ESTRO 38 Abstract book

S528 ESTRO 38

Figure 1.d shows the computational prostate tissue after 80 Gy. The tumour density (number of tumour cells divided by the total number of cells) at the end of the treatment was used as output of the model. The Morris sensitivity method calculated, on the 21 computational tissues (73500 simulations in total), the mean and standard values (μ i * ± σ i ) of 100 elementary effects for each parameter. The Euclidean distance of the point (μ i *, σ i ) to the origin was the indicator of the impact of parameter i.

Poster: Physics track: Intra-fraction motion management

PO-0970 Advances in intra-fraction movement detection during stereotactic radiosurgery O. Semeniuk 1 , P. Sadeghi 1,2 , J.D. Farah 1 , J. Robar 1,2,3 1 Nova Scotia Health Authority, Medical Physics, Halifax, Canada ; 2 Dalhousie University, Department of Physics & Atmospheric Science, halifax, Canada ; 3 Dalhousie University, Department of Radiation Oncology, Halifax, Canada Purpose or Objective Detection of cranial position during stereotactic radiosurgery (SRS) is crucial for correcting for intrafractional motion. This abstract describes a novel technology that takes advantage of capacitive motion sensing (CMS) for real-time, 3D motion detection during SRS. The CMS system senses the capacitive signal from the cranium and provides advantages over systems that are based on x-ray imaging or use skin as the surrogate. Here, we report our progress on performance optimization for the CMS system with an emphasis on advances in sensitivity, detection precision and dynamic range. Material and Methods The performance of two capacitive-to-digital converters was evaluated to determine the most suitable candidate for practical applications. We compared the MPR121 (Adafruit Industries) converter performance (used in the proof-of-concept study [1]) against FDC1004 (Texas Instruments) – a device possessing an active shield technology. The active shield enables whole-system noise minimization, thus boosting sensitivity. Here, sensitivity was our primary performance metric, defined as capacitance change per unit of displacement; detection precision was defined as ratio of noise amplitude to sensitivity; and dynamic range was defined as the largest distance from the target at which we were able to detect a sub-millimeter displacement. CMS prototypes were tested with 5x5 cm 2 sensors in a parallel plate set-up and a head phantom made in-house. The phantom was 3D printed from a CT scan of a volunteer’s thermoplastic mask, layered with a copper foil to provide a capacitive signal, and subsequently grounded. A micro-stage and 6D Hexapod were used to simulate subjects’ motion with millimeter steps.

Results The table shows a ranking of the 34 parameters of the model, according to their mean Euclidean distances over the 21 tissues.

Results The sensitivity and detection precision of FDC1004-based CMS system were found to be at least an order of magnitude greater than the MPR121-based CMS system. A precision of 0.1 mm was reached with the FDC1004 sensors even at a large separation distance (34 mm) between the sensor and the ground plate, outperforming the MPR121 sensor which achieved precision of ~5 mm. When tested with the head phantom, the FDC1004 system also showed superior performance with submillimeter detection precision in all directions including anterior motion (reported to be most challenging orientation to detect

Table. Ranking of the 34 parameters of the model, according to their mean Euclidean distance to the origin Conclusion The Morris sensitivity analysis identified the duration of the cycle of tumour cells and the dose per fraction as the parameters having the greatest effect on the final tumour density after 80 Gy. The VEGF Michaelis-Menten maximun rate, the VEGF Michaelis constant and the VEGF diffusion coefficient had the lowest impact.