An artificial neural network based approach for predicting the proton beam spot dosimetric characteristics of a pencil beam scanning technique.

Utilising Machine Learning (ML) models to predict dosimetric parameters in pencil beam scanning proton therapy presents a promising and practical approach. The study developed Artificial Neural Network (ANN) models to predict proton beam spot size and relative positional errors using 9000 proton spot data. The irradiation log files as input variables and corresponding scintillation detector measurements as the label values. The ANN models were developed to predict six variables: spot size in thex-axis,y-axis, major axis, minor axis, and relative positional errors in thex-axis andy-axis. All ANN models used a Multi-layer perception (MLP) network using one input layer, three hidden layers, and one output layer. Model performance was validated using various statistical tools. The log file recorded spot size and relative positional errors, which were compared with scintillator-measured data. The Root Mean Squared Error (RMSE) values for the x-spot and y-spot sizes were 0.356 mm and 0.362 mm, respectively. Additionally, the maximum variation for the x-spot relative positional error was 0.910 mm, while for the y-spot, it was 1.610 mm. The ANN models exhibit lower prediction errors. Specifically, the RMSE values for spot size prediction in the x, y, major, and minor axes are 0.053 mm, 0.049 mm, 0.053 mm, and 0.052 mm, respectively. Additionally, the relative spot positional error prediction model for the x and y axes yielded maximum errors of 0.160 mm and 0.170 mm, respectively. The normality of models was validated using the residual histogram and Q-Q plot. The data over fit, and bias were tested using K (k = 5) fold cross-validation, and the maximum RMSE value of the K fold cross-validation among all the six ML models was less than 0.150 mm (R-Square 0.960). All the models showed excellent prediction accuracy. Accurately predicting beam spot size and positional errors enhances efficiency in routine dosimetric checks.

Mean parotid dose prediction model using machine learning regression method for intensity-modulated radiotherapy in head and neck cancer.

Parotids are considered one of the major organs at risk in Head and Neck (HN) intensity-modulated radiotherapy (IMRT). Achieving proper target coverage with reduced mean parotid dose demands an elaborate time-consuming IMRT plan optimization. A parotid mean dose prediction model based on a machine-learning linear regression was developed and validated in this study. The model was developed using independent variables, such as parotid to PTV overlapping volume, dose coverage of the overlapping PTV, the ratio of overlapping parotid volume to total parotid volume, and volume of parotid overlapping with isotopically expanded PTV contours. The Pearson correlation coefficients between these independent variables and the mean parotid dose were calculated. Multicollinearity of the independent variables was checked by calculating the Variance Inflation Factor (VIF). All variables are having VIF less than ten were taken for the model. Fifty IMRT patient plans were used to develop the model. The mean parotid dose predicted by the model was in good agreement with the obtained mean parotid dose. The model is having a Root Mean Square Error (RMSE) of 2.89 Gy and an R-square of 0.7695. The model was successfully validated using the fivefold cross-validation method, resulting R-square value of 0.6179 and an RMSE of 2.93 Gy. The normality of the model's residuals was tested using Quartile-Quartile (Q-Q) plot and Shapiro Wilk test (p = 0.996, for null hypothesis ``residuals were normally distributed''). The data points in the Q-Q plot are falling approximately along the reference line. This model can be used in clinics to help the planner in the preplanning phase for efficient plan optimization.

A technique for quantifying the sensitivity of dosimetric tool gamma with 2D detector array in pretreatment IMRT plans by segment deletion method.

OBJECTIVES: Motivation of this study is to check the sensitivity of dosimetric tool gamma with 2D detector array combination when unexpected errors occur while transferring intensity-modulated radiation therapy treatment plans from planning system to treatment unit. METHODS: This study consists of 17 head and neck cancer patient's treatment plans. Nine types of verification plans are created for all 17 clinically approved treatment plans by consecutively deleting different segments (up to eight) one by one from each field of the plan. Decrement factor (χ) is introduced in our study which illustrated the degree of decay of gamma passing rate when intentional errors are introduced. We analyzed the data by two different methods-one without selecting the region of interest (ROI) in dose distributions and the other by selecting the region of interest. RESULTS: By linear regression, the absolute value of slopes is 0.025, 0.024 and 0.015 without ROI and 0.030, 0.027 and 0.015 with ROI for 2%/2 mm, 3%/3 mm and 5%/5 mm criteria, respectively. The higher absolute value of the fitted slope indicates the higher sensitivity of this method to identify erroneous plan in treatment unit. The threshold value for 2%/2 mm equivalent to 95% passing criteria in 3%/3 mm used in clinical practice is obtained as 83.44%. CONCLUSIONS: The 2D detector array with dosimetric tool gamma is less sensitive in detecting errors when unprecedented errors of segment deletion occur within the treatment plans.

Experimental determination of thermal neutron fluence around Elekta Versa HD linear accelerator for various photon energies.

A complex neutron spectrum generated along with a useful photon beam imposes an additional radiation protection risk around medical linear accelerators (linac). The thermal neutron component of this complex neutron spectrum formed during different photon modes of operation of Elekta Versa HD linac has been quantified using Indium foil activation technique. The thermal neutron fluence (Φ th ) at isocenter for 15 MV, 10 MV and 10 MV FFF beams was found to be 2.45 × 105, 4.35 × 104 and 3.2 × 104 neutrons cm-2 Gy-1, respectively. The analysis shows a reduction in the Φ th as the flattening filter is being taken out from the beam path. A negative correlation in Φ th with respect to field size has been observed with an average 18% reduction in Φ th per monitor units as field size changes from 10 cm × 10 cm to 40 cm × 40 cm. For particular field size and photon energy, Φ th was found to be uniform across the patient plane. From the measured gamma ray spectrum inside the treatment room six major isotopes have been identified which were 122Sb, 187W, 82Br, 56Mn, 24Na and 28Al.

Dosimetric properties of equivalent-quality flattening filter-free (FFF) and flattened photon beams of Versa HD linear accelerator.

This study presents the basic dosimetric properties of photon beams of a Versa HD linear accelerator (linac), which is capable of delivering flattening filter-free (FFF) beams with a beam quality equivalent to the corresponding flattened beams based on comprehensive beam data measurement. The analyzed data included the PDDs, profiles, penumbra, out-of-field doses, surface doses, output factors, head and phantom scatter factors, and MLC transmissions for both FFF and flattened beams of 6 MV and 10 MV energy from an Elekta Versa HD linac. The 6MVFFF and 10MVFFF beams had an equivalent mean energy to the flattened beams and showed less PDD variations with the field sizes. Compared with their corresponding flattened beams, Dmax was deeper for FFF beams for all field sizes; the ionization ratio variations with the field size were lower for FFF beams; the out-of-field doses were lower and the penumbras were sharper for the FFF beams; the off-axis profile variations with the depths were lesser for the FFF beams. Further, the 6MVFFF and 10MVFFF beams had 35.7% and 40.9% less variations in output factor with the field size, respectively. The collimator exchange effect was reduced in the FFF mode. The head scatter factor showed 59.1% and 73.6% less variations, on average, for the 6MVFFF and 10MVFFF beams, respectively; the variations in the phantom scatter factor were also smaller. The surface doses for all beams increased linearly with the field size. The 6MVFFF and 10MVFFF beams had higher surface doses than the corresponding flattened beams for field sizes of up to 10 ×10cm2 but had lower surface doses for larger fields. Both FFF beams had lower average MLC transmissions than the flattened beams. The finding that the FFF beams were of equivalent quality to the corresponding flattened beams indicates a significant dif-ference from the data on unmatched FFF beams.