Computer Vision Quality Assurance for Medical Linear Accelerators

Abstract

Radiation therapy is a non-invasive and highly versatile approach to combat tumor growth in cancer patients by utilizing high-energy X-ray beams targeted towards tumor tissues to inhibit growth. Radiation therapy is able to suppress tumor growth in almost all areas of the human body, making it a widely applicable treatment method. Medical linear accelerators (linacs) deliver radiation therapy to patients, with only 3,500 linacs in the US as of 2021 it is used by 60% of cancer patients, as they provide precise beam radiation. A radiation oncologist provides a detailed treatment plan tailored to each patient’s individual needs. The efficacy of the treatment strongly relies on the accuracy and precision of the linacs. However, through their continuous use, mechanical deviations will affect its positioning and calibration. Compliance is ensured through monthly Quality Assurance (QA) sessions where various tests are conducted with sub-millimeter precision and calibration thresholds. The current QA sessions utilizes a Computer Vision Quality Assurance (C.V.Q.A.) system that can conduct the required tests to ensure correct calibration. However, this current system exhibits a couple problems in terms of the use of high-cost instrumentation and the introduction of human subjectivity in distance measurements, more specifically, the Optical Distance Indicator (ODI). StriX intends to implement new and advanced design components such as laser detection methods as well as enhanced computer-vision code to perform C.V.Q.A. testing. It is estimated that there will be 2.04 million new cases by 2025. One patient can pay up to $50k for a course of radiation therapy, describing the incredibly large customer market as well as its high estimated market projection. The production market in 2025 is worth $4.21 billion, and is expected to grow to $6.22 billion by 2030.

Video