Table of contents

Establishment of organ doses from diagnostic and interventional examinations is a key component to quantifying the radiation risks from medical exposures and for formulating corresponding dose-reduction strategies. Radiation transport models of human anatomy provide a convenient method for simulating radiological examinations. At present, two classes of models exist: stylized mathematical models and tomographic voxel models. In the present study, organ dose comparisons are made for projection radiographs of both a stylized and a tomographic model of the newborn patient. Sixteen separate radiographs were simulated for each model at x-ray technique factors typical of newborn examinations: chest, abdomen, thorax and head views in the AP, PA, left LAT and right LAT projection orientation. For AP and PA radiographs of the torso (chest, abdomen and thorax views), the effective dose assessed for the tomographic model exceeds that for the stylized model with per cent differences ranging from 19% (AP abdominal view) to 43% AP chest view. In contrast, the effective dose for the stylized model exceeds that for the tomographic model for all eight lateral views including those of the head, with per cent differences ranging from 9% (LLAT chest view) to 51% (RLAT thorax view). While organ positioning differences do exist between the models, a major factor contributing to differences in effective dose is the models' exterior trunk shape. In the tomographic model, a more elliptical shape is seen thus providing for less tissue shielding for internal organs in the AP and PA directions, with corresponding increased tissue shielding in the lateral directions. This observation is opposite of that seen in comparisons of stylized and tomographic models of the adult.

Improving system efficiency without jeopardizing spatial resolution is one of the main problems of small animal PET scanners. In pursuit of this goal, the future LSO-APD-PET prototype MADPET-II will combine highly granulated detector modules with a dual layer structure. The individual readout of the LSO crystals allows separately handling multiple signals related to those photons scattering between different crystal units (inter-crystal scatter, ICS). The contribution of ICS events can significantly increase the system efficiency. Such coincidences are not characterized by a unique LOR. However, in order to minimize resolution degradation, it would be desirable to identify the primary path of the ICS events. Since ICS is geometry dependent, this work was aimed at investigating the effects of ICS in the performance of the dual layer prototype. Different recovery algorithms to select the primary crystal were implemented and developed, and applied to Monte Carlo simulated data. Some of these algorithms were based on the properties of Compton kinematics. For a centred point source and a 100 keV lower energy threshold, the absolute system efficiency was found to increase by 35% when including ICS events: from 1.8% without ICS events to 2.8% with ICS. Similarly, for a threshold of 200 keV, the contribution of ICS coincidences still represented ≈20% of the total detected coincidences, leading to an absolute system efficiency of almost 2%. The mispositioning introduced by processing ICS coincidences only led to a moderate broadening of the axial line spread function (LSF), especially at the tails of the profile (FWTM). This effect was also noticeable in the transaxial plane. In presence of scattering media (water-filled cylinder), the resolution degradation was dominated by the contribution of object scatter. The reconstructed images from a simulated homogeneous cylinder filled with activity with a non-active rod at its centre were employed to estimate the impact of ICS on the image quality. In general, the use of ICS coincidences increased the signal-to-noise ratio (SNR) but worsened contrast. The effects of ICS on resolution could be reduced by employing a new identification scheme based on the maximum signal and the Compton kinematics. This method yielded the highest identification rate for the correct photon trajectory, even for a finite energy resolution of 15% (511 keV). This technique also increased the SNR by 17% to 30% and preserved the image contrast. In conclusion, by combining individual crystal readout, a low energy threshold and an appropriate recovery scheme, the processing of ICS coincidences significantly increases the system efficiency without any substantial deterioration of the image quality.

Tumour blood flow is one of the important factors limiting the efficacy of radiation therapy (hypoxic radioresistance), chemotherapy (drug delivery) and thermal therapy (heat dissipation) in treating cancer. The modification of tumour blood flow has been an area of intense investigation. In the current study, the arterial carbon dioxide tension (P_aCO₂) was changed in order to investigate the tumour vascular response to carbon dioxide. Functional maps of blood flow, blood volume and mean transit time were generated at four P_aCO₂ levels in VX2 tumour in the rabbit thigh and normal soft tissue. The P_aCO₂ levels investigated were normocapnia (P_aCO₂ = 40.9 ± 1.2 mmHg), hypocapnia (27.2 ± 2.3 and 33.5 ± 2.3 mmHg) and hypercapnia (54.9 ± 4.4 mmHg). The carbon dioxide reactivity of the global tumour blood flow and mean transit time showed significant differences between normocapnia and the two levels of hypocapnia, but not between normocapnia and hypercapnia. The average fractional change of blood flow from normocapnia for the two levels of hypocapnia was −0.41 ± 0.06 and −0.29 ± 0.08, respectively (P < 0.05). In the case of mean transit time the fractional change was +0.39 ± 0.30 and +0.23 ± 0.24, respectively (P < 0.05). The fractional change of blood volume from normocapnia, however, was not significantly different at any capnic level, as was the case with respect to each of the functional parameters in normal tissue. The ability to reduce blood flow and increase mean transit time through hypocapnia has significant implications in thermal therapy, since heat dissipation is a major factor in limiting the effectiveness of treatment.

For proton dose calculations in heterogeneous media, it was shown in a previous work that the conventional pencil beam approach based on pathlength scaling does not properly account for scattering effects in nonwater media (Szymanowski and Oelfke 2002 Phys. Med. Biol.47 3313–30). A two-dimensional scaling method was therefore introduced, which is able to predict with high accuracy the propagation of proton pencil beams both along the depth and the lateral directions in inhomogeneous media. In order to integrate this improved pencil beam algorithm in a CT based treatment planning system, two CT calibration curves are needed. The first one relates the Hounsfield numbers to the relative stopping powers, as for the conventional pencil beam approach. The second curve is to relate the Hounsfield numbers to the material-specific lateral scaling factors. The purpose of this work is to provide the CT calibration curves needed for the integration of the pencil beam algorithm featuring the two-dimensional scaling method. Similarly to as suggested by Schneider et al (1996 Phys. Med. Biol.41 111–24) for the calibration curve in terms of stopping powers, we follow a stoichiometric procedure to get the calibration curve in terms of material-specific lateral scaling factors. The calibration curves for a CT scanner of the type Siemens Somatom Plus 4 are obtained from the analytical calculation of the CT Hounsfield numbers, relative stopping powers and material-specific lateral scaling factors for human biological tissues.

Heavy ion therapy has two definite advantages: good dose localization and higher biological effect. Range calculation of the heavy ions is an important factor in treatment planning. X-ray CT numbers are used to estimate the heavy ion range by looking up values in a conversion table which relates empirically photon attenuation in tissues to particle stopping power; this is one source of uncertainty in the treatment planning. Use of positron emitting radioactive beams along with a positron emission tomograph or a positron camera gives range information and may be used as a means of checking in heavy ion treatment planning. However, the metabolism of the implanted positron emitters in a living object is unpredictable because the chemical forms of these emitters are unknown and the metabolism is dependent on the organ species and may be influenced by many factors such as blood flow rate and fluid components present. In this paper, the washout rate of ¹¹C activity implanted by injecting energetic ¹¹C beams into thigh muscle of a rear leg of a rabbit is presented. The washout was found to consist of two components, the shorter one was about 4.2 ± 1.1 min and the longer one ranged from 91 to 124 min. About one third of the implanted β⁺ activity can be used for imaging and the rest was washed out of the target area.

Total skin electron (TSE) radiotherapy is routinely used to treat cutaneous T-cell lymphomas and can be implemented using a modified Stanford technique. In our centre, the composite depth dose for this technique is achieved by a combination of two patient positions per day over a three-day cycle, and two gantry angles per patient position. Due to patient morphology, underdosed regions typically occur and have historically been measured using multiple thermoluminescent dosimeters (TLDs). We show that radiochromic film can be used as a two-dimensional relative dosimeter to measure the percent depth dose in TSE radiotherapy. Composite depth dose curves were measured in a cylindrical, polystyrene phantom and compared with TLD data. Both multiple films (1 film per day) and a single film were used in order to reproduce a realistic clinical scenario. First, three individual films were used to measure the depth dose, one per treatment day, and then compared with TLD data; this comparison showed a reasonable agreement. Secondly, a single film was used to measure the dose delivered over three daily treatments and then compared with TLD data; this comparison showed good agreement throughout the depth dose, which includes doses well below 1 Gy. It will be shown that one piece of radiochromic film is sufficient to measure the composite percent depth dose for a TSE beam, hence making radiochromic film a suitable candidate for monitoring underdosed patient regions.

A modified sector-integration method has been developed that predicts electron beam output factor at any point on the beam central axis, for a given source to surface distance (SSD), as a function of the geometry of the irradiated field. The main concept of this method is that with the arbitrary field shape divided into small sectors, the individual contributions from each sector can be calculated based on the sector radius, using a dataset consisting of circular inserts of standard radii. A computer program was developed based on this algorithm. The program interfaces to a digital camera that is used to capture the shape of the electron insert. We compared the calculated and the measured output factors and per cent depth doses (PDDs) at different SSDs for various rectangular inserts and a typical irregularly shaped insert used in our clinic. To determine the geometric limitations of this algorithm, a series of rectangular inserts were designed with the long-to-short axis ratio between 1:1 and 7:1. The agreement between calculation and measurement for the electron output and PDD was generally within 2% (or 2 mm) for energies from 6 to 20 MeV.

Recently, energy- and intensity-modulated electron radiotherapy (MERT) has garnered a growing interest for the treatment of superficial targets. In this work, we carried out a comparative dosimetry study to evaluate MERT, photon beam intensity-modulated radiation therapy (IMRT) and conventional tangential photon beams for the treatment of breast cancer. A Monte Carlo based treatment planning system has been investigated, which consists of a set of software tools to perform accurate dose calculation, treatment optimization, leaf sequencing and plan analysis. We have compared breast treatment plans generated using this home-grown treatment optimization and dose calculation software for these treatment techniques. The MERT plans were planned with up to two gantry angles and four nominal energies (6, 9, 12 and 16 MeV). The tangential photon treatment plans were planned with 6 MV wedged photon beams. The IMRT plans were planned using both multiple-gantry 6 MV photon beams or two 6 MV tangential beams. Our results show that tangential IMRT can reduce the dose to the lung, heart and contralateral breast compared to conventional tangential wedged beams (up to 50% reduction in high dose volume or 5 Gy in the maximum dose). MERT can reduce the maximum dose to the lung by up to 20 Gy and to the heart by up to 35 Gy compared to conventional tangential wedged beams. Multiple beam angle IMRT can significantly reduce the maximum dose to the lung and heart (up to 20 Gy) but it induces low and medium doses to a large volume of normal tissues including lung, heart and contralateral breast. It is concluded that MERT has superior capabilities to achieve dose conformity both laterally and in the depth direction, which will be well suited for treating superficial targets such as breast cancer.

The inverse radiation treatment planning model for a dynamic multileaf collimator (MLC) is used to find the optimal solution of planning problem. The model for dynamic MLC is explained in Tervo et al (2003 Appl. Math. Comput.135 227–50). The advantage of this model is that it optimizes leaf velocity parameters directly. Our algorithm uses a gradient-based local optimization method. Two patient cases, prostate carcinoma and tonsilla carcinoma, are studied. Field arrangements are pre-selected and velocity parameters for MLC leaves are optimized to obtain the prescribed dose in the patient space. In both simulated cases, high dose distribution conforms the planning target volume well and organs-at-risk are saved in most parts. Simulations show that the model has its functionality in patient treatments, although it is still formal and needs further development.

A pre-clinical characterization of the first fission converter based epithermal neutron beam (FCB) designed for boron neutron capture therapy (BNCT) has been performed. Calculated design parameters describing the physical performance of the aluminium and Teflon® filtered beam were confirmed from neutron fluence and absorbed dose rate measurements performed with activation foils and paired ionization chambers. The facility currently provides an epithermal neutron flux of 4.6 × 10⁹ n cm⁻² s⁻¹ in-air at the patient position that makes it the most intense BNCT source in the world. This epithermal neutron flux is accompanied by very low specific photon and fast neutron absorbed doses of 3.5 ± 0.5 and 1.4 ± 0.2 × 10⁻¹³ Gy cm², respectively. A therapeutic dose rate of 1.7 RBE Gy min⁻¹ is achievable at the advantage depth of 97 mm when boronated phenylalanine (BPA) is used as the delivery agent, giving an average therapeutic ratio of 5.7. In clinical trials of normal tissue tolerance when using the FCB, the effective prescribed dose is due principally to neutron interactions with the nonselectively absorbed BPA present in brain. If an advanced compound is considered, the dose to brain would instead be predominately from the photon kerma induced by thermal neutron capture in hydrogen and advantage parameters of 0.88 Gy min⁻¹, 121 mm and 10.8 would be realized for the therapeutic dose rate, advantage depth and therapeutic ratio, respectively. This study confirms the success of a new approach to producing a high intensity, high purity epithermal neutron source that attains near optimal physical performance and which is well suited to exploit the next generation of boron delivery agents.

How to speed up Monte Carlo (MC) simulation in dose calculation without losing its intrinsic accuracy is one of the key issues of making a clinical MC dose engine. In this study we intensively investigated a special parallel computation technique, the vectorization technique, to boost simulation efficiency on a personal computer (PC) without extra hardware investment. A MC code, dose planning method (DPM), was extensively modified into a vectorized code, V-DPM, using the streaming single-instruction–multiple-data extension (SSE) parallel computation model. Comparative simulations were conducted for typical simulation cases in both DPM and V-DPM codes. We found that in every case the V-DPM code runs 1.5 times faster than the DPM code with variance of 0.6%.