Strategies to develop radiomics and machine learning models for lung cancer stage and histology prediction using small data samples.

Predictive models based on radiomics and machine-learning (ML) need large and annotated datasets for training, often difficult to collect. We designed an operative pipeline for model training to exploit data already available to the scientific community. The aim of this work was to explore the capability of radiomic features in predicting tumor histology and stage in patients with non-small cell lung cancer (NSCLC). We analyzed the radiotherapy planning thoracic CT scans of a proprietary sample of 47 subjects (L-RT) and integrated this dataset with a publicly available set of 130 patients from the MAASTRO NSCLC collection (Lung1). We implemented intra- and inter-sample cross-validation strategies (CV) for evaluating the ML predictive model performances with not so large datasets. We carried out two classification tasks: histology classification (3 classes) and overall stage classification (two classes: stage I and II). In the first task, the best performance was obtained by a Random Forest classifier, once the analysis has been restricted to stage I and II tumors of the Lung1 and L-RT merged dataset (AUC = 0.72 ± 0.11). For the overall stage classification, the best results were obtained when training on Lung1 and testing of L-RT dataset (AUC = 0.72 ± 0.04 for Random Forest and AUC = 0.84 ± 0.03 for linear-kernel Support Vector Machine). According to the classification task to be accomplished and to the heterogeneity of the available dataset(s), different CV strategies have to be explored and compared to make a robust assessment of the potential of a predictive model based on radiomics and ML.

Characterization of PTW-31015 PinPoint ionization chambers in photon and proton beams.

The increased use of complex forms of radiotherapy using small-field photon and proton beams has invoked a growing interest in the use of micro-ionization chambers. In this study, 48 PTW-TM31015 PinPoint-type micro-ionization chambers that are used in the commissioning and patient specific QA of a proton pencil beam scanning (PBS) delivery system have been characterized in proton and high-energy photon beams. In both beam modalities, the entire set of PinPoint chambers was characterized by imaging them, by evaluating their stability using check source measurements, by experimentally determining the ion recombination, polarity effect and by cross calibrating them in terms of absorbed dose to water against Farmer-type ionization chambers. Beam quality correction factors were theoretically derived for both beam modalities. None of the chambers' check source readings drifted by more than 0.5% over a one year period. Beam quality correction factors for the 6 MV photon with reference to 60Co were on average 1.0 ± 0.5% lower than the theoretical values calculated according to the data and procedures outlined in IAEA TRS-398. While this difference is within the overall dosimetric uncertainty, it is significant considering only uncorrelated uncertainties indicating inconsistencies in the theoretical data. Beam quality correction factors for the 179.2 MeV proton beam with reference to 60Co were in good agreement with the theoretical data. Ion recombination and polarity correction factors were very consistent for all chambers with standard deviations of 0.2% or below show that findings from more comprehensive investigations in the literature can be considered as representative for all the chambers of this type. The characterization of 48 PinPoint-type micro-ionization chambers performed in this study provided a unique opportunity to investigate chamber-to-chamber variations of calibration, beam quality correction factors, ion recombination and polarity correction factors for an unprecedented sample size of chambers for both high-energy photon and proton beams.

End-to-end tests using alanine dosimetry in scanned proton beams.

This paper describes end-to-end test procedures as the last fundamental step of medical commissioning before starting clinical operation of the MedAustron synchrotron-based pencil beam scanning (PBS) therapy facility with protons. One in-house homogeneous phantom and two anthropomorphic heterogeneous (head and pelvis) phantoms were used for end-to-end tests at MedAustron. The phantoms were equipped with alanine detectors, radiochromic films and ionization chambers. The correction for the 'quenching' effect of alanine pellets was implemented in the Monte Carlo platform of the evaluation version of RayStation TPS. During the end-to-end tests, the phantoms were transferred through the workflow like real patients to simulate the entire clinical workflow: immobilization, imaging, treatment planning and dose delivery. Different clinical scenarios of increasing complexity were simulated: delivery of a single beam, two oblique beams without and with range shifter. In addition to the dose comparison in the plastic phantoms the dose obtained from alanine pellet readings was compared with the dose determined with the Farmer ionization chamber in water. A consistent systematic deviation of about 2% was found between alanine dosimetry and the ionization chamber dosimetry in water and plastic materials. Acceptable agreement of planned and delivered doses was observed together with consistent and reproducible results of the end-to-end testing performed with different dosimetric techniques (alanine detectors, ionization chambers and EBT3 radiochromic films). The results confirmed the adequate implementation and integration of the new PBS technology at MedAustron. This work demonstrates that alanine pellets are suitable detectors for end-to-end tests in proton beam therapy and the developed procedures with customized anthropomorphic phantoms can be used to support implementation of PBS technology in clinical practice.

Physics, Techniques and Review of Neuroradiological Applications of Diffusion Kurtosis Imaging (DKI).

In recent years many papers about diagnostic applications of diffusion tensor imaging (DTI) have been published. This is because DTI allows to evaluate in vivo and in a non-invasive way the process of diffusion of water molecules in biological tissues. However, the simplified description of the diffusion process assumed in DTI does not permit to completely map the complex underlying cellular components and structures, which hinder and restrict the diffusion of water molecules. These limitations can be partially overcome by means of diffusion kurtosis imaging (DKI). The aim of this paper is the description of the theory of DKI, a new topic of growing interest in radiology. DKI is a higher order diffusion model that is a straightforward extension of the DTI model. Here, we analyze the physics underlying this method, we report our MRI acquisition protocol with the preprocessing pipeline used and the DKI parametric maps obtained on a 1.5 T scanner, and we review the most relevant clinical applications of this technique in various neurological diseases.

Comparison of EPR response of alanine and Gd2O3-alanine dosimeters exposed to TRIGA Mainz reactor.

In this work we report some preliminary results regarding the analysis of electron paramagnetic resonance (EPR) response of alanine pellets and alanine pellets added with gadolinium used for dosimetry at the TRIGA research reactor in Mainz, Germany. Two set-ups were evaluated: irradiation inside PMMA phantom and irradiation inside boric acid phantom. We observed that the presence of Gd2O3 inside alanine pellets increases the EPR signal by a factor of 3.45 and 1.24 in case of PMMA and boric acid phantoms, respectively. We can conclude that in the case of neutron beam with a predominant thermal neutron component the addition of gadolinium oxide can significantly improve neutron sensitivity of alanine pellets. Monte Carlo (MC) simulations of both response of alanine and Gd-added alanine pellets with FLUKA code were performed and a good agreement was achieved for pure alanine dosimeters. For Gd2O3-alanine deviations between MC simulations and experimental data were observed and discussed.

Testing and linearity calibration of films of phenol compounds exposed to thermal neutron field for EPR dosimetry.

This paper reports the preliminary results obtained by Electron Paramagnetic Resonance (EPR) measurements on films of IRGANOX® 1076 phenols with and without low content (5% by weight) of gadolinium oxide (Gd2O3) exposed in the thermal column of the Triga Mark II reactor of LENA (Laboratorio Energia Nucleare Applicata) of Pavia (Italy). Thanks to their size, the phenolic films here presented are good devices for the dosimetry of beams with high dose gradient and which require accurate knowledge of the precise dose delivered. The dependence of EPR signal as function of neutron dose was investigated in the fluence range between 10(11) cm(-2) and 10(14) cm(-2). Linearity of EPR response was found and the signal was compared with that of commercial alanine films. Our analysis showed that gadolinium oxide (5% by weight) can enhance the thermal neutron sensitivity more than 18 times. Irradiated dosimetric films of phenolic compound exhibited EPR signal fading of about 4% after 10 days from irradiation.

Neutron ESR dosimetry through ammonium tartrate with low Gd content.

This paper continues analyses on organic compounds for application in neutron dosimetry performed through electron spin resonance (ESR). Here, the authors present the results obtained by ESR measurements of a blend of ammonium tartrate dosemeters and gadolinium oxide (5 % by weight). The choice of low amount of Gd is due to the need of improving neutron sensitivity while not significantly influencing tissue equivalence. A study of the effect of gadolinium presence on tissue equivalence was carried out. The experiments show that the neutron sensitivity is enhanced by more than an order of magnitude even with this small additive content. Monte Carlo simulations on the increment of energy release due to gadolinium presence were carried, and the results were in good agreement with the experimental data.

Radical distributions in ammonium tartrate single crystals exposed to photon and neutron beams.

The radiation therapy carried out by means of heavy charged particles (such as carbon ions) and neutrons is rapidly becoming widespread worldwide. The success of these radiation therapies relies on the high density of energy released by these particles or by secondary particles produced after primary interaction with matter. The biological damages produced by ionising radiations in tissues and cells depend more properly on the energy released per unit pathlength, which is the linear energy transfer and which determines the radiation quality. To improve the therapy effectiveness, it is necessary to grasp the mechanisms of free radical production and distribution after irradiation with these particles when compared with the photon beams. In this work some preliminary results on the analysis of the spatial distributions of the free radicals produced after exposure of ammonium tartrate crystals to various radiation beams ((60)Co gamma photons and thermal neutrons) were reported. Electron spin resonance analyses were performed by the electron spin echo technique, which allows the determination of local spin concentrations and by double electron-electron resonance technique, which is able to measure the spatial distance distribution (range 1.5-8 nm) among pairs of radicals in solids. The results of these analyses are discussed on the basis of the different distributions of free radicals produced by the two different radiation beams used.

EPR/alanine pellets with low Gd content for neutron dosimetry.

This paper reports on results obtained by electron paramagnetic resonance (EPR) measurements and Monte Carlo (MC) simulation on a blend of alanine added with low content of gadolinium oxide (5 % by weight) to improve the sensitivity to thermal neutron without excessively affecting tissue equivalence. The sensitivity is enhanced by this doping procedure of more an order of magnitude. The results are compared with those obtained with the addition of boric acid (50 % by weight) where boron is in its natural isotopic composition in order to produce low-cost EPR dosemeters. The gadolinium addition influences neutron sensitivity more than the boron addition. The presence of additives does not substantially change the fading of the EPR signal induced by neutrons. The MC simulations agree the experimental results in case of gadolinium addition.

Sensitivity of alanine dosimeters with gadolinium exposed to 6 MV photons at clinical doses.

In this study we analyzed the ESR signal of alanine dosimeters with gadolinium exposed to 6 MV linear accelerator photons. We observed that the addition of gadolinium brings about an improvement in the sensitivity to photons because of its high atomic number. The experimental data indicated that the addition of gadolinium increases the sensitivity of the alanine to 6 MV photons. This enhancement was better observed at high gadolinium concentrations for which the tissue equivalence is heavily reduced. However, information about the irradiation setup and of the radiation beam features allows one to correct for this difference. Monte Carlo simulations were carried out to obtain information on the expected effect of the addition of gadolinium on the dose absorbed by the alanine molecules inside the pellets. These results are compared with the experimental values, and the agreement is discussed.

Review of retrospective dosimetry techniques for external ionising radiation exposures.

The current focus on networking and mutual assistance in the management of radiation accidents or incidents has demonstrated the importance of a joined-up approach in physical and biological dosimetry. To this end, the European Radiation Dosimetry Working Group 10 on 'Retrospective Dosimetry' has been set up by individuals from a wide range of disciplines across Europe. Here, established and emerging dosimetry methods are reviewed, which can be used immediately and retrospectively following external ionising radiation exposure. Endpoints and assays include dicentrics, translocations, premature chromosome condensation, micronuclei, somatic mutations, gene expression, electron paramagnetic resonance, thermoluminescence, optically stimulated luminescence, neutron activation, haematology, protein biomarkers and analytical dose reconstruction. Individual characteristics of these techniques, their limitations and potential for further development are reviewed, and their usefulness in specific exposure scenarios is discussed. Whilst no single technique fulfils the criteria of an ideal dosemeter, an integrated approach using multiple techniques tailored to the exposure scenario can cover most requirements.

Monte Carlo simulation of the response of ESR dosimeters added with gadolinium exposed to thermal, epithermal and fast neutrons.

Monte Carlo numerical calculations of the response of alanine and ammonium tartrate ESR (electron spin resonance) dosimeters exposed to neutron fields with different energy spectra are reported. Results have been obtained for various gadolinium concentrations inside the dosimeters. Furthermore, in order to simulate the in-phantom response we have carried out calculations by varying the depth of the dosimeter. We have found that a large enhancement is obtained for thermal neutrons, because of the very high capture cross section of gadolinium to thermal neutrons. A good enhancement was obtained for epithermal neutrons, whereas the sensitivity improvement in the case of fast neutron irradiation is poor. Furthermore, the simulations carried out by varying the depth suggests that an appreciable sensitivity to thermal and epithermal neutrons could be observed for in-phantom measurements in the 2-3 cm depth range. These results can provide useful insight for future experiments with epithermal neutron beams (such as those used in neutron capture therapy) and for future applications in neutron capture therapy dosimetry.

Seasonal variation of air kerma rate in Sicily.

Thermoluminescence dosimetry has been used to measure air kerma in 29 sites in Sicily. Four three month measurement campaigns have been carried out in order to assess seasonal variations. Average annual values between 20 and 90 nGy h(-1), after cosmic background subtraction, are reported. Average annual values are strongly dependent on site lithology, and we find that winter data are generally the highest, while spring and autumn rates are generally the lowest with very similar trends in any site. Summer values generally lay in between. Largest seasonal variations are found in sites along the southern coast of the island, probably because of stronger action of winds affecting radon, along with its progeny, concentrations at ground level.

Improvement of ESR dosimetry for thermal neutron beams through the addition of gadolinium.

In this paper, the addition of gadolinium is proposed as a useful tool to enhance the electron spin resonance (ESR) sensitivity of organic compounds to thermal neutrons. The target of this work is the detection, through the ESR technique, of the thermal neutron fluence in a mixed field of photons and neutrons. Gadolinium was chosen because it has a very high capture cross section to thermal neutrons; its nuclear reaction with thermal neutrons induces complex inner shell transitions that generate, besides other particles, Auger electrons, which in turn release their energy in the neighborhood (only several nanometers) of the place of reaction. Gadolinium was added to two organic molecules: alanine and ammonium tartrate. The main result obtained was a greater neutron sensitivity for dosimeters with gadolinium than for those without gadolinium for both organic molecules used. Since a dosimeter pair is required to discriminate between the two components of a mixed field, we studied the response of each dosimeter pair irradiated in a mixed field. Through a blind test we verified the usefulness of this dosimetric system and we obtained an estimate of the fluence in the mixed field with a relative uncertainty of 3%, when the pair composed of an alanine dosimeter and a dosimeter with alanine and gadolinium is used.

Alanine blends for ESR measurements of thermal neutron fluence in a mixed radiation field.

In this paper, the results of a study on the electron spin resonance (ESR) dosimetry to measure thermal neutron fluence in a mixed radiation field (neutron and photons) are presented. The ESR responses of alanine dosemeters with different additives are compared. In particular, the (10)B-acid boric and the Gd-oxide were chosen to enhance the sensitivity of alanine dosemeters to thermal neutrons. Irradiations were carried out inside the thermal column of the TAPIRO reactor of the ENEA center, Casaccia Rome. The main results are a greater neutron sensitivity and a smaller lowest detectable fluence for the dosemeters with gadolinium than for dosemeters of alanine with (10)B, which is well known to be much more sensitive to thermal neutrons than simple alanine.

Study of the glow curves of TLD exposed to thermal neutrons.

The glow curves of thermoluminescent dosimeters (TLD600, TLD700 and MCP), exposed to a mixed field of thermal neutrons and gamma photons are analysed. The fluence values of thermal neutrons used, comparable with those used in radiotherapy, allow one to define the reliability of the TLDs, in particular the most sensitive MCP, in this radiation field and to get information on the dose absorbed values. The glow curves obtained have been deconvoluted using general order kinetics and the observed differences for the different LET components have been analysed. In particular, the ratio of the n(0) parameter of two different peaks seems to allow to discriminate the different contributions of neutrons and gamma photons in the beam.

ESR response to gamma-rays of alanine pellets containing B(OH)3 or Gd2O3.

ESR response to gamma-irradiation (1-50 Gy) of blends containing alanine and either B(OH)(3) or Gd(2)O(3) is reported. The sensitivity of the alanine--B(OH)(3) blend is comparable to the sensitivity of pure alanine, although its lowest detectable dose, LDD, is smaller ( approximately 1.3 Gy) than that of pure alanine ( approximately 2.9 Gy). Alanine with Gd(2)O(3) is about two times more sensitive than pure alanine, and its LDD is 0.8 Gy. The better sensitivity and LDD are probably due to the high atomic number (Z=64) of gadolinium, which enhances the interaction probability with photons and, consequently, the radical yield. This study suggests that other high-Z atoms may be useful for increasing the sensitivity of the response of alanine to gamma-radiation.