Frédéric Beaulieu

Multiobjective optimization with a modified simulated annealing algorithm for external beam radiotherapy treatment planning

Inverse planning in external beam radiotherapy often requires a scalar objective function that incorporates importance factors to mimic the planners preferences between conflicting objectives. Defining those importance factors is not straightforward, and frequently leads to an iterative process in which the importance factors become variables of the optimization problem. In order to avoid this drawback of inverse planning, optimization using algorithms more suited to multiobjective optimization, such as evolutionary algorithms, has been suggested. However, much inverse planning software, including one based on simulated annealing developed at our institution, does not include multiobjective-oriented algorithms. This work investigates the performance of a modified simulated annealing algorithm used to drive aperture-based intensity-modulated radiotherapy inverse planning software in a multiobjective optimization framework. For a few test cases involving gastric cancer patients, the use of this new algorithm leads to an increase in optimization speed of a little more than a factor of 2 over a conventional simulated annealing algorithm, while giving a close approximation of the solutions produced by a standard simulated annealing. A simple graphical user interface designed to facilitate the decision-making process that follows an optimization is also presented.

Clinical Outcome of Adjuvant Treatment of Endometrial Cancer Using Aperture-Based Intensity-Modulated Radiotherapy

PURPOSE To assess disease control and acute and chronic toxicity with aperture-based intensity-modulated radiotherapy (AB-IMRT) for postoperative pelvic irradiation of endometrial cancer. METHODS AND MATERIALS Between January and July 2005, after hysterectomy for endometrial cancer, 15 patients received 45 Gy to the pelvis using AB-IMRT. The AB-IMRT plans were generated by an in-house treatment planning system (Ballista). The AB-IMRT plans were used for treatment and were dosimetrically compared with three other approaches: conventional four-field, enlarged four-field, and beamlet-based IMRT (BB-IMRT). Disease control and toxicity were prospectively recorded and compared with retrospective data from 30 patients treated with a conventional four-field technique. RESULTS At a median follow-up of 27 months (range, 23-30), no relapse was noted among the AB-IMRT group compared with five relapses in the control group (p = 0.1). The characteristics of each group were similar, except for the mean body mass index, timing of brachytherapy, and applicator type used. Patients treated with AB-IMRT experienced more frequent Grade 2 or greater gastrointestinal acute toxicity (87% vs. 53%, p = 0.02). No statistically significant difference was noted between the two groups regarding the incidence or severity of chronic toxicities. AB-IMRT plans significantly improved target coverage (93% vs. 76% of planning target volume receiving 45 Gy for AB-IMRT vs. conventional four-field technique, respectively). The sparing of organs at risk was similar to that of BB-IMRT. CONCLUSION The results of our study have shown that AB-IMRT provides excellent disease control with equivalent late toxicity compared with the conventional four-field technique. AB-IMRT provided treatment delivery and quality assurance advantages compared with BB-IMRT and could reduce the risk of second malignancy compared with BB-IMRT.

Simultaneous optimization of beam orientations, wedge filters and field weights for inverse planning with anatomy-based MLC fields

As an alternative between manual planning and beamlet-based IMRT, we have developed an optimization system for inverse planning with anatomy-based MLC fields. In this system, named Ballista, the orientation (table and gantry), the wedge filter and the field weights are simultaneously optimized for every beam. An interesting feature is that the system is coupled to Pinnacle3 by means of the PinnComm interface, and uses its convolution dose calculation engine. A fully automatic MLC segmentation algorithm is also included. The plan evaluation is based on a quasi-random sampling and on a quadratic objective function with penalty-like constraints. For efficiency, optimal wedge angles and wedge orientations are determined using the concept of the super-omni wedge. A bound-constrained quasi-Newton algorithm performs field weight optimization, while a fast simulated annealing algorithm selects the optimal beam orientations. Moreover, in order to generate directly deliverable plans, the following practical considerations have been incorporated in the system: collision between the gantry and the table as well as avoidance of the radio-opaque elements of a table top. We illustrate the performance of the new system on two patients. In a rhabdomyosarcoma case, the system generated plans improving both the target coverage and the sparing of the parotide, as compared to a manually designed plan. In the second case presented, the system successfully produced an adequate plan for the treatment of the prostate while avoiding both hip prostheses. For the many cases where full IMRT may not be necessary, the system efficiently generates satisfactory plans meeting the clinical objectives, while keeping the treatment verification much simpler.

Functional avoidance of lung in plan optimization with an aperture-based inverse planning system.

PURPOSE To implement SPECT-based optimization in an anatomy-based aperture inverse planning system for the functional avoidance of lung in thoracic irradiation. MATERIAL AND METHODS SPECT information has been introduced as a voxel-by-voxel modulation of lung importance factors proportionally to the local perfusion count. Fifteen cases of lung cancer have been retrospectively analyzed by generating angle-optimized non-coplanar plans, comparing a purely anatomical approach and our functional approach. Planning target volume coverage and lung sparing have been compared. Statistical significance was assessed by a Wilcoxon matched pairs test. RESULTS For similar target coverage, perfusion-weighted volume receiving 10 Gy was reduced by a median of 2.2% (p=0.022) and mean perfusion-weighted lung dose, by a median of 0.9 Gy (p=0.001). A separate analysis of patients with localized or non-uniform hypoperfusion could not show which would benefit more from SPECT-based treatment planning. Redirection of dose sometimes created overdosage regions in the target volume. Plans consisted of a similar number of segments and monitor units. CONCLUSIONS Angle optimization and SPECT-based modulation of importance factors allowed for functional avoidance of the lung while preserving target coverage. The technique could be also applied to implement PET-based modulation inside the target volume, leading to a safer dose escalation.

Automatic generation of anatomy-based MLC fields in aperture-based IMRT

We have developed an algorithm to automatically generate anatomy-based MLC fields. For each beam, a first field is adjusted to the projection of the target in a beams eye view, allowing subsequent fields to be derived from this conformal field by removing the overlapping surface of each organ at risk, respectively. The projections are based on a surface sampling of the anatomical structures. On top of the MLC mechanical constraints, verification constraints are imposed on the MLC segments, in order to get reliable dosimetry using a commercial dose calculation engine. Thus, in each direction, the apertures cross-section must be greater than a specified threshold, in our case 2 cm. Furthermore, junctions are not tolerated in order to avoid underdosage, for instance from the tongue-and-groove effect. The use of such MLC fields simplifies the verification process. The performance of the algorithm is illustrated for head and neck, thorax and prostate cases. Only a fraction of a second of CPU time is required to perform the segmentation for each beam.

Dose escalation in the radiotherapy of non-small-cell lung cancer with aperture-based intensity modulation and photon beam energy optimization for non-preselected patients.

PURPOSE To verify the potential of aperture-based intensity-modulated radiotherapy (AB-IMRT) to realize dose escalation plans for non-preselected non-small-cell lung cancer (NSCLC) patients, using photon beam energy optimization. METHODS AND MATERIALS Seven cases of NSCLC were retrospectively studied. Clinical reference plans were made at 60 Gy by an experienced dosimetrist. Dose escalation was applied to PTV2, a subvolume within the main PTV1. Escalation plans were optimized by considering beam angles (table and gantry), energy (6 and 23 MV) and weights, for an increasing dose to the PTV2, starting from 66 Gy and keeping 30 fractions. RESULTS In five cases, doses over 78 Gy could be achieved before exceeding organs at risk (OARs) standard tolerance. Peripheral overdosages, as well as lung and spinal cord tolerance doses, limited escalation. Means+/-SD V(95%) parameters were (97.3+/-0.9)% for PTV1s and (96.7+/-2.2)% for PTV2s. Doses to OARs were also maintained at acceptable levels. Optimized plans made use of both low- and high-energy beams and had a similar number of monitor units compared to the 60 Gy clinical plans. CONCLUSIONS The AB-IMRT system can successfully realize dose escalation for a sizeable number of cases. Plans produced contained few large segments, and are applicable to a wide range of tumor volumes and locations.

Optimization of photon beam energy in aperture‐based inverse planning

Optimal choice of beam energy in radiation therapy is easy in many well‐documented cases, but less obvious in some others. Low‐energy beams may provide better conformity around the target than their high‐energy counterparts due to reduced lateral scatter, but they also contribute to overdosage of peripheral normal tissue. Beam energy was added as an optimization parameter in an automatic aperture‐based inverse planning system. We have investigated a total of six cases for two sites (prostate and lung), representative of deep‐seated and moderately deep‐seated tumors. For one case for each site, different numbers of beam incidences were considered. The other cases for each site were optimized using a fixed number of incidences. Four types of plans were optimized: 6 MV, 23 MV, and mixed energy plans with one or two energies per incidence. Each plan was scored with a dose‐volume cost function. Cost function values, number of segments, monitor units, dose‐volume parameters, and isodose distributions were compared. For the prostate and lung cases, energy mixing improved plans in terms of cost function values, with a more important reduction for a small number of beam incidences. Use of high energy allowed better peripheral tissue sparing, while keeping similar target coverage and sensitive structures avoidance. Low energy contribution to monitor units usually increased with the number of beam incidences. Thus, for deep‐seated and moderately deep‐seated tumors, energy optimization can produce interesting plans with less peripheral dose and monitor units than for low energy alone. PACS numbers: 87.55.de, 87.55.dk, 87.56.N‐

SU‐FF‐T‐275: Improving IMRT Plans Delivery for Head and Neck Cases Using Aperture‐Based MLC Segments

Purpose: To investigate the possibility of performing IMRT in head and neck treatment sites with less segments and monitor units (MU). Materials and Methods: Six pharyngeal cases (n = 6) were analysed and four cases (n = 4), in the sinonasal region. For each one, an IMRT plan was first realized using a commercial software (P3IMRT, Pinnacle3 — IMFAST segmentation algorithm). These patients had to receive 32 fractions of simultaneous integrated boost external beam radiotherapy at 1.8 and 2.15 Gy/fraction, respectively to the low and high risk planning target volumes (PTV1 and PTV2). Then, an‐in‐house inverse planning system, called Ballista, based on predetermined segments, was used to realize comparable plans. Its segments are generated with the subtraction of the projection of the OARs with the PTV (planning target volume). Results: For the pharyngeal Ballista plans, the average volume of the PTV that received at least 100% of the prescribed dose (V100) was 85.0±4.5% for the first prescription (PTV1) and the V 100 for the second prescription (PTV2 — simultaneous integrated boost —) was 78.5±10.9%. With Pinnacle3,the V 100 value was 86.6±4.8% and 81.5±12.4% respectively for PTV1 and PTV2 (see figure 2a and 2b). On average, Ballista plans have required 932±124 MU and 52±10 segments compared to 1238±230 MU and 117±7 segments for Pinnacle3. For the sinonasal Ballista plans, the average V100 obtained was 80.0±3.1%. With Pinnacle3, the V 100 gave 75.7±2.7%. Ballista plans have required an average of 406±54 MU and 22±1 segments compared to 697±133 MU and 99±14 segments for beamlet‐based IMRT.Conclusion: In step‐and‐shoot head and neck IMRT, an anatomy‐based MLC optimization system can achieve similar dosimetric plans comparable to traditional beamlet‐based IMRT with less number of segments and MU.

An EPID-based method to determine mechanical deformations in a linear accelerator

PURPOSE Medical linear accelerators (linac) are delivering increasingly complex treatments using modern techniques in radiation therapy. Complete and precise mechanical QA of the linac is therefore necessary to ensure that there is no unexpected deviation from the gantrys planned course. However, state-of-the-art EPID-based mechanical QA procedures often neglect some degrees of freedom (DOF) like the in-plane rotations of the gantry and imager or the source movements inside the gantry head. Therefore, the purpose of this work is to characterize a 14 DOF method for the mechanical QA of linacs. This method seeks to measure every mechanical deformation in a linac, including source movements, in addition to relevant clinical parameters like mechanical and radiation isocenters. METHODS A widely available commercial phantom and a custom-made accessory inserted in the linacs interface mount are imaged using the electronic portal imaging device (EPID) at multiple gantry angles. Then, simulated images are generated using the nominal geometry of the linac and digitized models of the phantoms. The nominal geometry used to generate these images can be modified using 14 DOF (3 rigid rotations and 3 translations for the imager and the gantry, and 2 in-plane translations of the source) and any change will modify the simulated image. The set of mechanical deformations that minimizes the differences between the simulated and measured image is found using a genetic algorithm coupled with a gradient-descent optimizer. Phantom mispositioning and gantry angular offset were subsequently calculated and extracted from the results. Simulations of the performances of the method for different levels of noise in the phantom models were performed to calculate the absolute uncertainty of the measured mechanical deformations. The measured source positions and the center of collimation were used to define the beam central axis and calculate the radiation isocenter position and radius. RESULTS After the simultaneous optimization of the 14 DOF, the average distance between the center of the measured and simulated ball bearings on the imager was 0.086 mm. Over the course of a full counter-clockwise gantry rotation, all mechanical deformations were measured, showing sub-millimeter translations and rotations smaller than 1° along every axis. The average absolute uncertainty of the 14 DOF (1 SD) was 0.15 mm or degree. Phantom positioning errors were determined with more than 0.1 mm precision. Errors introduced in the experimental setup like phantom positioning errors, source movements or gantry angular offsets were all successfully detected by our QA method. The mechanical deformations measured are shown to be reproducible over the course of a few weeks and are not sensitive to the experimental setup. CONCLUSION This work presents of new method for an accurate mechanical QA of the linacs. It features a 14 DOF model of the mechanical deformations that is both more complete and precise than other available methods. It has demonstrated sub-millimeter accuracy through simulation and experimentation. Introduced errors were successfully detected with high precision.

SU‐FF‐T‐151: Perfusion‐Based Plan Optimization for Lung Cancer Using An Anatomy‐Based Aperture Inverse Planning System

Purpose: To implement SPECT‐based plan optimization in an anatomy‐based aperture inverse planning system (IPS) for the avoidance of functional pulmonary regions for cases of lungcancer.Method and Materials: The IPS allows simultaneous optimization of beam orientations and weights from apertures defined by an anatomy‐based segmentation. SPECT perfusion information has been integrated in the dose‐volume‐based cost function of the inverse planning system through a voxel‐by‐voxel linear spatial modulation of the importance factors (IFs) according to local perfusion score. For two cases of lungcancer, plans have been generated by the IPS using four non‐coplanar incidences (gantry and couch angles optimized) using a purely anatomical approach and the SPECT‐based approach. Planning target volume (PTV) coverage and lung avoidance (both volumetric and functional) have been compared. Results: Maximum dose to PTV is usually increased when increasing importance of functional lung regions in the optimization, creating boost regions. For the first case, the functional volume of lung receiving 20 Gy (F20) decreases from 28.4% to 22.0% while the mean lung perfused dose (MpLD) decreases from 16.5 Gy to 13.7 Gy. For the second case, the F20 does not vary (26.5%) and the MpLD decreases from 17.4 Gy to 16.6 Gy. All plans produced are simpler than typical IMRT plans, with few segments (5 or 10) and few monitor units (range 285–375) used. Conclusions: The system allows generation of simple aperture‐based IMRT plans with the addition of functional lung sparing when considering SPECT‐based information. However, the extent of the benefit is patient‐dependant and varies according to the perfusion pattern and proximity of other critical structures to the PTV. Boost regions created by the redistribution of dose might prove useful in the context of dose escalation in lung irradiation.