The Irregular Surface Compensator (ISC) - Dosimetric Verification

ISC plans have the same complexity as IMRT plans, which makes it nearly impossible to perform an independent Monitor Unit calculation. Our QA policy is to verify the treatment plan by measuring absolute dose, field by field. For this task, either the 2D-ARRAY or Portal Dosimetry can used. The latter is the most uncomplicated and quick way to verify IMRT dose.The former is more accurate. With the 2D-ARRAY, verifying a complete plan on the linac takes no more than 15-20 minutes, including setup and Array warmup. With Portal Dosimetry, it is even less.

Let's say this is the clinical plan which shall be verified with the 2D-ARRAY:

Fig.1: The clinical ISC plan. Two tangential ISC fields.
 

The first step is to prepare the verification plans in Eclipse. This is done by selecting the plan to verify and choose "Create verification plan ..." from the menu. This copies the plan to a CT-scanned phantom on a field-by-field-basis.

 

 Fig.2: Some decisions have to be made when creating a verification plan. Collimator is overridden to 0°, to get a better alignment between leaf movement and array chamber positions.
 

The next step is to calculate dose of both verification plans. The coronal dose plane in depth of the effective point of measurement "Planned Dose File" is exported as DICOM-file in absolute dose units.

 
 Fig.3: Each field in the verification plan receives the same number of Monitor Units as in the clinical plan (here: 170). With the MU fixed, absolute dose is fixed. The dose plane in the lower left corner is exported. Note that the central axis is not aligned with the central (14, 14) chamber, but between chamber rows 14 and 15. This matches chamber rows with MLC leaves.
 

The size of the plane is chosen to cover at least the full treatment field, but this is not critical. Also the resolution of the dose matrix is not critical, as long as it is well below the resolution of the 2D-ARRAY (VeriSoft will later interpolate the lower resolution distribution - the measurement - to the higher resolution distribution - the dose plane).

 
 Fig.4: Dose plane export: the plane size should at least cover the field. The "Burn marker pixel" is for poor people who still have to work with film.
 

In order to get rid of small calibration deviations of the linac, small inaccuracies during setup, small temperature and pressure misreadings, small HU inaccuracies etc., we copy one of the Verification Plans and paste it as Reference Plan. This gives the plan exactly the same isocenter as the Verification Plans. The fluence is deleted, the field size set to X=10cm and Y=11cm (why 11? see below). This plan is a conventional static plan. Dose is calculated, normalized to, e.g., 2Gy at isocenter (again, this is not critical, as long as it is absolute dose. One could as well calculate a FixedMU plan) and the coronal dose plane in depth of the effective point of measurement is exported as above. Taking this plan as reference means "believing that it is correctly calculated in the TPS".

 
 Fig.5: Reference plan which is believed to be "true". By referring to this plan, dose calibration deviations of the linac cancel out.
 

At last, one has to decide how the plan shall be delivered on the linac. Delivery can either be done automatically via ARIA or manually in Clinac Service Mode. If ARIA is chosen, one has to create an appointment in Time Planner. After the patient is checked in, she appears on the linac's Queue and can be loaded.

If one choses Service Mode for more flexibility (e.g., to repeat a measurement several times without fuzzing around with the verification system), a little more manual preparation is necessary: the dynamic MLC-files have to be exported so that they can later be loaded on the linac with MLC Software. This completes the preparation in the TPS. The files at this stage are organized in the following way:

 
 Fig.6: The file structure in the preparation phase, somewhere on the network. Keep your files organized!
 

There are only two files for each field that need to be accessed on the linac: the dynamic MLC file and the planned dose file. The dynamic MLC file is loaded to the MLC controller, the dose file is opened with VeriSoft.

How the measurement on the linac is done, can be seen elsewhere. Here, only some VeriSoft screenshots from the actual evaluation are shown:
Fig.7: The upper image is the calculated dose distribution (finer grid), the lower image is the measurement.
 
Fig.8: Clicking somewhere in the field draws two orthogonal dose profiles.
Fig.9: The Compare area of VeriSoft has been improved to show more information on one screen. The "failed" points are due to our Pencil Beam in Eclipse, which underestimates dose in the low dose area in the scatter region outside the field. This will change soon after commissioning of the AAA (Anisotropic Analytical Algorithm) is completed.
Fig.10: all colors can be chosen by the user - here dark green is selected for deviations below 1%
Fig.11: The Gamma Index Report displays the final result of the evaluation in the condensed form of a traffic sign.
 

next

 

 

back to Medical Physics home