Mark Langer

Improved leaf sequencing reduces segments or monitor units needed to deliver IMRT using multileaf collimators

Leaf sequencing algorithms may use an unnecessary number of monitor units or segments to generate intensity maps for delivery of intensity modulated radiotherapy (IMRT) using multiple static fields. An integer algorithm was devised to generate a sequence with the fewest possible segments when the minimum number of monitor units are used. Special hardware related restrictions on leaf motion can be incorporated. The algorithm was tested using a benchmark map from the literature and clinical examples. Results were compared to sequences given by the routine of Bortfeld that minimizes monitor units by treating each row independently, and the areal or reducing routines that use fewer segments at the price of more monitor units. The Bortfeld algorithm used on average 58% more segments than provided by the integer algorithm with bidirectional motion and 32% more segments than did an integer algorithm admitting only unidirectional sequences. The areal algorithm used 48% more monitor units and the reducing algorithm used 23% more monitor units than did the bidirectional integer algorithm, while the areal and reducing algorithms used 23% more segments than did the integer algorithm. Improved leaf sequencing algorithms can allow more efficient delivery of static field IMRT. The integer algorithm demonstrates the efficiencies possible with an improved routine and opens a new avenue for development.

A coupled column generation, mixed integer approach to optimal planning of intensity modulated radiation therapy for cancer

Abstract.Approximately 40% of all U.S. cancer cases are treated with radiation therapy. In Intensity-Modulated Radiation Therapy (IMRT) the treatment planning problem is to choose external beam angles and their corresponding intensity maps (showing how the intensity varies across a given beam) to maximize tumor dose subject to the tolerances of surrounding healthy tissues. Dose, like temperature, is a quantity defined at each point in the body, and the distribution of dose is determined by the choice of treatment parameters available to the planner. In addition to absolute dose limits in healthy tissues, some tissues have at least one dose-volume restriction that requires a fraction of its volume to not exceed a specified tighter threshold level for damage. There may also be a homogeneity limit for the tumor that restricts the allowed spread of dose across its volume. We formulate this planning problem as a mixed integer program over a coupled pair of column generation processes -- one designed to produce intensity maps, and a second specifying protected area choices for tissues under dose-volume restrictions. The combined procedure is shown to strike a balance between computing an approximately optimal solution and bounding its maximum possible suboptimality that we believe holds promise for implementations able to offer the power and flexibility of mixed-integer linear programming models on instances of practical scale.

Methodological issues in radiation dose–volume outcome analyses: Summary of a joint AAPM/NIH workshop

This report represents a summary of presentations at a joint workshop of the National Institutes of Health and the American Association of Physicists in Medicine (AAPM). Current methodological issues in dose-volume modeling are addressed here from several different perspectives. Areas of emphasis include (a) basic modeling issues including the equivalent uniform dose framework and the bootstrap method, (b) issues in the valid use of statistics, including the need for meta-analysis, (c) issues in dealing with organ deformation and its effects on treatment response, (d) evidence for volume effects for rectal complications, (e) the use of volume effect data in liver and lung as a basis for dose escalation studies, and (f) implications of uncertainties in volume effect knowledge on optimized treatment planning. Taken together, these approaches to studying volume effects describe many implications for the development and use of this information in radiation oncology practice. Areas of significant interest for further research include the meta-analysis of clinical data; interinstitutional pooled data analyses of volume effects; analyses of the uncertainties in outcome prediction models, minimal parameter number outcome models for ranking treatment plans (e.g., equivalent uniform dose); incorporation of the effect of motion in the outcome prediction; dose-escalation/isorisk protocols based on outcome models; the use of functional imaging to study radioresponse; and the need for further small animal tumor control probability/normal tissue complication probability studies.

Column generation for IMRT cancer therapy optimization with implementable segments

A radiation beam passes through normal tissue to reach tumor. The latest devices for the radiotherapy of cancer provide intensity modulated radiation treatment, or IMRT. This method refines cancer treatment by varying the intensity profile across the face of a radiation beam. Intensity modulation is usually accomplished by partitioning each beam, distinguished by its angle of entry, into an array of smaller sized units, called beamlets, assigned different intensities. Planning treatment calls for an optimization over beamlet intensities to maximize the dose delivered to the targeted tumor while keeping the distribution of dose throughout the various organs within physician prescribed bounds. The choice of beam angles can be entered into the optimization as well.A common method to produce an intensity pattern is to block out different parts of the beam for different amounts of time. This can be done sliding narrow blocks (leafs) of unit width into the beam from either of two opposing sides to create different beam shapes called segments. A sequence of segments with their exposure times is superimposed to yield the dose distribution actually received in the patient. Current two stage treatment is derived in separate steps: optimization over independently considered beamlet intensities, and generation of a sequence of segments to approximate the planned intensity map. The approximation degrades the solution, and the separate search for segments adds to planning time. We present a mixed integer programming alternative employing column generation to optimize dose over segments themselves. Only segments that can be realized with delivery devices are generated, and adjustments made for the effects of block edges, so that the optimized plans are directly implementable. Preliminary testing demonstrates gains in both planning efficiency and quality of the plans produced.

Survey: Real-Time Tumor Motion Prediction for Image-Guided Radiation Treatment

Tumor motion caused by patient breathing creates challenges for accurate radiation dose delivery to a tumor while sparing healthy tissues. Image-guided radiation therapy (IGRT) helps, but theres a lag time between tumor position acquisition and dose delivered to that position. An efficient and accurate predictive model is thus an essential requirement for IGRT success.

On probabilistically defined margins in radiation therapy

Margins about a target volume subject to external beam radiation therapy are designed to assure that the target volume of tissue to be sterilized by treatment is adequately covered by a lethal dose. Thus, margins are meant to guarantee that all potential variation in tumour position relative to beams allows the tumour to stay within the margin. Variation in tumour position can be broken into two types of dislocations, reducible and irreducible. Reducible variations in tumour position are those that can be accommodated with the use of modern image-guided techniques that derive parameters for compensating motions of patient bodies and/or motions of beams relative to patient bodies. Irreducible variations in tumour position are those random dislocations of a target that are related to errors intrinsic in the design and performance limitations of the software and hardware, as well as limitations of human perception and decision making. Thus, margins in the era of image-guided treatments will need to accommodate only random errors residual in patient setup accuracy (after image-guided setup corrections) and in the accuracy of systems designed to track moving and deforming tissues of the targeted regions of the patients body. Therefore, construction of these margins will have to be based on purely statistical data. The characteristics of these data have to be determined through the central limit theorem and Gaussian properties of limiting error distributions. In this paper, we show how statistically determined margins are to be designed in the general case of correlated distributions of position errors in three-dimensional space. In particular, we show how the minimal margins for a given level of statistical confidence are found. Then, how they are to be used to determine geometrically minimal PTV that provides coverage of GTV at the assumed level of statistical confidence. Our results generalize earlier recommendations for statistical, central limit theorem-based recommendations for margin construction that were derived for uncorrelated distributions of errors (van Herk, Remeijer, Rasch and Lebesque 2000 Int. J. Radiat. Oncol. Biol. Phys. 47 1121-35; Stroom, De Boer, Huizenga and Visser 1999 Int. J. Radiat. Oncol. Biol. Phys. 43 905-19).

Ability of the National Surgical Quality Improvement Program Risk Calculator to Predict Complications Following Total Laryngectomy.

Importance The accuracy of the American College of Surgeons National Surgical Quality Improvement Program (NSQIP) risk calculator has been assessed in multiple surgical subspecialties; however, there have been no publications doing the same in the head and neck surgery literature. Objective To evaluate the accuracy of the calculators predictions in a single institutions total laryngectomy (TL) population. Design, Setting, and Participants Total laryngectomies performed between 2013 and 2014 at a tertiary referral academic center were evaluated using the risk calculator. Predicted 30-day outcomes were compared with observed outcomes for return to operating room, surgical site infection, postoperative pneumonia, length of stay, and venous thromboembolism. Main Outcomes and Measures Comparison of the NSQIP risk calculators predicted postoperative complication rates and length of stay to what occurred in this patient cohort using percent error, Brier scores, area under the receiver operating characteristic curve, and Pearson correlation analysis. Results Of 49 patients undergoing TL, the mean (SD) age at operation was 59 (9.3) years, with 67% male. The risk calculator had limited efficacy predicting perioperative complications in this group of patients undergoing TL with or without free tissue reconstruction or preoperative chemoradiation or radiation therapy with a few exceptions. The calculator overestimated the occurrence of pneumonia by 165%, but underestimated surgical site infection by 7%, return to operating room by 24%, and length of stay by 13%. The calculator had good sensitivity and specificity of predicting surgical site infection for patients undergoing TL with free flap reconstruction (area under the curve, 0.83). For all other subgroups, however, the calculator had poor sensitivity and specificity for predicting complications. Conclusions and Relevance The risk calculator has limited utility for predicting perioperative complications in patients undergoing TL. This is likely due to the complexity of the treatment of patients with head and neck cancer and factors not taken into account when calculating a patients risk.

Laser-plasma generated very high energy electrons in radiation therapy of the prostate

Monte Carlo simulation experiments have shown that very high energy electrons (VHEE), 150-250 MeV, have potential advantages in prostate cancer treatment over currently available electrons, photon and proton beam treatment. Small diameter VHEE beamlets can be scanned, thereby producing a finer resolution intensity modulated treatment than photon beams. VHEE beams may be delivered with greater precision and accelerators may be constructed at significantly lower cost than proton beams. A VHEE accelerator may be optimally designed using laser-plasma technology. If the accelerator is constructed to additionally produce low energy photon beams along with VHEE, real time imaging, bioprobing, and dose enhancement may be performed simultaneously. This paper describes a Monte Carlo experiment, using the parameters of the electron beam from the UCLA laser-plasma wakefield accelerator, whereby dose distributions on a human prostate are generated. The resulting dose distributions of the very high energy electrons are shown to be comparable to photon beam dose distributions. This simple experiment illustrates that the nature of the dose distribution of electrons is comparable to that of photons. However, the main advantage of electrons over photons and protons lies in the delivery and manipulation of electrons, rather than the nature of the dose distribution. This paper describes the radiation dose delivery of electrons employing technologies currently in exploration and evaluates potential benefits as compared with currently available photon and protons beams in the treatment of prostate and other cancers, commonly treated with radiation.

State of dose prescription and compliance to international standard (ICRU-83) in intensity modulated radiation therapy among academic institutions

PURPOSE The purpose of this study was to evaluate dose prescription and recording compliance to international standard (International Commission on Radiation Units & Measurements [ICRU]-83) in patients treated with intensity modulated radiation therapy (IMRT) among academic institutions. METHODS AND MATERIALS Ten institutions participated in this study to collect IMRT data to evaluate compliance to ICRU-83. Under institutional review board clearance, data from 5094 patients-including treatment site, technique, planner, physician, prescribed dose, target volume, monitor units, planning system, and dose calculation algorithm-were collected anonymously. The dose-volume histogram of each patient, as well as dose points, doses delivered to 100% (D100), 98% (D98), 95% (D95), 50% (D50), and 2% (D2), of sites was collected and sent to a central location for analysis. Homogeneity index (HI) as a measure of the steepness of target and is a measure of the shape of the dose-volume histogram was calculated for every patient and analyzed. RESULTS In general, ICRU recommendations for naming the target, reporting dose prescription, and achieving desired levels of dose to target were relatively poor. The nomenclature for the target in the dose prescription had large variations, having every permutation of name and number contrary to ICRU recommendations. There was statistically significant variability in D95, D50, and HI among institutions, tumor site, and technique with P values < .01. Nearly 95% of patients had D50 higher than 100% (103.5 ± 6.9) of prescribed dose and varied among institutions. On the other hand, D95 was close to 100% (97.1 ± 9.4) of prescribed dose. Liver and lung sites had a higher D50 compared with other sites. Pelvic sites had a lower variability indicated by HI (0.13 ± 1.21). Variability in D50 is 101.2 ± 8.5, 103.4 ± 6.8, 103.4 ± 8.2, and 109.5 ± 11.5 for IMRT, tomotherapy, volume modulated arc therapy, and stereotactic body radiation therapy with IMRT, respectively. CONCLUSIONS Nearly 95% of patient treatments deviated from the ICRU-83 recommended D50 prescription dose delivery. This variability is significant (P < .01) in terms of treatment site, technique, and institution. To reduce dosimetric and associated radiation outcome variability, dose prescription in every clinical trial should be unified with international guidelines.

Strong valid inequalities for fluence map optimization problem under dose-volume restrictions

Fluence map optimization problems are commonly solved in intensity modulated radiation therapy (IMRT) planning. We show that, when subject to dose-volume restrictions, these problems are NP-hard and that the linear programming relaxation of their natural mixed integer programming formulation can be arbitrarily weak. We then derive strong valid inequalities for fluence map optimization problems under dose-volume restrictions using disjunctive programming theory and show that strengthening mixed integer programming formulations with these valid inequalities has significant computational benefits.