Determination of Radiation Delivery Parameters of Medical Linear Accelerators using Data Analytics Pipeline

• Published in: 2022 IEEE-EMBS Conference on Biomedical Engineering and Sciences (IECBES) pp. 40 - 45
• Main Authors: Zubair, H.T., Tan, Yi-Fei, Basaif, Azmi, Oresegun, Adebiyi, Zin, Hafiz, Bradley, David, Azhar, Hairul, Rashid, Abdul
• Format: Conference Proceeding
• Language: English
• Published: IEEE 07-12-2022
• Subjects: as well as QA parameters; Data analysis; Linear particle accelerator; Maintenance engineering; Pipelines; Radiation therapy; Real-time systems; Schedules; This work establishes the exploratory data analysis framework for the assessment of linac signals. It derives key characteristics of the pulse delivery that can associate with dosimetry
• DOI: 10.1109/IECBES54088.2022.10079318

Radiotherapy treatments involve the delivery of sharp radiation pulses of 2 to 4 microseconds duration over typical total periods of 30 to 300 seconds at a rate of up to 400 pulses per second. Recent developments in optical fiber-based radioluminescence/scintillator systems offer radiation-sensing capabilities that capture signals from individual pulses. Each of these signals has unique characteristics which provide insights into the parameters of radiation delivery. Current data acquisition methods commonly rely on hardware-based charge integration methods for radiation dose calculations and have limited utilization of the acquired data for further insights or applications. In this paper, a data analytics pipeline for the extraction and processing of data from a Ge-doped real-time dosimetry system is presented. The data, as obtained for an Elekta Synergy radiotherapy system, is then analyzed for dose distribution, dose-rates determination, and signal clustering. The gathering and processing of such time-resolved data would enable applications such as fault analysis, auto-calibration, and equipment fault prediction in medical radiation facilities in addition to enhancing the routine QA process.