Interobserver delineation variability of computed tomography-based radiomic features of the parotid gland

PURPOSE

To assess the interobserver delineation variability of radiomic features of the parotid gland from computed tomography (CT) images and evaluate the correlation of these features for head and neck cancer (HNC) radiotherapy patients.

MATERIALS AND METHODS

Contrast-enhanced CT images of 20 HNC patients were utilized. The parotid glands were delineated by treating radiation oncologists (ROs), a selected RO and AccuContour auto-segmentation software. Dice similarity coefficients (DSCs) between each pair of observers were calculated. A total of 107 radiomic features were extracted, whose robustness to interobserver delineation was assessed using the intraclass correlation coefficient (ICC). Pearson correlation coefficients (r) were calculated to determine the relationship between the features. The influence of excluding unrobust features from normal tissue complication probability (NTCP) modeling was investigated for severe oral mucositis (grade ≥3).

RESULTS

The average DSC was 0.84 (95% confidence interval, 0.83-0.86). Most of the shape features demonstrated robustness (ICC ≥0.75), while the first-order and texture features were influenced by delineation variability. Among the three observers investigated, 42 features were sufficiently robust, out of which 36 features exhibited weak correlation (|r|<0.8). No significant difference in the robustness level was found when comparing manual segmentation by a single RO or automated segmentation with the actual clinical contour data made by treating ROs. Excluding unrobust features from the NTCP model for severe oral mucositis did not deteriorate the model performance.

CONCLUSION

Interobserver delineation variability had substantial impact on radiomic features of the parotid gland. Both manual and automated segmentation methods contributed similarly to this variation.

CBCT-to-CT Translation Using Registration-Based Generative Adversarial Networks in Patients with Head and Neck Cancer

Recently, deep learning with generative adversarial networks (GANs) has been applied in multi-domain image-to-image translation. This study aims to improve the image quality of cone-beam computed tomography (CBCT) by generating synthetic CT (sCT) that maintains the patient’s anatomy as in CBCT, while having the image quality of CT. As CBCT and CT are acquired at different time points, it is challenging to obtain paired images with aligned anatomy for supervised training. To address this limitation, the study incorporated a registration network (RegNet) into GAN during training. RegNet can dynamically estimate the correct labels, allowing supervised learning with noisy labels. The study developed and evaluated the approach using imaging data from 146 patients with head and neck cancer. The results showed that GAN trained with RegNet performed better than those trained without RegNet. Specifically, in the UNIT model trained with RegNet, the mean absolute error (MAE) was reduced from 40.46 to 37.21, the root mean-square error (RMSE) was reduced from 119.45 to 108.86, the peak signal-to-noise ratio (PSNR) was increased from 28.67 to 29.55, and the structural similarity index (SSIM) was increased from 0.8630 to 0.8791. The sCT generated from the model had fewer artifacts and retained the anatomical information as in CBCT.

Correction: Suwanbut et al. Assessment of Fetal Dose and Health Effect to the Fetus from Breast Cancer Radiotherapy during Pregnancy. Life 2022, 12, 84

In the original publication [...].

Track-structure modes in particle and heavy ion transport code system (PHITS): application to radiobiological research

PURPOSE

In radiation physics, Monte Carlo radiation transport simulations are powerful tools to evaluate the cellular responses after irradiation. When investigating such radiation-induced biological effects, it is essential to perform track structure simulations by explicitly considering each atomic interaction in liquid water at the sub-cellular and DNA scales. The Particle and Heavy-Ion Transport code System (PHITS) is a Monte Carlo code which enables to calculate track structure at DNA scale by employing the track-structure modes for electrons, protons and carbon ions. In this paper, we review the recent development status and future prospects of the track-structure modes in the PHITS code.

CONCLUSIONS

To date, the physical features of these modes have been verified using the available experimental data and Monte Carlo simulation results reported in literature. These track-structure modes can be used for calculating microdosimetric distributions to estimate cell survival and for estimating initial DNA damage yields. The use of PHITS track-structure mode is expected not only to clarify the underlying mechanisms of radiation effects but also to predict curative effects in radiation therapy. The results of PHITS simulations coupled with biophysical models will contribute to the radiobiological studies by precisely predicting radiation-induced biological effects based on the Monte Carlo approach.

Verification of KURBUC-based ion track structure mode for proton and carbon ions in the PHITS code

The particle and heavy ion transport code system (PHITS) is a general-purpose Monte Carlo radiation transport simulation code. It has the ability to handle diverse particle types over a wide range of energy. The latest PHITS development enables the generation of track structure for proton and carbon ions (¹H⁺, ¹²C⁶⁺) based on the algorithms in the KURBUC code, which is considered as one of the most verified track-structure codes worldwide. This ion track-structure mode is referred to as the PHITS-KURBUC mode. In this study, the range, radial dose distributions, and microdosimetric distributions were calculated using the PHITS-KURBUC mode. Subsequently, they were compared with the corresponding data obtained from the original KURBUC and from other studies. These comparative studies confirm the successful inclusion of the KURBUC code in the PHITS code. As results of the synergistic effect between the macroscopic and microscopic radiation transport codes, this implementation enabled the detailed calculation of the microdosimetric and nanodosimetric quantities under complex radiation fields, such as proton beam therapy with the spread-out Bragg peak.

CHARACTERIZATION OF AN IN-HOUSE DEVELOPED MULTI-CYLINDRICAL MODERATOR NEUTRON SPECTROMETER

This article describes the characterization of an in-house developed multi-cylindrical moderator neutron spectrometer, which consists of a cylindrical 3He proportional counter and cylindrical moderator shells of different sizes. The response matrix of the spectrometer was calculated by Monte Carlo simulations for neutron energies from 1 × 10-8 to 10 MeV and verified with measurements in 0.144 MeV, 1.2 MeV and 241AmBe neutron standard fields. Influence of scattered neutrons was properly eliminated from the measured response using the shadow cone technique. The calculated and measured responses were in good agreement in most cases. Differences were <10% for all moderated counter configurations, while larger deviations were observed for the bare counter configuration. The performance of the neutron spectrometer in terms of spectrum unfolding was verified in the 241AmBe neutron standard field, showing reliable neutron spectrum and fluence rate in the energy range up to 10 MeV as investigated in this work.

Radiation track, DNA damage and response-a review

The purpose of this paper has been to review the current status and progress of the field of radiation biophysics, and draw attention to the fact that physics, in general, and radiation physics in particular, with the aid of mathematical modeling, can help elucidate biological mechanisms and cancer therapies. We hypothesize that concepts of condensed-matter physics along with the new genomic knowledge and technologies and mechanistic mathematical modeling in conjunction with advances in experimental DNA (Deoxyrinonucleic acid molecule) repair and cell signaling have now provided us with unprecedented opportunities in radiation biophysics to address problems in targeted cancer therapy, and genetic risk estimation in humans. Obviously, one is not dealing with 'low-hanging fruit', but it will be a major scientific achievement if it becomes possible to state, in another decade or so, that we can link mechanistically the stages between the initial radiation-induced DNA damage; in particular, at doses of radiation less than 2 Gy and with structural changes in genomic DNA as a precursor to cell inactivation and/or mutations leading to genetic diseases. The paper presents recent development in the physics of radiation track structure contained in the computer code system KURBUC, in particular for low-energy electrons in the condensed phase of water for which we provide a comprehensive discussion of the dielectric response function approach. The state-of-the-art in the simulation of proton and carbon ion tracks in the Bragg peak region is also presented. The paper presents a critical discussion of the models used for elastic scattering, and the validity of the trajectory approach in low-electron transport. Brief discussions of mechanistic and quantitative aspects of microdosimetry, DNA damage and DNA repair are also included as developed by the authors' work.

Microdosimetry of the full slowing down of protons using Monte Carlo track structure simulations

The article investigates two approaches in microdosimetric calculations based on Monte Carlo track structure (MCTS) simulations of a 160-MeV proton beam. In the first approach, microdosimetric parameters of the proton beam were obtained using the weighted sum of proton energy distributions and microdosimetric parameters of proton track segments (TSMs). In the second approach, phase spaces of energy depositions obtained using MCTS simulations in the full slowing down (FSD) mode were used for the microdosimetric calculations. Targets of interest were water cylinders of 2.3-100 nm in diameters and heights. Frequency-averaged lineal energies ([Formula: see text]) obtained using both approaches agreed within the statistical uncertainties. Discrepancies beyond this level were observed for dose-averaged lineal energies ([Formula: see text]) towards the Bragg peak region due to the small number of proton energies used in the TSM approach and different energy deposition patterns in the TSM and FSD of protons.

Microdosimetry of low-energy electrons

PURPOSE

To investigate differences in energy depositions and microdosimetric parameters of low-energy electrons in liquid and gaseous water using Monte Carlo track structure simulations.

MATERIALS AND METHODS

KURBUC-liq (Kyushu University and Radiobiology Unit Code for liquid water) was used for simulating electron tracks in liquid water. The inelastic scattering cross sections of liquid water were obtained from the dielectric response model of Emfietzoglou et al. (Radiation Research 2005;164:202-211). Frequencies of energy deposited in nanometre-size cylindrical targets per unit absorbed dose and associated lineal energies were calculated for 100-5000 eV monoenergetic electrons and the electron spectrum of carbon K edge X-rays. The results for liquid water were compared with those for water vapour.

RESULTS

Regardless of electron energy, there is a limit how much energy electron tracks can deposit in a target. Phase effects on the frequencies of energy depositions are largely visible for the targets with diameters and heights smaller than 30 nm. For the target of 2.3 nm by 2.3 nm (similar to dimension of DNA segments), the calculated frequency- and dose-mean lineal energies for liquid water are up to 40% smaller than those for water vapour. The corresponding difference is less than 12% for the targets with diameters ≥ 30 nm.

CONCLUSIONS

Condensed-phase effects are non-negligible for microdosimetry of low-energy electrons for targets with sizes smaller than a few tens of nanometres, similar to dimensions of DNA molecular structures and nucleosomes.

A model of carbon ion interactions in water using the classical trajectory Monte Carlo method

In this paper, model calculations for interactions of C(6+) of energies from 1 keV u(-1) to 1 MeV u(-1) in water are presented. The calculations were carried out using the classical trajectory Monte Carlo method, taking into account the dynamic screening of the target core. The total cross sections (TCS) for electron capture and ionisation, and the singly and doubly differential cross sections (SDCS and DDCS) for ionisation were calculated for the five potential energy levels of the water molecule. The peaks in the DDCS for the electron capture to continuum and for the binary-encounter collision were obtained for 500-keV u(-1) carbon ions. The calculated SDCS agree reasonably well with the z(2) scaled proton data for 500 keV u(-1) and 1 MeV u(-1) projectiles, but a large deviation of up to 8-folds was observed for 100-keV u(-1) projectiles. The TCS for ionisation are in agreement with the values calculated from the first born approximation (FBA) at the highest energy region investigated, but become smaller than the values from the FBA at the lower-energy region.

EDDIX--a database of ionisation double differential cross sections

The use of Monte Carlo track structure is a choice method in biophysical modelling and calculations. To precisely model 3D and 4D tracks, the cross section for the ionisation by an incoming ion, double differential in the outgoing electron energy and angle, is required. However, the double differential cross section cannot be theoretically modelled over the full range of parameters. To address this issue, a database of all available experimental data has been constructed. Currently, the database of Experimental Double Differential Ionisation Cross sections (EDDIX) contains over 1200 digitalised experimentally measured datasets from the 1960s to present date, covering all available ion species (hydrogen to uranium) and all available target species. Double differential cross sections are also presented with the aid of an eight parameter functions fitted to the cross sections. The parameters include projectile species and charge, target nuclear charge and atomic mass, projectile atomic mass and energy, electron energy and deflection angle. It is planned to freely distribute EDDIX and make it available to the radiation research community for use in the analytical and numerical modelling of track structure.