This dataset comprises the data reported on by Werner et al. (2019) in Phys. Med. Biol. 64 105023, 20pp (https://doi.org/10.1088/1361-6560/ab176d). Please refer to this publication for details on the experimental setup, data acquisition and preprocessing. The process is summarised in the following.

A static, pulsed pencil beam was delivered to a target without and with cylindrical air cavities of 5 to 20 mm thickness and prompt gamma-ray timing distributions were acquired.

Experimental setup:

A homogeneous cylindrical phantom comprised of poly(methylmethacrylate) was used. Air cavities of varying thickness ∆R ∈ {0 mm, 5 mm, 10 mm, 20 mm} were successively introduced into the phantom to mimic anatomical variations leading to range deviations. For each air cavity thickness, the phantom was irradiated with proton pencil beams of two different kinetic energies (E_1 = 162 MeV and E_2 = 227 MeV) and a micropulse repetition rate of 106.3 MHz. Prompt-gamma ray timing distributions were measured with a detection unit consisting of a single ∅2 ” × 2 ” CeBr_3 crystal by Scionix, a Hamamatsu R13089-100 photomultiplier and a U100 digital spectrometer by Target Systemelektronik, which was placed at a backward angle of 130° . A static pencil beam was directed centrally at the phantom. The beam was pulsed in spots with a spot duration of 69 ms, a period of 72 ms and 1e9 (!) protons per spot (corresponding approximately to the combined signal of 8 prompt-gamma ray detection units for one strongly weighted clinical pencil beam scanning spot). One measurement consisted of 100 spots. Overall, the experiment comprised eight measurements covering the set of four cavity thicknesses ∆R and two beam energies E_1 and E_2. Experiments were carried out in the patient treatment room of OncoRay, Dresden.

Data preprocessing:

The raw data of each measurement was preprocessed as follows: The binary data was converted to ROOT. The photomultiplier gain drift was corrected for and the integral signal charge was converted into deposited energy. Time digitalisation nonlinearities were corrected for. The calibrated data was then saved in list-mode format. The data was assigned to the spot number and the detection time relative to the accelerator radiofrequency (fine time) was used to populate a prompt gamma-ray timing histogram for each spot. No background or phase shift correction were applied.

Data structure:

The dataset contains one root file for each measurement, named by the detector number in the format u100-p00XX and the measurement time. The spreadsheet MeasurementIndex_20160716_SingleSpot.xlsx contains the details of each measurement. The corrected and calibrated PGT spectra can be found in the root file at analysis/05_PGT_for_Layers_and_Spots.

Each root file contains the following directories:

analysis



    01_Layers_and_Spots_Detection: association between spot number and measurement time


    02_Gain_Correction: energy gain drift correction curve


    03_Energy_Calibration: energy calibration curve


    04_Fine_Time_Linearization: timing non-linearity calibration curve


    05_PGT_for_Layers_and_Spots: final PGT spectra - for each spot of each layer:



        PGT_*_all: timing spectrum of the whole energy range


        PGT_*_2,5to7MeV:  timing spectrum for events between 2.5 and 7 MeV only


        PGT_*_3to5MeV: timing spectrum for events between 3 and 5 MeV only


        ESpec: energy spectrum


        EoT: two-dimensional energy-timing spectrum






data: list-mode data (not histogrammed)



    uncorrected: before the correction and calibration steps


    corrected: after the correction and calibration steps




meta: measurement meta data (log file containing applied detector HV etc.)


histograms: selected example histograms

For further questions, please refer to the contact persons stated in the Contributors section.