Data publication: Performance of the Prompt Gamma-ray Timing system prototype under clinical-like conditions

The dataset contains the data used for evaluating the performance of the Prompt Gamma-ray Timing (PGT) system under clinical-like conditions.

<strong><em>Experimental setup</em>:</strong> Clinically realistic dose plans were applied to an anthropomorphic head phantom at the pencil beam scanning (PBS) beamline. Two phantom positioning schemes were employed:

nose<em>φ </em>setup: the geometric center of the head phantom was aligned with the beamline isocenter, and the phantom’s nose pointed in a given direction defined by an angle <em>φ</em> (in the bird’s-eye view)
gantry-like G<em>θ</em> setup: the phantom was placed according to a positioning template so that a hypothetical tumor, contoured on the phantom’s CT images, was aligned with a beamline isocenter, and the PBS nozzle position relative to the phantom corresponded then to a gantry rotational angle <em>θ.</em>

The photographs of the experimental setup and the schematic of the target positioning are provided in Figure 1 of the <strong>0_Materials_and_Methods.zip</strong> file. The positioning template for the G<em>θ</em> setups is given in Figure 2 in <strong>0_Materials_and_Methods.zip</strong>.

Three types of irradiation fields were used for the study: 

EqualMU fields: square fields of about 8.4 cm × 8.4 cm, comprising 15×15 spots arranged on a regular grid with a lateral spacing of 6 mm. Spots within the same energy layer share an identical weight. 
DistalLayer fields: fields comprising 5+15×15+5 spots arranged on a regular grid with a lateral spacing of 6 mm. The main sequence of spots (15×15) forms a square field of about 8.4 cm × 8.4 cm and has varying spot weights. The additional 10 outermost lateral spots (5 before and after the main sequence) are used to determine the field orientation.
G<em>θ</em> fields: these are treatment fields developed to target a hypothetical tumor delineated in the phantom’s CT images. They define complex field shapes consisting of multiple energy layers and spots with widely varying weights.

The employed irradiation fields are provided as *.pld files in <strong>0_Materials_and_Methods.zip</strong>.

For several measurements, a beam range shifter with a water-equivalent thickness of 7.38 cm was inserted into the beamline. It was rigidly attached to the snout holding the detection units, ensuring a fixed position throughout the measurements.

Produced gamma rays were measured with eight scintillation detectors placed at: 

0° (detector p0012);            180° (detector p0008);

45° (detector p0017);          225° (detector p0006);

90° (detector p0015);         270° (detector p0013);

135° (detector p0009);       315° (detector p0019).

<strong><em>Measurements</em>:</strong> the four experimental studies were conducted, and the data from these studies are given in the corresponding zip archives:

Evaluate the <strong>count-rate capacity</strong> of the PGT system: the phantom was in the G270 orientation; an EqualMU plan comprising 9 energy layers (combinations of {150, 120, 90}MeV and {0.01, 0.1, 1}MU was used. Due to the limitations on the minimal spot weight imposed by the beam delivery system, the 0.01 MU spots actually weighed 0.0101 MU. Data only for the 7 detectors employed in this experiment are provided in <strong>1_Count_rate_capacity.zip</strong>. Experimental and clinical machine log files are not given (due to internal regulations).
Investigate the <strong>range shifter contribution</strong> to the PGT data: The data are provided only for the detector p0006 (at 225°) placed inside a hollow cylindrical lead collimator (<em>r₁</em>=2'', <em>r₂</em>=2''+1 cm). The range shifter was inserted in the beamline; the phantom was in nose45 orientation; two DistalLayer plans with 104 MeV and 187 MeV energy layers were applied. After passing the range shifter, these correspond to proton energies of 30 MeV and 150 MeV, respectively. Each plan comprised 24 identical energy layers and delivered a total of 1009 MU. Data from these measurements are provided in <strong>2_Range_shifter_contribution.zip</strong>.
Study <strong>spot-position dependence in scanned fields</strong>: the phantom was positioned as nose0; the range shifter was removed from the beamline to ensure only a single (target-related) peak in time distributions; EqualMU fields of {90, 120, 150} MeV and with spots of 1 MU weight were applied, each field comprised 8 identical layers and was delivered 2 times. Note that during the second repetition of the 120 MeV field, the file for p0012 was corrupted; therefore, the field was applied for the third time, and for this repetition, the file for p0015 was corrupted. Therefore, there are 3 data files for all detectors except for p0012 and p0015. Data files are in <strong>3_Spot_position_dependence_in_scanned_fields.zip</strong>.
Investigate the <strong>stability of the PGT mean</strong> with irradiation time: phantom was in the nose0 orientation; the range shifter was removed from the beamline; EqualMU fields with energy layers of {90, 120, 150}MeV and spot weights of either 0.2 MU or 1 MU were delivered. Fields with 0.2 MU spots included 40 identical energy layers, while those with 1 MU spots included 8 layers. Each field was delivered twice, in a random order. Since studies 3 and 4 overlap (they comprise the same measurements with {90, 120, 150} MeV and 1 MU fields), only the data from {90, 120, 150} MeV and 0.2 MU fields are included in <strong>4_Stability_of_PGT_mean.zip</strong>. The remaining files for {90, 120, 150} MeV and 1 MU fields have already been given in 3_Spot_position_dependence_in_scanned_fields.zip.

<strong>Data preprocessing: </strong>The raw data of each measurement were converted from the binary list-mode format to ROOT TTrees. The data were corrected for the photomultiplier gain drift and digitalization time non-linearities. The integral signal was converted into deposited energy. The data were assigned to individual corresponding spots.

<strong>Data structure: </strong>The ROOT files are named u100-p00<strong>XX</strong>-yyyy-mm-dd_HH.MM.SS+TZ.root, where p00<strong>XX</strong> is the detector’s number, yyyy-mm-dd_HH.MM.SS is the time of the measurement, and TZ is the time zone.

In general, the data structure inside the ROOT files includes:

<strong>data</strong> (TTree) contains list-mode data, which comprises

uncorrected (original measured) data. It contains branches:

<strong><em>Triggertime </em></strong>(in time stamps, when the event triggered the data acquisition)
<strong><em>Livetime </em></strong>(in time stamps, when the detector was idle)
<strong><em>Energy </em></strong>(in a.u., normalized integral over the pulse)
<strong><em>HeadEnergy</em></strong> (in a.u.)

corrected and calibrated data. It comprises branches:

<strong><em>EnergyGainCorrected </em></strong>(in a.u., pulse integral after applying correction for a photomultiplier gain drift).
<strong><em>EnergyCalibrated</em></strong> (in MeV, calibrated pulse integral).
<strong><em>FineTimeCorrected </em></strong>(in ns, detection time within the cyclotron acceleration period after correcting for time non-linearities).
<strong><em>GlobalSpotID</em></strong> (in a.u., assigns a global ID to a spot, which incrementally increases for each new spot. If there is no beam, the counter is 0).
<strong><em>LayerID</em></strong> (in a.u., an ID of the current energy layer. Outside the layer (no beam), the counter is 0).
<strong><em>LocalSpotID</em></strong> (in a.u., a spot ID within the current layer. Outside the spot (no beam), the counter is 0).
<strong><em>SpotMU</em></strong> (in MU, a spot weight of the current spot extracted from machine log files. If there was no spot irradiated, this value is 0).
<strong><em>SpotEnergy </em></strong>(in MeV, the energy of the current energy layer taken from machine log files. Outside energy layers, this value is 0).
<strong><em>SpotXCoordinate</em></strong>,<strong><em> SpotYCoordinate</em> </strong>(in mm, the measured X- and Y-coordinates of the current spot. Outside the spot (no beam), these values are 10000).

<strong>meta </strong>(TTree) is measurement metadata (applied detector voltage, the start time of the measurements, etc.);
<strong>histograms </strong>is a directory with selected example histograms (uncorrected);
<strong>analysis </strong>is a directory with histograms to correct and calibrate data, which are later saved into the <strong>data</strong> TTree. The main subdirectories here are:

<strong><em>00_General_Information</em></strong> contains data from machine log files: how many energy layers were irradiated, of which energies, how many spots each layer comprised, etc.
<strong><em>01_Layers_and_Spots_Detection</em></strong> contains histograms with the start and stop time of every energy layer and spot.
<strong><em>02_Gain_Correction</em></strong> includes histograms used to correct for photomultiplier gain drift. The procedure is described in Werner <em>et al</em>. (2019) in Phys. Med. Biol. 64 105023, 20pp (https://doi.org/10.1088/1361-6560/ab176d).
<strong><em>03_Energy_Calibration</em></strong> contains data of the performed energy calibration of the detector. 
<strong><em>04_Fine_Time_Linearization</em></strong> comprises histograms used to correct for differential and integral time non-linearities. The procedure is described in Werner <em>et al</em>. (2019) in Phys. Med. Biol. 64 105023, 20pp (https://doi.org/10.1088/1361-6560/ab176d).

Further, the authors typically employed an energy selection window of 0.7-7.40 MeV and subtracted time-uncorrelated background using the closest neighbor algorithm, as described in the dedicated publication.

For further questions, please contact the persons stated above.