Treatment data and technical process challenges for practical big data efforts in radiation oncology.

The term Big Data has come to encompass a number of concepts and uses within medicine. This paper lays out the relevance and application of large collections of data in the radiation oncology community. We describe the potential importance and uses in clinical practice. The important concepts are then described and how they have been or could be implemented are discussed. Impediments to progress in the collection and use of sufficient quantities of data are also described. Finally, recommendations for how the community can move forward to achieve the potential of big data in radiation oncology are provided.

A stochastic approach to estimate the uncertainty involved in B-spline image registration.

Uncertainties in image registration may be a significant source of errors in anatomy mapping as well as dose accumulation in radiotherapy. It is, therefore, essential to validate the accuracy of image registration. Here, we propose a method to detect areas where mono modal B-spline registration performs well and to distinguish those from areas of the same image, where the registration is likely to be less accurate. It is a stochastic approach to automatically estimate the uncertainty of the resulting displacement vector field. The coefficients resulting from the B-spline registration are subject to moderate and randomly performed variations. A quantity is proposed to characterize the local sensitivity of the similarity measure to these variations. We demonstrate the statistical dependence between the local image registration error and this quantity by calculating their mutual information. We show the significance of the statistical dependence with an approach based on random redistributions. The proposed method has the potential to divide an image into subregions which differ in the magnitude of their average registration error.

Image registration and data fusion in radiation therapy.

This paper provides an overview of image registration and data fusion techniques used in radiation therapy, and examples of their use. They are used at all stages of the patient management process; for initial diagnosis and staging, during treatment planning and delivery, and after therapy to help monitor the patients' response to treatment. Most treatment planning systems now support some form of interactive or automated image registration and provide tools for mapping information, such as tissue outlines and computed dose from one imaging study to another. To complement this, modern treatment delivery systems offer means for acquiring and registering 2D and 3D image data at the treatment unit to aid patient setup. Techniques for adapting and customizing treatments during the course of therapy using 3D and 4D anatomic and functional imaging data are currently being introduced into the clinic. These techniques require sophisticated image registration and data fusion technology to accumulate properly the delivered dose and to analyse possible physiological and anatomical changes during treatment. Finally, the correlation of radiological changes after therapy with delivered dose also requires the use of image registration and fusion techniques.

Inverse plan optimization accounting for random geometric uncertainties with a multiple instance geometry approximation (MIGA).

Radiotherapy treatment plans that are optimized to be highly conformal based on a static patient geometry can be degraded by setup errors and/or intratreatment motion, particularly for IMRT plans. To achieve improved plans in the face of geometrical uncertainties, direct simulation of multiple instances of the patient anatomy (to account for setup and/or motion uncertainties) is used within the inverse planning process. This multiple instance geometry approximation (MIGA) method uses two or more instances of the patient anatomy and optimizes a single beam arrangement for all instances concurrently. Each anatomical instance can represent expected extremes or a weighted distribution of geometries. The current implementation supports mapping between instances that include distortions, but this report is limited to the use of rigid body translations/ rotations. For inverse planning, the method uses beamlet dose calculations for each instance, with the resulting doses combined using a weighted sum of the results for the multiple instances. Beamlet intensities are then optimized using the inverse planning system based on the cost for the composite dose distribution. MIGA can simulate various types of geometrical uncertainties, including random setup error and intratreatment motion. A limited number of instances are necessary to simulate Gaussian-distributed errors. IMRT plans optimized using MIGA show significantly less degradation in the face of geometrical errors, and are robust to the expected (simulated) motions. Results for a complex head/neck plan involving multiple target volumes and numerous normal structures are significantly improved when the MIGA method of inverse planning is used. Inverse planning using MIGA can lead to significant improvements over the use of simple PTV volume expansions for inclusion of geometrical uncertainties into inverse planning, since it can account for the correlated motions of the entire anatomical representation. The optimized plan results reflect the differing patient geometry situations which can be important near the surface or heterogeneities. For certain clinical situations, the MIGA optimization approach can correct for a significant part of the degradation of the plan caused by the setup uncertainties.

Automated generation of a four-dimensional model of the liver using warping and mutual information.

The use of mutual information (MI) based alignment to map changes in liver shape and position from exhale to inhale was investigated. Inhale and exhale CT scans were obtained with intravenous contrast for six patients. MI based alignment using thin-plate spine (TPS) warping was performed between each inhale and exhale image set. An expert radiation oncologist identified corresponding vessel bifurcations on the exhale and inhale CT image and the transformation for identified points was determined. This transformation was then used to determine the accuracy of the MI based alignment. The reproducibility of the vessel bifurcation identification was measured through repeat blinded vessel bifurcation identification. Reproducibility [standard deviation (SD)] in the L/R, A/P, and I/S directions was 0.11, 0.09, and 0.14 cm, respectively. The average absolute difference between the transformation obtained using MI based alignment and the vessel bifurcation in the L/R, A/P, and I/S directions was 0.13 cm (SD=0.10 cm), 0.15 cm (SD=0.12 cm), and 0.15 cm (SD-0.14 cm), respectively. These values are comparable to the reproducibility of bifurcation identification, indicating that MI based alignment using TPS warping is accurate to within measurement error and is a reliable tool to aid in describing deformation that the liver undergoes from the exhale to inhale state.

Is uniform target dose possible in IMRT plans in the head and neck?

PURPOSE: Various published reports involving intensity-modulated radiotherapy (IMRT) plans developed using automated optimization (inverse planning) have demonstrated highly conformal plans. These reported conformal IMRT plans involve significant target dose inhomogeneity, including both overdosage and underdosage within the target volume. In this study, we demonstrate the development of optimized beamlet IMRT plans that satisfy rigorous dose homogeneity requirements for all target volumes (e.g., +/-5%), while also sparing the parotids and other normal structures. METHODS AND MATERIALS: The treatment plans of 15 patients with oropharyngeal cancer who were previously treated with forward-planned multisegmental IMRT were planned again using an automated optimization system developed in-house. The optimization system allows for variable sized beamlets computed using a three-dimensional convolution/superposition dose calculation and flexible cost functions derived from combinations of clinically relevant factors (costlets) that can include dose, dose-volume, and biologic model-based costlets. The current study compared optimized IMRT plans designed to treat the various planning target volumes to doses of 66, 60, and 54 Gy with varying target dose homogeneity while using a flexible optimization cost function to minimize the dose to the parotids, spinal cord, oral cavity, brainstem, submandibular nodes, and other structures. RESULTS: In all cases, target dose uniformity was achieved through steeply varying dose-based costs. Differences in clinical plan evaluation metrics were evaluated for individual cases (eight different target homogeneity costlets), and for the entire cohort of plans. Highly conformal plans were achieved, with significant sparing of both the contralateral and ipsilateral parotid glands. As the homogeneity of the target dose distributions was allowed to decrease, increased sparing of the parotids (and other normal tissues) may be achieved. However, it was shown that relatively few patients would benefit from the use of increased target inhomogeneity, because the range of improvement in the parotid dose is relatively limited. Hot spots in the target volumes are shown to be unnecessary and do not assist in normal tissue sparing. CONCLUSION: Sparing of both parotids in patients receiving bilateral neck radiation can be achieved without compromising strict target dose homogeneity criteria. The geometry of the normal tissue and target anatomy are shown to be the major factor necessary to predict the parotid sparing that will be possible for any particular case.

Planning, computer optimization, and dosimetric verification of a segmented irradiation technique for prostate cancer.

PURPOSE: To develop and verify a multisegment technique for prostate irradiation that results in better sparing of the rectal wall compared to a conventional three-field technique, for patients with a concave-shaped planning target volume (PTV) overlapping the rectal wall. METHODS AND MATERIALS: Five patients have been selected with various degrees of overlap between PTV and rectal wall. The planned dose to the ICRU reference point is 78 Gy. The new technique consists of five beams, each having an open segment covering the entire PTV and several smaller segments in which the rectum is shielded. Segment weights are computer-optimized using an algorithm based on simulated annealing. The score function to be minimized consists of dose-volume constraints for PTV, rectal wall, and femoral heads. The resulting dose distribution is verified for each patient by using point measurements and line scans made with an ionization chamber in a water tank and by using film in a cylindrical polystyrene phantom. RESULTS: The final number of segments in the five-field technique ranges from 7 to 9 after optimization. Compared to the standard three-field technique, the maximum dose to the rectal wall decreases by approximately 3 Gy for patients with a large overlap and 1 Gy for patients with no overlap, resulting in a reduction of the normal tissue complication probability (NTCP) by a factor of 1.3 and 1.2, respectively. The mean dose to the PTV is the same for the two techniques, but the dose distribution is slightly less homogeneous with the five-field technique (Average standard deviation of five patients is 1.1 Gy and 1.7 Gy for the three-field and five-field technique, respectively). Ionization chamber measurements show that in the PTV, the calculated dose is in general within 1% of the measured dose. Outside the PTV, systematic dose deviations of up to 3% exist. Film measurements show that for the complete treatment, the position of the isodose lines in sagittal and coronal planes is calculated fairly accurately, the maximum distance between measured and calculated isodoses being 4 mm. CONCLUSIONS: We developed a relatively simple multisegment "step-and-shoot" technique that can be delivered within an acceptable time frame at the treatment machine (Extra time needed is approximately 3 minutes). The technique results in better sparing of the rectal wall compared to the conventional three-field technique. The technique can be planned and optimized relatively easily using automated procedures and a predefined score function. Dose calculation is accurate and can be verified for each patient individually.

Estimation of tumor control probability model parameters from 3-D dose distributions of non-small cell lung cancer patients.

Tumor control probability (TCP) model calculations may be used in a relative manner to evaluate and optimize three-dimensional (3-D) treatment plans. Using a mathematical model which makes a number of simplistic assumptions, TCPs can be estimated from a 3-D dose distribution of the tumor given the dose required for a 50% probability of tumor control (D50) and the normalized slope (gamma) of the sigmoid-shaped dose-response curve at D50. The purpose of this work was to derive D50 and gamma from our clinical experience using 3-D treatment planning to treat non-small cell lung cancer (NSCLC) patients. Our results suggest that for NSCLC patients, the dose to achieve significant probability of tumor control may be large (on the order of 84 Gy) for longer (> 30 months) local progression-free survival.

Optimization and clinical use of multisegment intensity-modulated radiation therapy for high-dose conformal therapy.

Intensity-modulated radiation therapy (IMRT) may be performed with many different treatment delivery techniques. This article summarizes the clinical use and optimization of multisegment IMRT plans that have been used to treat more than 350 patients with IMRT over the last 4.5 years. More than 475 separate clinical IMRT plans are reviewed, including treatments of brain, head and neck, thorax, breast and chest wall, abdomen, pelvis, prostate, and other sites. Clinical planning, plan optimization, and treatment delivery are summarized, including efforts to minimize the number of additional intensity-modulated segments needed for particular planning protocols. Interactive and automated optimization of segmental and full IMRT approaches are illustrated, and automation of the segmental IMRT planning process is discussed.

Metastatic carcinoma of the prostate mimicking primary carcinoid tumor of the lung and mediastinum.

Although prostatic carcinomas rarely present as intrathoracic metastases, they may occasionally exhibit clinical and radiographic findings suggestive of a primary pulmonary carcinoid, particularly when they have a cribriform pattern. This report describes three patients who presented with lung and mediastinal neoplasms initially diagnosed as primary carcinoid tumors. These tumors were later proven to be metastatic prostate carcinoma by the use of immunohistochemical studies, including stains for chromogranin, carcinoembryogenic antigen and prostate specific antigen. These findings emphasize the importance of considering metastatic prostate adenocarcinoma in the differential diagnosis of carcinoid or neuroendocrine carcinoma with a cribriform pattern.

The impact of treatment complexity and computer-control delivery technology on treatment delivery errors.

PURPOSE: To analyze treatment delivery errors for three-dimensional (3D) conformal therapy performed at various levels of treatment delivery automation and complexity, ranging from manual field setup to virtually complete computer-controlled treatment delivery using a computer-controlled conformal radiotherapy system (CCRS). METHODS AND MATERIALS: All treatment delivery errors which occurred in our department during a 15-month period were analyzed. Approximately 34,000 treatment sessions (114,000 individual treatment segments [ports]) on four treatment machines were studied. All treatment delivery errors logged by treatment therapists or quality assurance reviews (152 in all) were analyzed. Machines "M1" and "M2" were operated in a standard manual setup mode, with no record and verify system (R/V). MLC machines "M3" and "M4" treated patients under the control of the CCRS system, which (1) downloads the treatment delivery plan from the planning system; (2) performs some (or all) of the machine set up and treatment delivery for each field; (3) monitors treatment delivery; (4) records all treatment parameters; and (5) notes exceptions to the electronically-prescribed plan. Complete external computer control is not available on M3; therefore, it uses as many CCRS features as possible, while M4 operates completely under CCRS control and performs semi-automated and automated multi-segment intensity modulated treatments. Analysis of treatment complexity was based on numbers of fields, individual segments, nonaxial and noncoplanar plans, multisegment intensity modulation, and pseudoisocentric treatments studied for a 6-month period (505 patients) concurrent with the period in which the delivery errors were obtained. Treatment delivery time was obtained from the computerized scheduling system (for manual treatments) or from CCRS system logs. Treatment therapists rotate among the machines; therefore, this analysis does not depend on fixed therapist staff on particular machines. RESULTS: The overall reported error rate (all treatments, machines) was 0.13% per segment, or 0.44% per treatment session. The rate (per machine) depended on automation and plan complexity. The error rates per segment for machines M1 through M4 were 0.16%, 0.27%, 0.12%, 0.05%, respectively, while plan complexity increased from M1 up to machine M4. Machine M4 (the most complex plans and automation) had the lowest error rate. The error rate decreased with increasing automation in spite of increasing plan complexity, while for the manual machines, the error rate increased with complexity. Note that the real error rates on the two manual machines are likely to be higher than shown here (due to unnoticed and/or unreported errors), while (particularly on M4) virtually all random treatment delivery errors were noted by the CCRS system and related QA checks (including routine checks of machine and table readouts for each treatment). Treatment delivery times averaged from 14 min to 23 min per plan, and depended on the number of segments/plan, although this analysis is complicated by other factors. CONCLUSION: Use of a sophisticated computer-controlled delivery system for routine patient treatments with complex 3D conformal plans has led to a decrease in treatment delivery errors, while at the same time allowing delivery of increasingly complex and sophisticated conformal plans with little increase in treatment time. With renewed vigilance for the possibility of systematic problems, it is clear that use of complete and integrated computer-controlled delivery systems can provide improvements in treatment delivery, since more complex plans can be delivered with fewer errors, and without increasing treatment time.

Fraction size and dose parameters related to the incidence of pericardial effusions.

PURPOSE: The influence of treatment parameters, such as (a) fraction size and (b) average and maximum dose (as derived from three-dimensional (3D) distributions), on the incidence of pericarditis was analyzed. To understand and predict the dose and volume effect on the pericardium, a normal tissue-complication probability model was tested with these complication data. METHODS AND MATERIALS: Patients (n = 57) entered in 3 consecutive University of Michigan protocols of combined modality for treatment of localized esophageal carcinoma, and having 3D treatment planning for radiation therapy were the subject of this study. Univariate and multivariate analyses were performed to determine the significance of the effect of fraction size and dose parameters on the development of any grade of pericarditis. Dose distributions were corrected for the biological effect of fraction size using the linear-quadratic method. Normal tissue complication probability (NTCP) was calculated with the Lyman model. RESULTS: Nonmalignant pericardial effusions occurred in 5 of the 57 patients; all effusions were in patients who received treatment with 3.5 Gy daily fractions. On multivariate analysis, no dose factor except fraction size predicted pericarditis, until the dose distributions were corrected for the effect of fraction size ("bio"-dose). Then, both "bio-average" and "bio-maximum" dose were significant predictive factors (p = 0.014). NTCPs for the patients with pericarditis range from 62% to 99% for the calculations with the "bio"-dose distributions vs. 0.5% to 27% for the uncorrected distributions. DISCUSSION: A normal tissue complication probability (NTCP) model predicts a trend towards a high incidence of radiation pericarditis for patients who have high complication probabilities. It is important to correct the dose distribution for the effects of fractionation, particularly when the fraction size deviates greatly from standard (2.0 Gy) fractionation.

A phase I trial of hepatic arterial bromodeoxyuridine and conformal radiation therapy for patients with primary hepatobiliary cancers or colorectal liver metastases.

PURPOSE: We have previously found that conformal radiation therapy (RT) and hepatic arterial fluorodeoxyuridine was associated with durable responses and long-term survival for patients treated for nondiffuse primary hepatobiliary tumors and colorectal liver metastases. Further improvements in hepatic control may result from the addition of selective radiosensitization using bromodeoxyuridine (BrdU) infused through the hepatic artery (HA) concurrently with RT. This is a Phase I study of escalating doses of HA BrdU combined with our standard hepatic RT. METHODS AND MATERIALS: Patients with unresectable primary hepatobiliary cancer or colorectal liver metastases were treated with concurrent HA BrdU and conformal RT (1.5 Gy per fraction, twice a day). Three-dimensional treatment planning was used to define both the target and normal liver volumes. The total dose of RT (24, 48, or 66 Gy) was determined by the fractional volume of normal liver excluded from the high dose volume. HA BrdU was escalated in standard Phase I fashion with at least three patients receiving each combination of RT dose and BrdU dose. The starting dose of HA BrdU was 10 mg/kg/day, with two potential escalations to a maximum of 25 mg/kg/day (the maximum tolerable dose of HA BrdU when given alone on this same schedule). Grade > or = 3 toxicity was considered dose limiting. Patients receiving 24 Gy had one cycle of HA BrdU, while those receiving either 48 or 66 Gy had two cycles. Patients were followed for toxicity, complications, and response (when evaluable). RESULTS: A total of 41 patients (18 with colorectal liver metastases, 16 with cholangiocarcinoma and 7 with hepatoma) were treated. Five patients were removed from the protocol (three had HA catheter complications, one developed atrial fibrillation, and one was removed due to recurrent Grade 4 toxicity), although all five are included for toxicity purposes. Dose-limiting toxicity was primarily thrombocytopenia and there was no obvious relationship with the RT dose. Only 2 of 17 cycles given at 25 mg/kg/day had Grade > or = 3 toxicity. Complications developed in four patients, including one patient with radiation-induced liver disease. Response rates were not improved compared to our previous experience. CONCLUSIONS: The appropriate dose of HA BrdU for Phase II evaluation is 25 mg/kg/day. Neither the hepatic parenchyma nor the gastrointestinal mucosa appeared to be sensitized by this method of BrdU administration. It is anticipated that these, or still newer methods of therapy, can improve treatment results in the near future.

Long-term results of hepatic artery fluorodeoxyuridine and conformal radiation therapy for primary hepatobiliary cancers.

PURPOSE: We have previously shown that conformal radiation therapy (RT) combined with hepatic artery (HA) fluorodeoxyuridine (FdUrd) had encouraging hepatic control and survival rates for patients with nondiffuse primary hepatobiliary malignancies. With longer follow-up, we were particularly interested if long-term hepatic control and disease-free survival could be achieved, and if late hepatic complications due to radiation therapy were observed. METHODS AND MATERIALS: Patients with unresectable primary hepatobiliary cancer were treated with concurrent HA FdUrd (0.2 mg/kg/day) and conformal RT (1.5-1.65 Gy per fraction, twice a day), directed only to the liver abnormalities. Three-dimensional treatment planning was used to define both the target and normal liver volumes. The total dose of radiation (48 or 66 Gy) was determined by the fractional volume of normal liver excluded from the high dose volume. Patients were followed routinely for response, patterns of failure, long-term toxicity, and survival. The median potential follow-up was 54 months. RESULTS: A total of 22 patients (11 with hepatocellular carcinoma and 11 with cholangiocarcinoma) were treated. There were 10 objective responses in the 11 evaluable patients. The overall freedom from hepatic progression at more than 2 years was about 50%. The median survival was 16 months with an actuarial 4-year survival of about 20%. Gastrointestinal bleeding was the most common long-term toxicity. Late hepatic toxicity was not observed; in fact, hypertrophy of the untreated liver was seen. CONCLUSIONS: Combined conformal RT and HA FdUrd can produce long-term freedom from hepatic progression and survival in patients with unresectable, nondiffuse primary hepatobiliary malignancies. There were no long-term liver complications observed.

Optimal placement of radioisotopes for permanent prostate implants.

PURPOSE: To determine which of four loading techniques most efficiently yields the prescribed dose to the prostate volume while limiting dose to the central urethral volume. MATERIALS AND METHODS: The four techniques included (a) equal activity and equal spacing with nomogram, (b) differential loading, (c) peripheral loading, and (d) spiked loading of the lobes. They were evaluated with regard to target coverage urethra dose, tolerance to error, and complexity of procedure. RESULTS: All ideal plans delivered the prescribed dose of 160 Gy to 99% of the prostate volume. With prostate-volume expansion and source-placement errors, all strategies indicated that at least 71% of the target volume received the prescribed dose and greater than 92% of the target volume received 120 Gy. CONCLUSION: With source-placement errors and glandular swelling, peripheral loading yields the best target coverage while limiting dose to the central urethral volume.

Advanced interactive planning techniques for conformal therapy: high level beam descriptions and volumetric mapping techniques.

PURPOSE: To aid in design of conformal radiation therapy treatment plans involving many conformally shaped fields, this work investigates the use of two methodologies to enhance the ease of interactive treatment planning: high-level beam constructs and beam's-eye view volumetric mapping. METHODS AND MATERIALS: High-performance computer graphics running on various workstations using a graphical visualization system (AVS) have been used in this work. Software specific to this application has been written in standard FORTRAN and C languages. A new methodology is introduced by defining radiation therapy "fields" to be composed of multiple beam "segments." Fields can then be defined as higher-level entities such as arcs, cones, and other shapes. A "segmental cone" field, for example, is defined by a symmetry axis and a cone angle, and can be used to rapidly place a series of beam segments that converge at the target volume, while reducing the degree of overlap elsewhere. A new beam's-eye view (BEV) volumetric mapping technique is presented to aid in selecting the placement of conformal radiation fields. With this technique, the relative average dose within an organ of interest is calculated for a sampling of isocentric, conformally shaped beams and displayed either as a "globe," which can be combined with the display of anatomical surfaces, or as a two-dimensionally mapped projection. The dose maps from multiple organs can be generated, stacked, or composited with relative weightings to aid in the placement of fields that minimize overlap with critical structures. RESULTS: The use of these new methodologies is demonstrated for prostate and lung treatment sites and compared to conventional planning techniques. DISCUSSION: The use of many beams for conformal treatment delivery is difficult with current interactive planning. The use of high-level beam constructs provides a means to quickly specify, place, and configure multiple beam arrangements. The BEV volumetrics aids in the placing of fields, which minimize involvement with critical normal tissues. CONCLUSIONS: Early experience with the new methodologies suggest that the new methods help to enhance (or at least speed up) the ability of a treatment planner to create optimal radiation treatment field arrangements.

A computer-controlled conformal radiotherapy system. I: Overview.

PURPOSE: Equipment developed for use with computer-controlled conformal radiotherapy (CCRT) treatment techniques, including multileaf collimators and/or computer-control systems for treatment machines, are now available. The purpose of this work is to develop a system that will allow the safe, efficient, and accurate delivery of CCRT treatments as routine clinical treatments, and permit modifications of the system so that the delivery process can be optimized. METHODS AND MATERIALS: The needs and requirements for a system that can fully support modern computer-controlled treatment machines equipped with multileaf collimators and segmental or dynamic conformal therapy capabilities have been analyzed and evaluated. This analysis has been used to design and then implement a complete approach to the delivery of CCRT treatments. RESULTS: The computer-controlled conformal radiotherapy system (CCRS) described here consists of a process for the delivery of CCRT treatments, and a complex software system that implements the treatment process. The CCRS system described here includes systems for plan transfer, treatment delivery planning, sequencing of the actual treatment delivery process, graphical simulation and verification tools, as well as an electronic chart that is an integral part of the system. The CCRS system has been implemented for use with a number of different treatment machines. The system has been used clinically for more than 2 years to perform CCRT treatments for more than 200 patients. CONCLUSIONS: A comprehensive system for the implementation and delivery of computer-controlled conformal radiation therapy (CCRT) plans has been designed and implemented for routine clinical use with multisegment, computer-controlled, multileaf-collimated conformal therapy. The CCRS system has been successfully implemented to perform these complex treatments, and is considered quite important to the clinical use of modern computer-controlled treatment techniques.

A computer-controlled conformal radiotherapy system. II: Sequence processor.

PURPOSE: A sequence processor (SP) is described as part of a larger computer-controlled conformal radiotherapy system (CCRS). The SP provides the means to accept and then translate highly sophisticated radiation therapy treatment plans into vendor specific instructions to control treatment delivery on a computer-controlled treatment machine. METHODS AND MATERIALS: The sequence processor (SP) is a small workstation computer that interfaces to the control computer of computer-controlled treatment machines, and to other parts of the larger CCRS system. The system reported here has been interfaced to a computer-controlled racetrack microtron with two treatment gantries, and also to other linear accelerator treatment machines equipped with multileaf collimators. An extensive design process has been used in defining the role of the SP within the context of the larger CCRS project. Flexibility and integration with various components of the project, including databases, treatment planning system, graphical simulator, were key factors in the development. In conjunction with the planned set of treatment fields, a procedural scripting language is used to define the sequence of treatment events that are performed, including operator interactions, communications to other systems such as dosimetry and portal imaging devices, and database management. RESULTS: A flexible system has been developed to allow investigation into procedural steps required for simulating and delivering complex radiation treatments. The system has been used to automate portions of the acceptance testing for the control system of the microtron, and is used for routine daily quality assurance testing. The sequence processor system described here has been used to deliver all clinical treatments performed on the microtron system in 2 years of clinical treatment (more than 200 patients treated to a variety of treatment sites). CONCLUSIONS: The sequence processor system has enabled the delivery of complex treatment using computer-controlled treatment machines. The flexibility of the system allows integration with secondary devices and modification of procedural steps, making it possible to develop effective techniques for insuring safe and efficient computer-controlled conformal radiation therapy treatments.