Search

Objective. Modern PET scanners offer precise TOF information, improving the SNR of the reconstructed images. Timing calibrations are performed to reduce the worsening effects of the system components and provide valuable TOF information. Traditional calibration procedures often provide static or linear corrections, with the drawback that higher-order skews or event-to-event corrections are not addressed. Novel research demonstrated significant improvements in the reachable timing resolutions when combining conventional calibration approaches with machine learning, with the disadvantage of extensive calibration times infeasible for a clinical application. In this work, we made the first steps towards an in-system application and analyzed the effects of varying data sparsity on a machine learning timing calibration, aiming to accelerate the calibration time. Furthermore, we demonstrated the versatility of our calibration concept by applying the procedure for the first time to analog readout technology. Approach. We modified experimentally acquired calibration data used for training regarding their statistical and spatial sparsity, mimicking reduced measurement time and variability of the training data. Trained models were tested on unseen test data, characterized by fine spatial sampling and rich statistics. In total, 80 decision tree models with the same hyperparameter settings, were trained and holistically evaluated regarding data scientific, physics-based, and PET-based quality criteria. Main results. The calibration procedure can be heavily reduced from several days to some minutes without sacrificing quality and still significantly improving the timing resolution from(304±5)psto(216±1)pscompared to conventionally used analytical calibration methods. Significance. This work serves as the first step in making the developed machine learning-based calibration suitable for an in-system application to profit from the method's capabilities on the system level. Furthermore, this work demonstrates the functionality of the methodology on detectors using analog readout technology. The proposed holistic evaluation criteria here serve as a guideline for future evaluations of machine learning-based calibration approaches., (Creative Commons Attribution license.)

Compared to photon therapy, proton therapy allows a better conformation of the dose to the tumor volume with reduced radiation dose to co-irradiated tissues. In vivo verification techniques including positron emission tomography (PET) have been proposed as quality assurance tools to mitigate proton range uncertainties. Detection of differences between planned and actual dose delivery on a short timescale provides a fast trigger for corrective actions. Conventional PET-based imaging of 15 O (T 1/2 = 2 min) and 11 C (T 1/2 = 20 min) distributions precludes such immediate feedback. We here present a demonstration of near real-time range verification by means of PET imaging of 12 N (T 1/2 = 11 ms). PMMA and graphite targets were irradiated with a 150 MeV proton pencil beam consisting of a series of pulses of 10 ms beam-on and 90 ms beam-off. Two modules of a modified Siemens Biograph mCT PET scanner (21 × 21 cm 2 each), installed 25 cm apart, were used to image the beam-induced PET activity during the beam-off periods. The modifications enable the detectors to be switched off during the beam-on periods. 12 N images were reconstructed using planar tomography. Using a 1D projection of the 2D reconstructed 12 N image, the activity range was obtained from a fit of the activity profile with a sigmoid function. Range shifts due to modified target configurations were assessed for multiples of the clinically relevant 10 8 protons per pulse (approximately equal to the highest intensity spots in the pencil beam scanning delivery of a dose of 1 Gy over a cubic 1 l volume). The standard deviation of the activity range, determined from 30 datasets obtained from three irradiations on PMMA and graphite targets, was found to be 2.5 and 2.6 mm (1σ) with 10 8 protons per pulse and 0.9 and 0.8 mm (1σ) with 10 9 protons per pulse. Analytical extrapolation of the results from this study shows that using a scanner with a solid angle coverage of 57%, with optimized detector switching and spot delivery times much smaller than the 12 N half-life, an activity range measurement precision of 2.0 mm (1σ) and 1.3 mm (1σ) within 50 ms into an irradiation with 4 × 10 7 and 10 8 protons per pencil beam spot can be potentially realized. Aggregated imaging of neighboring spots or, if possible, increasing the number of protons for a few probe beam spots will enable the realization of higher precision range measurement.