Characterisation of a CMOS APS for Application in image guided radiotherapy

Advanced radiotherapy techniques such as intensity modulated radiation therapy (IMRT) and intensity modulated arc therapy (IMAT) requires verification of complex treatments and tracking of moving tumours in real time. Current techniques using electronic portal imaging devices (EPID) and fluoroscopic based imaging have several limitations such as low image contrast and unnecessary skin dose. This work aims to investigate the feasibility of complementary metal oxide semiconductor active pixel sensor (CMOS APS) in overcoming these limitations, by utilising on pixel intelligence such as an on pixel analog-to-digital converter (ADC) and region of interest (ROI) functionality. Characterisation of two novel APS is presented and full performance parameters of the sensors had been calculated. Both APSs have a good dynamic range of ∼60 dB with gain of 19.9 e-/DN for sensor 1 (Vanilla) and 5.4 e-/DN for sensor 2 (LAS). Vanilla exhibits over 100 fps digital frame rate and over 20 kHz ROI acquisition speed which is superior to less than 30 fps using standard EPIDs or fluoroscopy. LAS stands for Large Area Sensor and is fabricated using stitching technology. Both sensors exhibit low spatial dark noise i.e. read noise which can be eliminated through dark image calibration. The problem of high temporal noise i.e. spiking pixels in Vanilla can be reduced through frame averaging. Future work will be carried on to exploit these sensors for radiotherapy treatment verification and tumour tracking through a combination of fast tracking algorithms and image enhancement techniques. The practicability of the system to incorporate image pre-processing in a field-programmable gate array (FPGA) is expected to allow the automation of treatment verification and real time tumour tracking. (author)