MathResolutions, LLC

5975 Gales Lane

Columbia, MD 21045




Testing of MLC Gap Test Program

Feb. 5, 2009


Purpose: To test the validity and accuracy of using an EPID to measure the position of the leaf edge of a multi-leaf collimator (MLC).


Equipment: Elekta, Varian, or Siemens accelerator with an MLC; program MlcGapTest; x-ray film with x-ray film method.


Issues Involved:



The pixel size of the EPID will introduce an uncertainty of the order of + or half the pixel dimension. The program does sub-pixel interpolation from center of pixel to center of pixel. Removing the effect of the flattening filter will give a different shape of the field than what is otherwise measured, as with x-ray film. The Varian Portal Vision system is usually set to multiply the off axis effect back into the data from a profile measured in water at dmax. Program MlcGapTest can otherwise multiply back in the same effect when doing a deconvolution from in air measured data. In either case, the field shape is not directly measured and this may have some effect on the end result as the 50% point may be off axis considerably depending upon the field size set. Since the EPID has internal scatter, the effect will be similar to putting an x-ray film at depth in a phantom and measuring the profile at depth instead of in air. Adding scatter to both a point on axis and a point off will change the relative difference between the points. This can effect the location of the 50% fluence point. Program MlcGapTest provides a deconvolution process, whereby the image is convolved with the inverse of the point spread function (the kernel). In effect, this is a high spatial frequency pass operation. However, the process used to develop the point spread function, the kernel, fits the kernel to optimize getting the fluence so that the correct dose in water is computed from the converted image. Edge effects are not so important nor controlled in the fitting process and this may have an effect on results when looking for a field edge to the accuracy expected for testing leaf positions. Further more, we have not seen a significant difference between using the raw image versus the deconvolved image. At present we have not provide a way to restore the in air off axis ratios separate from doing the deconvolution.


All of these considerations suggest the need for a validation comparing results with that of using x-ray film. However, in measuring the leaf position in a narrow gap, the considerations above will have less an effect.


Method: Program MlcGapTest provides a means to find the central axis and major axes from images of a slit formed with the collimator jaw system. This coordinate system is then transferred to the test image of the MLC. One may choose to use only the test image for the same centering operation. Move the jaws that move parallel to the leaves to form a narrow slit image. Or rotate the collimator 90 degrees if the MLC replaced a jaw. You can take two images to product two slit images at 90 degrees to each other to further refine finding the central axis. Both EPID images and x-ray film images are to be taken. Care must be taken so that multiple images will all have the same exact registration.


Then take an image of the MLC opening, both EPID and x-ray film. You should image the MLC opening with the EPID and then with x-ray film without moving the leaves.


Use the tools provided in MlcGapTest to locate the central axis and any rotational correction that might be required. Locate the same leaf edges with the x-ray film method, and on both the MLC opening image and the deconvoluted MLC opening image, and record the results below.



Data: Do at least 10 leaf pairs. Field size set:


Date: Machine:








Epid De-





Epid De-