Inferring positions of tumor and nodes in Stage III lung cancer from multiple anatomical surrogates using four-dimensional computed tomography

PURPOSE

To investigate the feasibility of modeling Stage III lung cancer tumor and node positions from anatomical surrogates.

METHODS AND MATERIALS

To localize their centroids, the primary tumor and lymph nodes from 16 Stage III lung cancer patients were contoured in 10 equal-phase planning four-dimensional (4D) computed tomography (CT) image sets. The centroids of anatomical respiratory surrogates (carina, xyphoid, nipples, mid-sternum) in each image set were also localized. The correlations between target and surrogate positions were determined, and ordinary least-squares (OLS) and partial least-squares (PLS) regression models based on a subset of respiratory phases (three to eight randomly selected) were created to predict the target positions in the remaining images. The three-phase image sets that provided the best predictive information were used to create models based on either the carina alone or all surrogates.

RESULTS

The surrogate most correlated with target motion varied widely. Depending on the number of phases used to build the models, mean OLS and PLS errors were 1.0 to 1.4 mm and 0.8 to 1.0 mm, respectively. Models trained on the 0%, 40%, and 80% respiration phases had mean (+/- standard deviation) PLS errors of 0.8 +/- 0.5 mm and 1.1 +/- 1.1 mm for models based on all surrogates and carina alone, respectively. For target coordinates with motion >5 mm, the mean three-phase PLS error based on all surrogates was 1.1 mm.

CONCLUSIONS

Our results establish the feasibility of inferring primary tumor and nodal motion from anatomical surrogates in 4D CT scans of Stage III lung cancer. Using inferential modeling to decrease the processing time of 4D CT scans may facilitate incorporation of patient-specific treatment margins.

The minimum knowledge base for predicting organ-at-risk dose-volume levels and plan-related complications in IMRT planning

IMRT treatment planning requires consideration of two competing objectives: achieving the required amount of radiation for the planning target volume and minimizing the amount of radiation delivered to all other tissues. It is important for planners to understand the tradeoff between competing factors so that the time-consuming human interaction loop (plan-evaluate-modify) can be eliminated. Treatment-plan-surface models have been proposed as a decision support tool to aid treatment planners and clinicians in choosing between rival treatment plans in a multi-plan environment. In this paper, an empirical approach is introduced to determine the minimum number of treatment plans (minimum knowledge base) required to build accurate representations of the IMRT plan surface in order to predict organ-at-risk (OAR) dose-volume (DV) levels and complications as a function of input DV constraint settings corresponding to all involved OARs in the plan. We have tested our approach on five head and neck patients and five whole pelvis/prostate patients. Our results suggest that approximately 30 plans were sufficient to predict DV levels with less than 3% relative error in both head and neck and whole pelvis/prostate cases. In addition, approximately 30-60 plans were sufficient to predict saliva flow rate with less than 2% relative error and to classify rectal bleeding with an accuracy of 90%.

A two-stage sequential linear programming approach to IMRT dose optimization

The conventional IMRT planning process involves two stages in which the first stage consists of fast but approximate idealized pencil beam dose calculations and dose optimization and the second stage consists of discretization of the intensity maps followed by intensity map segmentation and a more accurate final dose calculation corresponding to physical beam apertures. Consequently, there can be differences between the presumed dose distribution corresponding to pencil beam calculations and optimization and a more accurately computed dose distribution corresponding to beam segments that takes into account collimator-specific effects. IMRT optimization is computationally expensive and has therefore led to the use of heuristic (e.g., simulated annealing and genetic algorithms) approaches that do not encompass a global view of the solution space. We modify the traditional two-stage IMRT optimization process by augmenting the second stage via an accurate Monte Carlo-based kernel-superposition dose calculations corresponding to beam apertures combined with an exact mathematical programming-based sequential optimization approach that uses linear programming (SLP). Our approach was tested on three challenging clinical test cases with multileaf collimator constraints corresponding to two vendors. We compared our approach to the conventional IMRT planning approach, a direct-aperture approach and a segment weight optimization approach. Our results in all three cases indicate that the SLP approach outperformed the other approaches, achieving superior critical structure sparing. Convergence of our approach is also demonstrated. Finally, our approach has also been integrated with a commercial treatment planning system and may be utilized clinically.

Towards respiration management in radiation treatment of lung tumors: transferring regions of interest from planning CT to kilovoltage X-ray images

Tracking of lung tumors is imperative for improved radiotherapy treatment. However, the motion of the thoracic organs makes it a complicated task. 4D CT images acquired prior to treatment provide valuable information regarding the motion of organs and tumor, since it is manually annotated. In order to track tumors using treatment-day X-ray images (kV images), we need to find the correspondence with CT images so that projection of tumor region of interest will provide a good estimate about the position of the tumor on the X-ray image. In this study, we propose a method to estimate the alignment and respiration phase corresponding to X-ray images using 4D CT data. Our approach generates Digitally Reconstructed Radiographs (DRRs) using bilateral filter smoothing and computes rigid registration with kV images since the position and orientation of patient might differ between CT and treatment-day image acquisition processes. Instead of using landmark points, our registration method makes use of Kernel Density Estimation over the edges that are not affected much by respiration. To estimate the phase of X-ray, we apply template matching techniques between the lung regions of X-ray and registered DRRs. Our approach gives accurate results for rigid registration and provides a starting point to track tumors using the X-ray images during the treatment.

Moderate predictive value of demographic and behavioral characteristics for a diagnosis of HPV16-positive and HPV16-negative head and neck cancer

Patients with HPV-positive and HPV-negative head and neck squamous cell carcinoma (HNSCC) are significantly different with regard to sociodemographic and behavioral characteristics that clinicians may use to assume tumor HPV status. Machine learning methods were used to evaluate the predictive value of patient characteristics and laboratory biomarkers of HPV exposure for a diagnosis of HPV16-positive HNSCC compared to in situ hybridization, the current gold-standard. Models that used a combination of demographic characteristics such as age, tobacco use, gender, and race had only moderate predictive value for tumor HPV status among all patients with HNSCC (positive predictive value [PPV]=75%, negative predictive value [NPV]=68%) or when limited to oropharynx cancer patients (PPV=55%, NPV=65%) and thus included a sizeable number of false positive and false negative predictions. Prediction was not improved by the addition of other demographic or behavioral factors (sexual behavior, income, education) or biomarkers of HPV16 exposure (L1, E6/7 antibodies or DNA in oral exfoliated cells). Patient demographic and behavioral characteristics as well as HPV biomarkers are not an accurate substitute for clinical testing of tumor HPV status.

Investigation of motion sickness and inertial stability on a moving couch for intra-fraction motion compensation

BACKGROUND. Respiration-induced tumor motion compensation using a treatment couch requires moving the patient at non-trivial speeds. The purpose of this work was to investigate motion sickness and stability of the patient's external surface due to a moving couch with respiration-comparable velocities and accelerations. MATERIAL AND METHODS. A couch was designed to move with a peak-peak displacement of 5 cm and 1 cm in the S-I and A-P directions, respectively, and a period of 3.6 s. Fifty patients completed a 16-question motion sickness assessment questionnaire (MSAQ) prior to, during, and after the study. Seven optical reflectors affixed to the abdomen of each patient were monitored by infrared cameras. The relationship between reflector positions under stationary and moving conditions was evaluated to assess the stability of the patient's external surface. RESULTS AND DISCUSSION. Among the 4800 responses, 95% were 1 (no discomfort) of 9, and there were no scores of 6 or higher. Mild discomfort (scores of 4-5) was similar during couch motion and before couch motion (p = 0.39). Mild discomfort was less common after couch motion (p = 0.039) than before or during couch movement. There was a near 1:1 correspondence between marker-pair regression coefficients and phase offset values during couch-stationary and couch-moving conditions. Our results show that patients do not suffer motion sickness or external surface instability on a moving couch.

Optimal beam arrangement for stereotactic body radiation therapy delivery in lung tumors

PURPOSE

To compare the different beam arrangement and delivery techniques for stereotactic body radiation therapy (SBRT) of lung lesions using the criteria of Radiation Therapy Oncology Group (RTOG) 0236 protocol.

MATERIAL AND METHODS

Thirty-seven medically inoperable lung cancers were evaluated with various planning techniques including multiple coplanar multiple static beams, multiple non-coplanar static beams and arc delivery. Twelve plans were evaluated for each case, including five plans using coplanar fixed beams, six plans using non-coplanar fixed beams and one plan using arc therapy. These plans were compared using the target prescription isodose coverage, high and low dose volumes, and critical organ dose-volume limits.

RESULTS

The prescription isodose coverage, high dose evaluation criteria and dose to critical organs were similar among treatment delivery techniques. However, there were differences in low dose criteria, especially in the ratio of the volume of 50% isodose of the prescription dose to the volume of planning treatment volume (R(50%)). The R(50%) in plans using non-coplanar static beams was lower than other plans in 30 of 37 cases (81%).

CONCLUSION

Based on the dosimetric criteria outlined in RTOG 0236, the treatment technique using non-coplanar static beams showed the most preferable results for SBRT of lung lesions.

Modeling plan-related clinical complications using machine learning tools in a multiplan IMRT framework

PURPOSE

To predict organ-at-risk (OAR) complications as a function of dose-volume (DV) constraint settings without explicit plan computation in a multiplan intensity-modulated radiotherapy (IMRT) framework.

METHODS AND MATERIALS

Several plans were generated by varying the DV constraints (input features) on the OARs (multiplan framework), and the DV levels achieved by the OARs in the plans (plan properties) were modeled as a function of the imposed DV constraint settings. OAR complications were then predicted for each of the plans by using the imposed DV constraints alone (features) or in combination with modeled DV levels (plan properties) as input to machine learning (ML) algorithms. These ML approaches were used to model two OAR complications after head-and-neck and prostate IMRT: xerostomia, and Grade 2 rectal bleeding. Two-fold cross-validation was used for model verification and mean errors are reported.

RESULTS

Errors for modeling the achieved DV values as a function of constraint settings were 0-6%. In the head-and-neck case, the mean absolute prediction error of the saliva flow rate normalized to the pretreatment saliva flow rate was 0.42% with a 95% confidence interval of (0.41-0.43%). In the prostate case, an average prediction accuracy of 97.04% with a 95% confidence interval of (96.67-97.41%) was achieved for Grade 2 rectal bleeding complications.

CONCLUSIONS

ML can be used for predicting OAR complications during treatment planning allowing for alternative DV constraint settings to be assessed within the planning framework.

Inferential modeling and predictive feedback control in real-time motion compensation using the treatment couch during radiotherapy

Tumor motion induced by respiration presents a challenge to the reliable delivery of conformal radiation treatments. Real-time motion compensation represents the technologically most challenging clinical solution but has the potential to overcome the limitations of existing methods. The performance of a real-time couch-based motion compensation system is mainly dependent on two aspects: the ability to infer the internal anatomical position and the performance of the feedback control system. In this paper, we propose two novel methods for the two aspects respectively, and then combine the proposed methods into one system. To accurately estimate the internal tumor position, we present partial-least squares (PLS) regression to predict the position of the diaphragm using skin-based motion surrogates. Four radio-opaque markers were placed on the abdomen of patients who underwent fluoroscopic imaging of the diaphragm. The coordinates of the markers served as input variables and the position of the diaphragm served as the output variable. PLS resulted in lower prediction errors compared with standard multiple linear regression (MLR). The performance of the feedback control system depends on the system dynamics and dead time (delay between the initiation and execution of the control action). While the dynamics of the system can be inverted in a feedback control system, the dead time cannot be inverted. To overcome the dead time of the system, we propose a predictive feedback control system by incorporating forward prediction using least-mean-square (LMS) and recursive least square (RLS) filtering into the couch-based control system. Motion data were obtained using a skin-based marker. The proposed predictive feedback control system was benchmarked against pure feedback control (no forward prediction) and resulted in a significant performance gain. Finally, we combined the PLS inference model and the predictive feedback control to evaluate the overall performance of the feedback control system. Our results show that, with the tumor motion unknown but inferred by skin-based markers through the PLS model, the predictive feedback control system was able to effectively compensate intra-fraction motion.

Automatic segmentation of phase-correlated CT scans through nonrigid image registration using geometrically regularized free-form deformation

Conventional radiotherapy is planned using free-breathing computed tomography (CT), ignoring the motion and deformation of the anatomy from respiration. New breath-hold-synchronized, gated, and four-dimensional (4D) CT acquisition strategies are enabling radiotherapy planning utilizing a set of CT scans belonging to different phases of the breathing cycle. Such 4D treatment planning relies on the availability of tumor and organ contours in all phases. The current practice of manual segmentation is impractical for 4D CT, because it is time consuming and tedious. A viable solution is registration-based segmentation, through which contours provided by an expert for a particular phase are propagated to all other phases while accounting for phase-to-phase motion and anatomical deformation. Deformable image registration is central to this task, and a free-form deformation-based nonrigid image registration algorithm will be presented. Compared with the original algorithm, this version uses novel, computationally simpler geometric constraints to preserve the topology of the dense control-point grid used to represent free-form deformation and prevent tissue fold-over. Using mean squared difference as an image similarity criterion, the inhale phase is registered to the exhale phase of lung CT scans of five patients and of characteristically low-contrast abdominal CT scans of four patients. In addition, using expert contours for the inhale phase, the corresponding contours were automatically generated for the exhale phase. The accuracy of the segmentation (and hence deformable image registration) was judged by comparing automatically segmented contours with expert contours traced directly in the exhale phase scan using three metrics: volume overlap index, root mean square distance, and Hausdorff distance. The accuracy of the segmentation (in terms of radial distance mismatch) was approximately 2 mm in the thorax and 3 mm in the abdomen, which compares favorably to the accuracies reported elsewhere. Unlike most prior work, segmentation of the tumor is also presented. The clinical implementation of 4D treatment planning is critically dependent on automatic segmentation, for which is offered one of the most accurate algorithms yet presented.

The Use of Gated and 4D CT Imaging in Planning for Stereotactic Body Radiation Therapy

The localization of treatment targets is of utmost importance for patients receiving stereotactic body radiation therapy (SBRT), where the dose per fraction is large. While both setup or respiration-induced motion components affect the localization of the treatment volume, the purpose of this work is to describe our management of the intrafraction localization uncertainty induced by normal respiration. At our institution, we have implemented gated computed tomography (CT) acquisition with an active breathing control system (ABC), and 4-dimensional (4D) CT using a skin-based marker and retrospective respiration phase-based image sorting. During gated simulation, 3D CT images were acquired corresponding to end-inhalation and end-exhalation. For 4D CT imaging, 3D CT images were acquired corresponding to 8 phases of the respiratory cycle. In addition to gated or 4D CT images, we acquired a conventional free-breathing CT (FB). For both gated and 4D CT images, the target contours were registered to the FB scan in the planning system. These contours were then combined in the FB image set to form the internal target volume (ITV). Dynamic conformal arc treatment plans were generated for the ITV using the FB scan and the gated or 4D scans with an additional 7-mm margin for patient setup uncertainty. We have described our results for a pancreas and a lung tumor case. Plans were normalized so that the PTV received 95% of the prescription dose. The dose distribution for all the critical structures in the pancreas and lung tumor cases resulted in increased sparing when the ITV was defined using gated or 4D CT images than when the FB scan was used. Our results show that patient-specific target definition using gated or 4D CT scans lead to improved normal tissue sparing.

A multiplan treatment-planning framework: a paradigm shift for intensity-modulated radiotherapy

PURPOSE

To describe a multiplan intensity-modulated radiotherapy (IMRT) planning framework, and to describe a decision support system (DSS) for ranking multiple plans and modeling the planning surface.

METHODS AND MATERIALS

One hundred twenty-five plans were generated sequentially for a head-and-neck case and a pelvic case by varying the dose-volume constraints on each of the organs at risk (OARs). A DSS was used to rank plans according to dose-volume histogram (DVH) values, as well as equivalent uniform dose (EUD) values. Two methods for ranking treatment plans were evaluated: composite criteria and pre-emptive selection. The planning surface determined by the results was modeled using quadratic functions.

RESULTS

The DSS provided an easy-to-use interface for the comparison of multiple plan features. Plan ranking resulted in the identification of one to three "optimal" plans. The planning surface models had good predictive capability with respect to both DVH values and EUD values and generally, errors of <6%. Models generated by minimizing the maximum relative error had significantly lower relative errors than models obtained by minimizing the sum of squared errors. Using the quadratic model, plan properties for one OAR were determined as a function of the other OAR constraint settings. The modeled plan surface can then be used to understand the interdependence of competing planning objectives.

CONCLUSION

The DSS can be used to aid the planner in the selection of the most desirable plan. The collection of quadratic models constructed from the plan data to predict DVH and EUD values generally showed excellent agreement with the actual plan values.

An analysis of the treatment couch and control system dynamics for respiration-induced motion compensation

Sophisticated methods for real-time motion compensation include using the linear accelerator, MLC, or treatment couch. To design such a couch, the required couch and control system dynamics need to be investigated. We used an existing treatment couch known as the Hexapod to gain insight into couch dynamics and an internal model controller to simulate feedback control of respiration-induced motion. The couch dynamics, described using time constants and dead times, were investigated using step inputs. The resulting data were modeled as first and second order systems with dead time. The couch was determined to have a linear response for step inputs < or = 1 cm. Motion data from 12 patients were obtained using a skin marker placed on the abdomen of the patient and the marker data were assumed to be an exact surrogate of tumor motion. The feedback system was modeled with the couch as a second-ordersystem and the controller as a first order system. The time constants of the couch and controller and the dead times were varied starting with parameters obtained from the Hexapod couch and the performance of the feedback system was evaluated. The resulting residual motion under feedback control was generally <0.3 cm when a fast enough couch was simulated.

A difference-matrix metaheuristic for intensity map segmentation in step-and-shoot IMRT delivery

At an intermediate stage of radiation treatment planning for IMRT, most commercial treatment planning systems for IMRT generate intensity maps that describe the grid of beamlet intensities for each beam angle. Intensity map segmentation of the matrix of individual beamlet intensities into a set of MLC apertures and corresponding intensities is then required in order to produce an actual radiation delivery plan for clinical use. Mathematically, this is a very difficult combinatorial optimization problem, especially when mechanical limitations of the MLC lead to many constraints on aperture shape, and setup times for apertures make the number of apertures an important factor in overall treatment time. We have developed, implemented and tested on clinical cases a metaheuristic (that is, a method that provides a framework to guide the repeated application of another heuristic) that efficiently generates very high-quality (low aperture number) segmentations. Our computational results demonstrate that the number of beam apertures and monitor units in the treatment plans resulting from our approach is significantly smaller than the corresponding values for treatment plans generated by the heuristics embedded in a widely use commercial system. We also contrast the excellent results of our fast and robust metaheuristic with results from an 'exact' method, branch-and-cut, which attempts to construct optimal solutions, but, within clinically acceptable time limits, generally fails to produce good solutions, especially for intensity maps with more than five intensity levels. Finally, we show that in no instance is there a clinically significant change of quality associated with our more efficient plans.

Gated CT imaging using a free-breathing respiration signal from flow-volume spirometry

Respiration-induced tumor motion is known to cause artifacts on free-breathing spiral CT images used in treatment planning. This leads to inaccurate delineation of target volumes on planning CT images. Flow-volume spirometry has been used previously for breath-holds during CT scans and radiation treatments using the active breathing control (ABC) system. We have developed a prototype by extending the flow-volume spirometer device to obtain gated CT scans using a PQ 5000 single-slice CT scanner. To test our prototype, we designed motion phantoms to compare image quality obtained with and without gated CT scan acquisition. Spiral and axial (nongated and gated) CT scans were obtained of phantoms with motion periods of 3-5 s and amplitudes of 0.5-2 cm. Errors observed in the volume estimate of these structures were as much as 30% with moving phantoms during CT simulation. Application of motion-gated CT with active breathing control reduced these errors to within 5%. Motion-gated CT was then implemented in patients and the results are presented for two clinical cases: lung and abdomen. In each case, gated scans were acquired at end-inhalation, end-exhalation in addition to a conventional free-breathing (nongated) scan. The gated CT scans revealed reduced artifacts compared with the conventional free-breathing scan. Differences of up to 20% in the volume of the structures were observed between gated and free-breathing scans. A comparison of the overlap of structures between the gated and free-breathing scans revealed misalignment of the structures. These results demonstrate the ability of flow-volume spirometry to reduce errors in target volumes via gating during CT imaging.

Using a photon phase-space source for convolution/superposition dose calculations in radiation therapy

For a given linac design, the dosimetric characteristics of a photon beam are determined uniquely by the energy and radial distributions of the electron beam striking the x-ray target. However, in the usual commissioning of a beam from measured data, a large number of variables can be independently tuned, making it difficult to derive a unique and self-consistent beam model. For example, the measured dosimetric penumbra in water may be attributed in various proportions to the lateral secondary electron range, the focal spot size and the transmission through the tips of a non-divergent collimator; the head-scatter component in the tails of the transverse profiles may not be easy to resolve from phantom scatter and head leakage; and the head-scatter tails corresponding to a certain extra-focal source model may not agree self-consistently with in-air output factors measured on the central axis. To reduce the number of adjustable variables in beam modelling, we replace the focal and extra-focal sources with a single phase-space plane scored just above the highest adjustable collimator in a EGS/BEAM simulation of the linac. The phase-space plane is then used as photon source in a stochastic convolution/superposition dose engine. A photon sampled from the uncollimated phase-space plane is first propagated through an arbitrary collimator arrangement and then interacted in the simulation phantom. Energy deposition kernel rays are then randomly issued from the interaction points and dose is deposited along these rays. The electrons in the phase-space file are used to account for electron contamination. 6 MV and 18 MV photon beams from an Elekta SL linac are used as representative examples. Except for small corrections for monitor backscatter and collimator forward scatter for large field sizes (<0.5% with <20 x 20 cm2 field size), we found that the use of a single phase-space photon source provides accurate and self-consistent results for both relative and absolute dose calculations.

Real-time intra-fraction-motion tracking using the treatment couch: a feasibility study

Significant differences between planned and delivered treatments may occur due to respiration-induced tumour motion, leading to underdosing of parts of the tumour and overdosing of parts of the surrounding critical structures. Existing methods proposed to counter tumour motion include breath-holds, gating and MLC-based tracking. Breath-holds and gating techniques increase treatment time considerably, whereas MLC-based tracking is limited to two dimensions. We present an alternative solution in which a robotic couch moves in real time in response to organ motion. To demonstrate proof-of-principle, we constructed a miniature adaptive couch model consisting of two movable platforms that simulate tumour motion and couch motion, respectively. These platforms were connected via an electronic feedback loop so that the bottom platform responded to the motion of the top platform. We tested our model with a seven-field step-and-shoot delivery case in which we performed three film-based experiments: (1) static geometry, (2) phantom-only motion and (3) phantom motion with simulated couch motion. Our measurements demonstrate that the miniature couch was able to compensate for phantom motion to the extent that the dose distributions were practically indistinguishable from those in static geometry. Motivated by this initial success, we investigated a real-time couch compensation system consisting of a stereoscopic infra-red camera system interfaced to a robotic couch known as the Hexapod, which responds in real time to any change in position detected by the cameras. Optical reflectors placed on a solid water phantom were used as surrogates for motion. We tested the effectiveness of couch-based motion compensation for fixed fields and a dynamic arc delivery cases. Due to hardware limitations, we performed film-based experiments (1), (2) and (3), with the robotic couch at a phantom motion period and dose rate of 16 s and 100 MU min(-1), respectively. Analysis of film measurements showed near-equivalent dose distributions (<or=2 mm agreement of corresponding isodose lines) for static geometry and motion-synchronized real-time robotic couch tracking-based radiation delivery.

A stochastic convolution/superposition method with isocenter sampling to evaluate intrafraction motion effects in IMRT

Current methods to calculate dose distributions with organ motion can be broadly classified as "dose convolution" and "fluence convolution" methods. In the former, a static dose distribution is convolved with the probability distribution function (PDF) that characterizes the motion. However, artifacts are produced near the surface and around inhomogeneities because the method assumes shift invariance. Fluence convolution avoids these artifacts by convolving the PDF with the incident fluence instead of the patient dose. In this paper we present an alternative method that improves the accuracy, generality as well as the speed of dose calculation with organ motion. The algorithm starts by sampling an isocenter point from a parametrically defined space curve corresponding to the patient-specific motion trajectory. Then a photon is sampled in the linac head and propagated through the three-dimensional (3-D) collimator structure corresponding to a particular MLC segment chosen randomly from the planned IMRT leaf sequence. The photon is then made to interact at a point in the CT-based simulation phantom. Randomly sampled monoenergetic kernel rays issued from this point are then made to deposit energy in the voxels. Our method explicitly accounts for MLC-specific effects (spectral hardening, tongue-and-groove, head scatter) as well as changes in SSD with isocentric displacement, assuming that the body moves rigidly with the isocenter. Since the positions are randomly sampled from a continuum, there is no motion discretization, and the computation takes no more time than a static calculation. To validate our method, we obtained ten separate film measurements of an IMRT plan delivered on a phantom moving sinusoidally, with each fraction starting with a random phase. For 2 cm motion amplitude, we found that a ten-fraction average of the film measurements gave an agreement with the calculated infinite fraction average to within 2 mm in the isodose curves. The results also corroborate the existing notion that the interfraction dose variability due to the interplay between the MLC motion and breathing motion averages out over typical multifraction treatments. Simulation with motion waveforms more representative of real breathing indicate that the motion can produce penumbral spreading asymmetric about the static dose distributions. Such calculations can help a clinician decide to use, for example, a larger margin in the superior direction than in the inferior direction. In the paper we demonstrate that a 15 min run on a single CPU can readily illustrate the effect of a patient-specific breathing waveform, and can guide the physician in making informed decisions about margin expansion and dose escalation.

Feasibility of dose escalation using intensity-modulated radiotherapy in posthysterectomy cervical carcinoma

PURPOSE

To evaluate retrospectively the utility of intensity-modulated radiotherapy (IMRT) in reducing the volume of normal tissues receiving radiation at varying dose levels when the female pelvis after hysterectomy is treated to doses of 50.4 Gy and 54 Gy.

METHODS AND MATERIALS

Computed tomography scans from 10 patients who had previously undergone conventional postoperative RT were selected. The clinical tumor volume (vaginal apex and iliac nodes) and organs at risk were contoured. Margins were added to generate the planning tumor volume. The Pinnacle and Corvus planning systems were used to develop conventional and IMRT plans, respectively. Conventional four-field plans were prescribed to deliver 45 Gy (4F(45 Gy)) or 50.4 Gy; eight-field IMRT plans were prescribed to deliver 50.4 Gy (IMRT(50.4 Gy)) or 54 Gy (IMRT(54 Gy)) to the planning tumor volume. All plans were normalized so that > or =97% of the planning tumor volume received the prescribed dose. Student's t test was used to compare the volumes of organs at risk receiving the same doses with different plans.

RESULTS

The mean volume of bowel receiving > or =45 Gy was lower with the IMRT(50.4 Gy) (33% lower) and IMRT(54 Gy) (18% lower) plans than with the 4F(45 Gy) plan. The mean volume of rectum receiving > or =45 Gy or > or =50 Gy was also significantly reduced with the IMRT plans despite an escalation of the prescribed dose from 45 Gy with the conventional plans to 54 Gy with IMRT. The mean volume of bladder treated to 45 Gy was the same or slightly lower with the IMRT(50.4 Gy) and IMRT(54 Gy) plans compared with the 4F(45 Gy) plan. Compared with the 4F(45 Gy) plan, the IMRT(50.4 Gy) plan resulted in a smaller volume of bowel receiving 35-45 Gy and a larger volume of bowel receiving 50-55 Gy. Compared with the 4F(45 Gy) plan, the IMRT(54 Gy) plan resulted in smaller volumes of bowel receiving 45-50 Gy; however, small volumes of bowel received 55-60 Gy with the IMRT plan.

CONCLUSION

Intensity-modulated RT may permit an increase in the radiation dose that can safely be delivered to the central pelvis and pelvic lymph nodes after hysterectomy. However, dose-volume calculations using individual CT scans do not account for internal organ motion. Detailed data concerning the relationships among radiation dose, treatment volume, and treatment effects are lacking, and prospective studies of pelvic IMRT are needed to determine the safety and efficacy of this treatment.

Selection of beam orientations in intensity-modulated radiation therapy using single-beam indices and integer programming

While the process of IMRT planning involves optimization of the dose distribution, the procedure for selecting the beam inputs for this process continues to be largely trial-and-error. We have developed an integer programming (IP) optimization method to optimize beam orientation using mean organ-at-risk (MOD) data from single-beam plans. Two test cases were selected in which one organ-at-risk (OAR) and four OARs were simulated, respectively, along with a PTV. Beam orientation space was discretized in 10 degrees increments. For each beam orientation, a single-beam plan without intensity modulation and without constraints on OAR dose was generated and normalized to yield a mean PTV dose of 2 Gy and the corresponding MOD was calculated. The degree of OAR sparing was related to the average OAR MODs resulting from the beam orientations utilized with improvements of up to 10% at some dose levels. On the other hand, OAR DVHs in the IMRT plans were insensitive to beam numbers (in the 6-9 range) for similar average single-beam MODs. These MOD data were input to an IP optimization process, which then selected specified numbers of beam angles as inputs to a treatment planning system. Our results show that sets of beam angles with lower average single-beam MODs produce IMRT plans with better OAR sparing than manually selected beam angles. To optimize beam orientations, weights were assigned to each OAR following MOD input to the IP which was subsequently solved using the branch-and-cut algorithm. Seven-beam orientations obtained from solving the IP were applied to the test case with four OARs and the resulting plan with a dose prescription of 63 Gy was compared with an equi-spaced beam plan. The IP selected beams produced dose-volume improvements of up to 40% for OARs proximal to the PTV. Further improvement in the DVH can be obtained by increasing the weights assigned to these OARs but at the expense of the remaining OARs.

Dose-volume conundrum for response of prostate cancer to brachytherapy: summary dosimetric measures and their relationship to tumor control probability

PURPOSE

Although it is known that brachytherapy dose distributions are highly heterogeneous, the effect of particular dose distribution patterns on tumor control probability (TCP) is unknown. It is unlikely that clinical results will throw light on the question in the near future, given the long follow-up and detailed dosimetry required for each patient. We used detailed dose distribution data from 50 patients combined with radiobiologic parameters consistent with what is known about TCP curves for prostate cancer to study the changes in TCP that accompany gross dosimetric measures and particular dosing irregularities (e.g., moderate underdosing of large volumes vs. extreme underdosing of small volumes).

METHODS AND MATERIALS

For each of the 50 patients with organ-confined prostate cancer who had undergone 125I prostate implants alone at our clinic, postimplant CT scans were obtained approximately 1 month after implantation. Dose distribution information was obtained from postimplant dosimetry. The percentage of the prostate volume receiving a specified dose was recorded from the respective differential dose-volume histograms in 10-Gy bins. In addition, the percentage of prostate volume underdosed at varying fractions of the prescription dose were determined, as was the minimal prostate dose. The log-normal distributions of the radiobiologic parameters [ln(initial clonogen number), alpha, and alpha/beta] were adjusted so that the predicted population parameters (steepness and location) of the dose-response curves for external beam radiotherapy agreed with the published estimates. The variability in the dose-volume details was increased by scaling the dose distributions by factors ranging from 0.7 to 1.5, thereby simulating, for each of the patients, nine new patients with different total doses but identical relative distributions of the dose over the voxels. Radiobiologic variability between the selected dose distributions was then removed by averaging >50 randomly chosen sets of radiobiologic parameters from the log-normal distributions to estimate the TCP for each of the dose distributions, giving some insight into the TCP variations with conventional dosimetric indexes and different patterns of underdosing.

RESULTS

Using the 450 dose distributions created by expanding the 50-patient data set, the volume of the prostate that was extremely underdosed (between 50% and 70% of the prescription dose) was related to the volume that was moderately underdosed (between 80% and 100% of the prescription dose). We found that the individual TCP is greatly dependent on the inhomogeneous dose distribution and the dosimetric indexes, such as the volume of prostate receiving 100% of the prescribed dose (V100) and the maximal dose received by 90% of the prostate volume (D90), which, by themselves, are not always accurate predictors of control probabilities. In a multivariate analysis of the dependence of TCP on these parameters (V100, D90, minimal dose, and moderately and severely underdosed volumes), only D90 and the minimal dose were statistically significant. Generally speaking, however, a lower minimal dose means a lower TCP.

CONCLUSION

The work described here was an hypothesis-generating study. Our results showed that even if the V100 and D90 are nearly identical for 2 patients, there can be (and frequently are) significant differences in the dose distributions in the subvolumes of the prostate. Under simulated dose-response conditions (i.e., with variations in the dose distribution), the D90 and minimal dose significantly affected the TCP but the V100 and the volumes moderately or severely underdosed did not. In general, one must consider the totality of the dose distribution to evaluate the dosimetric quality of a low-dose-rate prostate implant. TCP is not a monotonic function of extreme or moderate underdosing. In some instances, extreme underdosing of relatively small volumes may result in a greater TCP than moderate underdosing of relatively large volumes and vice versa.