Authors (including presenting author) :
Lee WY Venus(1), Wong WC Dennis(1), Steven Cheung(1), Dennis Ngar(1), Wong CS(1)
Affiliation :
(1) Clinical Oncology Department, Tuen Mun Hospital
Introduction :
Brachytherapy is a clinical procedure to place a radioactive substance inside the patient's body usually next to a tumor for treatment. The radioactive substance used widely is Ir-192 and an afterloader is used to deliver the source to patient without requires manual handling. Due to the high dose rate and radioactive nature of the source, daily quality assurance of Ir-192 source position is required before brachytherapy treatment according to the report number 56 issued by American Association of Physicists in Medicine. The test is usually performed using radiographic or radiochromic film or source position check ruler to verify if it is within the tolerance of 1mm. However, source position relative to the reference position is observer-dependent. Furthermore, radiographic film is going to be obsolete and the cost of radiochromic film is very high. The RavenQA system (LAP Laser, Germany) is a quality assurance device with a phosphor screen detector for performing the quality assurance tasks in external beam radiotherapy.
Objectives :
This study aims to demonstrate the feasibility of using the RavenQA system for brachytherapy quality assurance by using it with an in-house position benchmark phantom and image-based automatic source localization algorithm to give a quantitative measurement for the position of the source. This study also aims to eliminate the observer-dependency and the use of radiographic or radiochromic films from the traditional quality assurance procedure.
Methodology :
The in-house phantom is made by a 255 x 255 x 3 mm^3 acrylic panel which fits tightly with the detector panel. There are 13 catheters along the acrylic panel and each catheter is separated by 15mm. 10 metal beads are seeded beneath each catheter and all beads are separated by 10mm along the catheter. Therefore, there are totally 130 positions are tested. The position of each catheters are benchmarked by the reference marker from the factory. The phantom is first irradiated by an X-ray system on the RavenQA to capture a reference image, which is used for obtaining the benchmark coordinates of all the dwell positions. All catheters are connected to the microSelectron afterloader (Elekta, Sweden) and the source is stopped by 1 second for each dwell position. RavenQA records all the dwell time and positions for analysis. An image-based automatic source localization algorithm in python3 is then developed in order to extract the pixel coordinates of the source for each dwell position in each channel. The algorithm divides into 4 parts. The first part is preprocessing, a median filter is applied for each image in order to remove the salt-and-pepper noise. Secondly, the algorithm performs a channel clustering is to differentiate each channel from all the images. Thirdly, the algorithm performs a position clustering to get the center of each source position. Last but not least, by comparing the pixel coordinates of the center for source positions with the benchmark position coordinates, the position error for each position can be acquired.
Result & Outcome :
The result shows all the difference in positions between the reference and source positions in both x and y directions and the result of all the differences in positions are within 1mm. The algorithm and Raven successfully demonstrated the feasibility of automatically detecting the Ir-192 source center and correlated with the reference positions in the phantom to give a quantitative result. This would be highly useful as a daily quality assurance tool to give accurate result and highly increase the cost effectiveness. Surprisingly, the data not only shows the result of offset for individual position but also shows the accuracy of the gain of afterloader.