Room: ePoster Forums
Purpose: Commercial software often fails in starshot analysis in case of blurred beam edges. One example is the beam from MRI guided Cobalt machines. This study is to develop a method to analyze starshot films/images with reduced uncertainty and improved efficiency.
Methods: The starshot films were digitized with a flat scanner. The analysis takes place in Matlab and has three steps. First, it plots a profile along a circle that transverses all beams. Second, it identifies the center of each beam at both entrance and exit based on the profile. The beam entrance and exit appear as valleys due to radiation induced hypointensity. A line connecting the two centers determines the beam central axis. Third, a circle with a minimum radius is found that is tangential to or crosses all beam central axis. A typical commercial software relies on finding a single global threshold for detecting all beam-legs. Therefore, it becomes challenging when the beams have blurred edges. Our program, instead, relies on a circular line profile that covers the entire intensity range.
Results: The analysis method was validated against commercial software using LINAC starshot films which showed clear separation between beam-legs. Based on 10 films for collimator, gantry, couch, and MLC starshot QA, the radius calculated from the commercial software was 0.23Â±0.04mm, and from our program was 0.20Â±0.04mm. For the gantry starshot on a Cobalt ViewRay machine, the commercial software either could not identify the beam-legs or failed the test. Our program, however, rendered reasonable results, 0.82Â±0.03mm (n=12).
Conclusion: An accurate starshot analysis method has been developed and validated. It is particularly beneficial in any situation a global threshold cannot be found to identify all beam-legs. This method can be applied to other beam types with substantial edge blurring such as electron, proton, and carbon ion.