The use of machine learning and other sophisticated models to aid in prediction and decision making has become widely popular across a breadth of disciplines. Within the greater diagnostic radiology, radiation oncology, and medical physics communities promising work is being performed in tissue classification and cancer staging, outcome prediction, automated segmentation, treatment planning, and quality assurance as well as other areas. In this article, machine learning approaches are explored, highlighting specific applications in machine and patient-specific quality assurance (QA). Machine learning can analyze multiple elements of a delivery system on its performance over time including the multileaf collimator (MLC), imaging system, mechanical and dosimetric parameters. Virtual Intensity-Modulated Radiation Therapy (IMRT) QA can predict passing rates using different measurement techniques, different treatment planning systems, and different treatment delivery machines across multiple institutions. Prediction of QA passing rates and other metrics can have profound implications on the current IMRT process. Here we cover general concepts of machine learning in dosimetry and various methods used in virtual IMRT QA, as well as their clinical applications.

Due to their Bragg peak, proton beams are capable of delivering a targeted dose of radiation to a narrow volume, but range uncertainties currently limit their accuracy. One promising beam characterization technique, protoacoustic range verification, measures the acoustic emission generated by the proton beam. We simulated the pressure waves generated by proton radiation passing through water. We observed that the proton-induced acoustic signal consists of two peaks, labeled α and γ, with two originating sources. The α acoustic peak is generated by the pre-Bragg peak heated region whereas the source of the γ acoustic peak is the proton Bragg peak. The arrival time of the α and γ peaks at a transducer reveals the distance from the beam propagation axis and Bragg peak center, respectively. The maximum pressure is not observed directly above the Bragg peak due to interference of the acoustic signals. Range verification based on the arrival times is shown to be more effective than determining the Bragg peak position based on pressure amplitudes. The temporal width of the α and γ peaks are linearly proportional to the beam diameter and Bragg peak width, respectively. The temporal separation between compression and rarefaction peaks is proportional to the spill time width. The pressure wave expected from a spread out Bragg peak dose is characterized. The simulations also show that acoustic monitoring can verify the proton beam dose distribution and range by characterizing the Bragg peak position to within ~1 mm.

No abstract

This is the first time 2D dose surface maps have been produced for the duodenum and provide spatial dose distribution information which can be explored to create models that may improve toxicity prediction in treatments for locally advanced pancreatic cancer.