261 poster Initial experiences in a CT simulation only department

S96

Posters

moving core driven by a square wave from a function generator (HP 3312A) were recorded with the detector and simultaneously both the signal from the function generator and the end-of-line and end-of-frame signals from the frame grabber were recorded with the A/D card. Time stamps based on a system-wide monotonic counter (win32 API function GetTickCount0, which has a resolution of milliseconds) were added to the A/D signal and in the callback for each acquired image. The motion of the core in the coil is virtually instantaneous, compared to the acquisitionspeed of the detector. Results: By comparing the signals and the images, it follows that each callback is made about 38 ms after the image acquisition is completed. By recording the end-of-line and end-of-frame signal from the frame grabber together with one or more analogue signals, it is possible to find the analogue value exactly at the time each line of the x-ray image was recorded. Conclusion: We combined a portal imaging system with an A/D converter and validated the synchronization between images and external signals. We will use this system in future for cone-beam CT, dynamic IMRT verification, respiratory correlated imaging and dosimetric experiments. 258

poster

Comparison of attenuation correction using transmission PET and planning CT for detection of lung cancer I. Fitton 1, M. van Herk 1, R.J.H.M. Steenbakkers 1, J.C. Duppen 1, K.E.I. Deurloo 1, E.F.I. Comans 3, S.H. Muller 1, P.J:C.M. Nowak 2, C.R.N. Rasch 1 1The Netherlands Cancer Institute, Radiotherapy, Amsterdam, The Netherlands 2Erasmus University Medical Center, Radiotherapy, Rotterdam, The Netherlands 3VU University Medical Center, Nuclear Medicine, Amsterdam, The Netherlands Backaround and uurDose: For PET scanners without a built-in CT, transmission PET images are used for photon attenuation correction. The purpose of this study is to Compare attenuation correction using transmission PET (ACPET) and using the planning CT (ACCT) for detection of lung cancer. Methods: We applied an attenuation correction method based on the Chang algorithm, using local attenuation coefficients estimated from the spatially registered planning CT data. We evaluated matched PET and CT images of 20 lung cancer patients. First, the contribution of the attenuation correction method to detect lung cancers was studied visually by comparing the images with delineations by several physicians from CT images. Second, a quantitative comparison of the noise and the activity levels was performed. To determine the signal to noise ratio (SNR), we calculated the ratio between the standard deviation of data extracted from a window inside the lung, and the mean of pixels values extracted from a window inside the tumor. Third, the contrast between the tumor and the surrounding tissues was measured for tumor and lymph nodes. Results: Visually the ACCT images show a better discrimination of the tumors and the lymph nodes from the background than both the ACPET and uncorrected images. They appear with a higher intensity and it is possible to detect them more easily and earlier in the sequence of images. These observations are confirmed by the quantitative study: the SNR on ACCT images is 1 to 5 times better than on the uncorrected images and from 1 to 8 times better than on the ACPET images. The improvement of the contrast between a lung tumor and its neighborhood on ACCT images compared with those on ACPET images is on average 13.9% with a standard deviation of 15.6%. In the case of the lymph nodes, the improvement of the contrast between ACCT images and ACPET images is on average 16.0% with a standard deviation of 6.5%. Conclusions: We have demonstrated the efficacy of an attenuation correction method based on the use of the registered planning CT images, highlighting firstly the reduction of the noise, and secondly the improvement of the contrast in PET images. Thus, lung tumors and lymph nodes are better differentiated from the background of the PET images. 259

poster

CT image segmentation using information theoretic criteria L. Hibbard Computerized Medical Systems, Inc., Physics & Research, St. Louis, U.S.A. Image segmentations based on maximum likelihood or maximum a postedori analyses of object textures usually assume parametric models (e.g., Gaussian) for distributions of these features. For real images, parameter accuracy and model stationarity may be elusive, so that model-free inference methods ought to have an advantage over those that are model-

dependent. Functions of the relative entropy from information theory can produce minimum error, model-free inferences, and can detect the boundary of an image object by maximizing the RE between the pixel distributions inside and outside a flexible curve contour. A generalization of the RE--the Jensen-Renyi divergence--computes optimal n-way decisions and can contour multiple objects in an image simultaneously. Seed regions expand naturally and multiple contours tend not to overlap. We apply these functions to contour patient anatomy in X-ray computed tomography (CT) for radiotherapy treatment planning. 260

poster

An opening fully exchange of data in a radiotherapy department: methods and current benefit C. Fiandra 1, S. Anglesio2, F.R. Giglioli2, R. Ragona 1, G. Rossi 3, U. Ricardi 1 1University of Turin, Radiotherapy, Turin, Italy 2Az. Osp. S. Giovanni Battista, Medical Physics, Turin, Italy 3Az. Osp. S. Giovanni Battista, Radiotherapy, Turin, Italy A fully integrated system for the multimedia management of images and patient clinical information has been implemented in our radiotherapy department (Medfolio system). It provides new or updated demographic and visit information about patients, reducing data entry, and updating that information to ensure clinical data synchronization with'technical data supplied from radiotherapy planning. It supports DICOM protocol, of course, to capture images from virtually any DICOM source, including electronic portal imagers, simulation imaging systems, CT, MRI, X-ray, and film digitizers, hence the patient images are stored as part of the integrated electronic medical record. To do that the department must provide a really covering network extending everywhere. In fact a radiotherapy network addresses many objectives: first of all managing both the information necessary for treatment planning (target volumes definition, planning dosimetry) and the control of all parameters involved during the patient's treatment under the treatment unit (R&V systems); then with a network based on the Ethernet and TCP/IP protocol is possible to create a unique database for anagrafic data, diagnosis and eventually radiological or istological exams previous to radiotherapy treatment, furthermore on the same database the patient will be scheduled by the system for all services that he needs (TC, simulation, treatment...) We remind that most of the information circulating in a radiotherapy department are codified by the supplement to the Dicom standard, Dicom RT, that specifies five data objects known in Dicom as Information Object Definition for Relevant Radiotherapy by which a wide range of device types are represented, (treatment planning systems, portal imaging devices, linear accelerators, recording and verifying systems, conventional and virtual simulators). So Dicom RT represents an absolutly necessary tool for standardized transfer of most of the information wich circulates in a radiotherapy department. Future developments in implementation of the system will be aimed to a better integration both with the Hospital Information and our intranet too, to simplifiing ever more communication outside and inside our department. 261

poster

Initial experiences in a CT simulation only department G. Baker, P. Marshall, F. Mall, K. Brasted, M. Lipsham, J. Sorunt Kent Oncology Centre, Medical Physics Department, Canterbury, United Kingdom With the move towards conformal radiotherapy for more treatment sites and the development of true 3-dimensional treatment planning systems there is an increased need to acquire 3-dimensional data sets for each patient. Historically, radiotherapy departments have often had to rely on limited access to a CT scanner in the diagnostic radiology department. The need for CT scans for an increasing number of patients has led some departments to purchase their own scanner in addition to their conventional simulator. At Canterbury, space is at a premium and when the old simulator needed replacement, the decision was made to replace the conventional simulator with a CT Simulator; i.e. a CT scanner and Virtual Simulation software package. The perceived advantages are: a) One fewer visit for the patient during the planning phase b) A reduction in the time spent on acquiring images for each patient enabling a greater throughput of patients c) A true 3-D data set available for each patient which should improve identification and Iocalisation of the target and organs at risk d) Oncologists do not need to be present at the time of imaging but can sim-

Posters

$97

ulate the patient at their convenience. The only major drawback anticipated was the increased time required to verify the plan on the treatment set. We present our initial experiences of running a small department with CT Simulation alone. In particular we will describe changes in our working practices, any unforeseen problems that arose and audit our experiences against the perceived advantages.

IMRT 262

poster

Influence of the MLC leaf width on the dose distribution: a Monte Carlo study A. Leal 1,2, F. S~nchez-Doblado 1,2, R. Arr~ns2, R. Capote 1,2 E. Carrasco 1,2, J.L Lagares 1,2, J. Rosell63, M. Perucha 1, E. Molina 1, J.A. Terr6n 2 1Universidad de Sevilla, Dept. Fisiologfa Medica y Bioffsica, Sevilla, Spain 2Hospital Virgen Macarena, Radiophysics, Sevilla, Spain 3Hospital General ERESA, Radiotherapy, Valencia, Spain Introduction: iMRT techniques and the use of MLCs allow the delivery of complex fluence distribution. This combination administers the accurate dose to the PTV, while minimizes the dose on normal tissues. Some MLCs are integrated in the lineal accelerator head using a leaf width which projecting to 1 cm at the isocenter. Other companies include mini-MLC or micro-MLC designs with leaves having a width of 0.5 cm or less projected at the isocenter. Nowadays, TPS are developed to calculate treatments using different MLC geometries but the transmission through the leaves and the exact shape of the penumbra regions are difficult to be modelled with accuracy. The Monte Carlo (MC) method can be used to simulate the particle transport calculating the dose distribution and then, to isolate the effect of using variable leaf width. Material and Method: The BEAM/EGS4 code was employed to model a Siemens PRIMUS linac at the Hospital Virgen Macarena, featuring an MLC with a leaf width projecting 1 cm at the isocenter. Based on this real model, a virtual head was designed allowing for a variation of the leaf width projection. Both the real tinac as well as a virtual linac with leaves projecting 0.5 cm were used to calculate several real cases treated with both: IMRT and classic solutions. For each treatment, the positioning of the leaves to achieve the best fitting was the only difference between the two models. Several inverse planned treatments were also analysed to study the way the inverse algorithm works for different geometries. The cases were chosen to include different degrees of irregularity and inhomogeneity. Gamma analysis and comparative study was made between the results obtained using each MLC leaf width. Results: Differences were found close to concavities in the volumes where the narrower leaf can shape better. Several cases showed differences of millimetres between isodoses involved around these concavities. Nevertheless, the gamma analysis (dose and distance criteria of 3% and 3 mm, respectively) showed minor disagreement. Despite these differencies, larger in the head and neck cases, the impact on the corresponding DVHs was low. Conclusions: The results found showed a very low influence when using the MLC model with a leaf proyecting 0.5 cm at the isocenter instead of the MLC model with a leaf proyecting 1 cm. 263

technique, the conception and the building of a representative head and neck phantom and finally DICOM network developments. This phantom is sent to each participating center and a multicentric dosimetry quality assurance programme is being installed according to clear protocols. Moreover, to make such intercomparisons much easier and more efficient, this last subgroup is collaborating and working on the conception of a new DICOM RT quality assurance platform. This project will enable exchange of information possible between different treatment planning systems ansd will be used for QA procedure of various treatment parameters in the framework of multi-institutional clinical trials. Such tool will enable collection of data on patient anatomy, contouring of various organs and target volumes, beams and dose distributions. Of course, such facility will be used for remote specialist support in the framework of a network of cooperating centers. Finally, other activities are still in progress and are related to the establishment of a clinical implementation program, QA routine procedures, and report on incidents and accidents that occurred during IMRT treatments. Obviously, some specific topics related to IMRT in head and neck still remain such as dose constraints on certain volumes, fractionation of IMRT treatments (concept of simultaneous integrated boost), extension of target volumes under the skin, etc. A consensus has to occur and all these topics will be discussed in this communication. 266

poster

Intensity modulated radiation therapy using a variable aperture collimator (VAC) S. Webb 1, G_ Hartmann2, G. Echnet2, IV. Schleget 2 1Joint Department of Physics, Institute of Cancer Research, London and Royal Marsden NHS Trust, UK 2Medizinische Physik, DKFZ , Heidelberg, Germany Puroose: the study aimed to investigate whether intensity modulated radiation therapy (IMRT) could be performed without the use of a conventional multileaf collimator (MLC). It extended some earlier concepts of using a tertiary mask plus jaws for delivering IMRT without a MLC (Webb 2002 PMB 47, 257-275 and Webb 2002 PMB 47, 1869-1879). Method: the new concept is to sweep a variable aperture collimator (VAC) across the space of the intensity-modulated beam (IMB) to be delivered and to strip this IMB down into multiple-static-field components, each deliverable with the VAC (patent pending). For an N-columns by M-rows VAC, the VAC comprises two banks of single-bixel attenuators that are parked in rows 1 and M but can be independently driven into any of the otherwise open bixel spaces. The number of possible patterns that can be created is ON + C N'I + C N'2 4- C where C=M(M-1)/2. The VAC is able to be located in any (N x M) patch in the IMB via a translating cradle double-focused in any (N x M) pathc in the IMB via a translating cradle double-focused to the source. This study modelled the performance of a variety of such VACs, varying N and M, allowing rotation and a special form of hybrid VAC with variable N. Results: the stripping algorithm will be described and it has been shown, for several designs of VAC, that the mean number of field components and mean number of monitor units isless using the VAC than would be required for a jaws-only (JO) decomposition. The VAC would be simpler to construct than several previously suggested jaws-plus-mask (J+M) combinations. We have proposed a design concept of a hybrid VAC and shown that it is advantageous to add the potential to rotate the simple or hybrid VAC for some components relative to the field to be modulated. Conclusion: construction of a prototype is about to commence following the detailed modelling study. .

.

.

.

poster

IMRT in head and neck tumours: a new physics working group "GORTEC IMRT physics": activities and results M. Tomsei 1, P. Aletti 2 1U.C.L. St-Luc University Hospital, Radiation Oncology - Medical Physics Unit, Brussels, Belgium 2C.A. V., Radiation oncology- Medical Physics unit, Nancy, France Basically, GORTEC "Groupe Oncologie Radiotherapie Tete Et Cou" is a french medical task group which is dealing with clinical trials related to head and neck cancer. Following a willingness to move towards conformal and intensity modulated radiotherapy, a new physics group was created to go forward and in a parrallel way. As a result, GORTEC IMRT physics was born through a first meeting in Brussels in 2001 and is currently gathering 21 physicists spread over 1 belgian and 13 french radiotherapy departments. Among these, many commercial treatment planning systems and different linac technologies are represented. Several activities in this group are distributed into subgroups working on dosimetry intercomparison studies, quality assurance programmes for this

267

poster

High leaf-speed effects in delivery of intensity modulated radiotherapy with a micro-multileaf collimator C. Garibaldi 1, R. Azzeroni 1, F. Cattani 1, R. Cambria 1, G. Tosi 1, R. Orecchia2

1European Institute of Oncology, Medical Physics, Milano, Italy 2European Institute of Oncology, Radiation Oncology, Milano, Italy Puroose: To investigate main characteristics of high leaf-speed IMRT delivery with a dynamic micro-multileaf collimator (mMLC). Materials and methods: At our Institute, a BrainLAB m3 mMLC, attached to a Varian Clinac 600 CD linac and combined with BrainSCAN inverse planning module, was installed to deliver IMRT. The leaf sequence algorithm implemented in the inverse planning module does not take into account the maximum physical leaf-speed of the mMLC. To obtain the desired leaf and monitor units (MU) sequences, the linac control system drives the leaves at their maximum speed while modulating the dose rate, if the leaf speed exceeds the limit. The characteristics of high