dose units in cGy

RPC Film vs Eclipse TPS

MGDR (green) vs Eclipse TPSRight Left Profile

RPC Film vs Eclipse TPS

MDGR vs TPSAnterior‐Posterior Profile

RPC Film vs Eclipse TPS

MGDR vs Eclipse TPSSuperior Inferior Profile

3DVH analysis

Comparison of QA for Patient H & N VMAT plans

Patient 

Arc check QA local gamma pass rate

Arc check Global gamma

3D DVH local 

gamma pass

3D DVH2% 2mm

1 92.70% 100% 98.20% 93.80%2 81.50% 99.50% 97.5% 92%3 83.30% 100% 99.40% 96.20%4 90.80% 99.80% 99.20% 96.80%

ArcCHECK Phantom QA Patient 2

3DVH QA‐Patient 2

3DVH analysis

ArcCHECK Phantom QA‐Patient 3

3DVH‐Patient 3

CONCLUSIONS

• For VMAT, phantom geometry based gamma QA metrics have a weak correlation with actual dose delivered to patient.

• Need to take a step back and think what is the goal of QA? Are you checking the TPS model indirectly or errors that have patient specific clinical impact?

• If the answer is latter, then MGDR analysis is the likely solution.

Is current QA/methodology adequate?