Improving the Quality of Care in Radiation Oncology using Artificial Intelligence.

Radiation therapy is a complex process involving multiple professionals and steps from simulation to treatment planning to delivery, and these procedures are prone to error. Additionally, the imaging and treatment delivery equipment in radiotherapy is highly complex and interconnected and represents another risk point in the quality of care. Numerous quality assurance tasks are carried out to ensure quality and to detect and prevent potential errors in the process of care. Recent developments in artificial intelligence provide potential tools to the radiation oncology community to improve the efficiency and performance of quality assurance efforts. Targets for artificial intelligence enhancement include the quality assurance of treatment plans, target and tissue structure delineation used in the plans, delivery of the plans and the radiotherapy delivery equipment itself. Here we review recent developments of artificial intelligence applications that aim to improve quality assurance processes in radiation therapy and discuss some of the challenges and limitations that require further development work to realise the potential of artificial intelligence for quality assurance.

A feasibility study of spatiotemporally integrated radiotherapy using the LQ model.

This paper investigates the feasibility of spatiotemporally modulated radiotherapy (STMRT)-integrated model with explicit constraints on the tumor dose heterogeneity. In particular, we demonstrate the effect of the tumor dose heterogeneity on the tumor biologically effective dose (BED) achievable and optimal fractionation. We propose an STMRT model that simultaneously optimizes the dose distributions and fractionation schedule for each individual case with the maximum and minimum constraints on the tumor BED to explicitly control the level of tumor dose heterogeneity. Sixteen thoracic phantom cases were planned using (1) STMRT and (2) standard fractionation (60 Gy in 30 fractions fixed) IMRT. Constraints on the organs-at-risk (OAR) BED were identical for both plans. BEDs were calculated using the [Formula: see text] ratio of 10 Gy for the tumor and 3 Gy for all OARs. The maximum tumor BED for STMRT plans was constrained to be less than 100%-150% of the maximum tumor BED resulted from the standard fractionation plans. The mean tumor BED from STMRT plans was up to 110.7%, 128.3%, 135.0% and 148.0% of that from the standard fractionation plans when the maximum tumor BED was constrained to be less than 100%, 120%, 130% and 150% of the maximum BED achieved using the standard plans. The optimal number of fractions varied widely for different phantom geometries for the same radiobiological parameter values. The increase in the tumor BED and the range of optimal fractionation was larger with a larger tumor dose heterogeneity allowed. The results have shown the feasibility of personalizing fractionation schedule using an STMRT integrated model to deliver a maximum feasible BED to the tumor for a fixed OAR BED. The potential increase in the tumor BED was positively correlated to the tumor dose heterogeneity allowed.

A Markov decision process approach to temporal modulation of dose fractions in radiation therapy planning.

The current state of the art in cancer treatment by radiation optimizes beam intensity spatially such that tumors receive high dose radiation whereas damage to nearby healthy tissues is minimized. It is common practice to deliver the radiation over several weeks, where the daily dose is a small constant fraction of the total planned. Such a 'fractionation schedule' is based on traditional models of radiobiological response where normal tissue cells possess the ability to repair sublethal damage done by radiation. This capability is significantly less prominent in tumors. Recent advances in quantitative functional imaging and biological markers are providing new opportunities to measure patient response to radiation over the treatment course. This opens the door for designing fractionation schedules that take into account the patient's cumulative response to radiation up to a particular treatment day in determining the fraction on that day. We propose a novel approach that, for the first time, mathematically explores the benefits of such fractionation schemes. This is achieved by building a stylistic Markov decision process (MDP) model, which incorporates some key features of the problem through intuitive choices of state and action spaces, as well as transition probability and reward functions. The structure of optimal policies for this MDP model is explored through several simple numerical examples.

Automatic selection of non-coplanar beam directions for three-dimensional conformal radiotherapy.

An algorithm is described, based on ray-tracing and the beam's-eye-view, that exhaustively searches all permitted beam directions. The evaluation of the search is based on a general cost function that can be adapted to the clinical objectives by means of parameters and weighting factors. The approach takes into account the constraints of the linear accelerator by discarding beam directions that are not permitted. A sensitivity analysis was carried out to determine appropriate parameters for different sized organs, and a prostate case was used to benchmark the approach. The algorithm was also applied to two clinical cases (brain and sinus) to test the benefits of the approach compared with manual angle selection. The time to perform a beam direction search was approximately 2 min for the coplanar and 12 min for the non-coplanar beam space. The angles obtained for the prostate case compared well with reports in the literature. For the brain case, the mean dose to the right and left optic nerves was reduced by 12% and 50%, respectively, whilst the target dose uniformity was improved. For the sinus case, the mean doses to the right and left parotid glands were reduced by 54% and 46%, respectively, to the right and left optic nerves by 37% and 62%, respectively, and to the optic chiasm by 39%, whilst the target dose uniformity was also improved. For the clinical cases the plans based on optimized beam directions were simpler and resulted in better sparing of critical structures compared with plans based on manual angle selection. The approach provides a practical alternative to elaborate and time consuming beam angle optimization schemes and is suitable for routine clinical usage.

Use of intensity modulation for missing tissue compensation of pediatric spinal fields.

Irradiation of the cranio-spinal axis is often one of the treatment modalities of certain childhood cancers, e.g., medulloblastoma. In order to achieve a uniform dose to the spinal cord, missing tissue compensators are required. In the past, our practice was to fabricate compensators out of strips of lead. We report on the use of intensity modulated fields to achieve the desired compensation. Seven cases of pediatric cancer whose treatment involved irradiation of the cranio-spinal axis had compensators designed using a beam intensity modulation method rather than making mechanical compensators. The compensators only adjusted for missing tissue along the spinal axis. Comparisons between calculated and measured doses were made at depth in phantoms and on the surface of the patient. The intensity modulated fields were delivered using a step-and-shoot delivery on an Elekta SL20 accelerator equipped with multileaf collimator. The intensity-modulated compensators provided more flexibility in design than the physical compensator method. Finer intensity steps were achievable, more accurate dose distributions were able to be calculated, and adjustments during treatment, e.g., junction changes, were more easily implemented. Convolution/superposition dose calculations were within +/-3% of measurements. Intensity modulated fields are a practical and more efficient method of delivering uniform doses to the spine in pediatric cancer treatments. They provide many advantages over mechanical compensators with regard to time and flexibility.

Reduction of computational dimensionality in inverse radiotherapy planning using sparse matrix operations.

For dynamic multileaf collimator-based intensity modulated radiotherapy in which small beam elements are used to generate continuous modulation, the sheer size of the dose calculation matrix could pose serious computational challenges. In order to circumvent this problem, the dose calculation matrix was reduced to a sparse matrix by truncating the weakly contributing entries below a certain cutoff to zero. Subsequently, the sparse matrix was compressed and matrix indexing vectors were generated to facilitate matrix-vector and matrix-matrix operations used in inverse planning. The application of sparsity permitted the reduction of overall memory requirement by an order of magnitude. In addition, the effect of disregarding the small scatter components on the quality of optimization was investigated by repeating the inverse planning using the dense dose calculation matrix. Comparison of dense and sparse matrix-based plans revealed an insignificant difference in optimization outcome, thus demonstrating the feasibility and usefulness of the sparse method in inverse planning. Furthermore, two additional methods of memory minimization are suggested, namely hexagonal dose sampling and limited normal tissue sampling.

Verification of dynamic intensity-modulated beam deliveries in canine subjects.

Our objective in this work was to assess the precision and degree of accuracy with which intensity modulated radiation therapy (IMRT) can deliver highly localized dose distributions to tumors near critical structures using the dynamic sliding window technique. Measurements of dose distribution were performed both in vivo and in vitro using a combination of dosimeters [thermoluminescent dosimeters (TLD's), films, and diodes]. In vivo measurements were performed in two groups of purpose-bred dogs: one receiving four-field three-dimensional (3D) conformal treatment and the other receiving IMRT. The algorithms used in the inverse planning process included the Macro Pencil Beam (MPB) model and Projections onto Convex Sets (POCS). Single beam measurements were performed in phantoms to verify the accuracy of monitor unit settings required for delivering the desired doses. The composite doses from the delivery of the seven beam intensity modulated plans were measured in phantoms and cadavers, Biological end points (spinal cord toxicity and neurologic deficits due to irradiation) were evaluated at the end of one year to determine the spatial accuracy of the IMRT treatments over a fractionated course in live subjects. Results in single beam measurements were used at first to improve the dose calculation and translation algorithms. Results of the measurements for the delivery of all seven beams in phantoms confirmed that the system was capable of accurate spatial and dosimetric IMRT delivery. The in vivo results showed dramatic differences between control and IMRT-treated dogs, with the IMRT group showing no adverse effects and the control animals showing severe spinal cord injuries due to irradiation. The measurements presented in this paper have helped to verify the successful and accurate delivery of IMRT in a clinically related model using the University of Washington Medical Center (UWMC) system.

Dynamic and omni wedge implementation on an Elekta SL linac.

The concommitant use of a multileaf collimator (MLC) and a wedge can result in conflicts in the optimal collimator angle if both MLC and wedge are fixed relative to one another. This is particularly true of linacs in which a single wedge orientation is provided. In this paper, a solution is provided that makes use of two orthogonal universal wedges (omni wedge). Although this technique can be applied regardless of the means by which the wedged fields are implemented, the measurements reported in this paper were performed using a fixed, internal mechanical wedge coupled with a dynamic wedge, formed by the motion of one of the backup jaws. An implementation of a dynamic wedge for the Elekta SL series of linear accelerators is presented. Results of measurements of the dosimetric characteristics of both the particular implementation of the dynamic wedge and of the omni field are presented. For the dynamic wedge, measurements were made of the wedge factor and dose profile as a function of field size and depth. In addition, the effects of variables, such as dynamic delivery technique and direction of diaphragm motion, on the dynamic wedge profiles were studied and discussed. For the omni wedge, measurements were made of the degree to which the mathematical formalism for describing an omni wedge matches the measured isodose distributions. Comparisons between mechanical wedge dose distributions and the omni wedge were also made.

Commissioning an image-guided localization system for radiotherapy.

PURPOSE: To describe the design and commissioning of a system for the treatment of classes of tumors that require highly accurate target localization during a course of fractionated external-beam therapy. This system uses image-guided localization techniques in the linac vault to position patients being treated for cranial tumors using stereotactic radiotherapy, conformal radiotherapy, and intensity-modulated radiation therapy techniques. Design constraints included flexibility in the use of treatment-planning software, accuracy and precision of repeat localization, limits on the time and human resources needed to use the system, and ease of use. METHODS AND MATERIALS: A commercially marketed, stereotactic radiotherapy system, based on a system designed at the University of Florida, Gainesville, was adapted for use at the University of Washington Medical Center. A stereo pair of cameras in the linac vault were used to detect the position and orientation of an array of fiducial markers that are attached to a patient's biteblock. The system was modified to allow the use of either a treatment-planning system designed for stereotactic treatments, or a general, three-dimensional radiation therapy planning program. Measurements of the precision and accuracy of the target localization, dose delivery, and patient positioning were made using a number of different jigs and devices. Procedures were developed for the safe and accurate clinical use of the system. RESULTS: The accuracy of the target localization is comparable to that of other treatment-planning systems. Gantry sag, which cannot be improved, was measured to be 1.7 mm, which had the effect of broadening the dose distribution, as confirmed by a comparison of measurement and calculation. The accuracy of positioning a target point in the radiation field was 1.0 +/- 0.2 mm. The calibration procedure using the room-based lasers had an accuracy of 0.76 mm, and using a floor-based radiosurgery system it was 0.73 mm. Target localization error in a phantom was 0.64 +/- 0.77 mm. Errors in positioning due to couch rotation error were reduced using the system. CONCLUSION: The system described has proven to have acceptable accuracy and precision for the clinical goals for which it was designed. It is robust in detecting errors, and it requires only a nominal increase in setup time and effort. Future work will focus on evaluating its suitability for use in the treatment of head-and-neck cancers not contained within the cranial vault.

Validation of K-edge 125I brachytherapy enhancement with silver compounds.

Brachytherapy with radioactive seeds implanted within the tumour volume has demonstrated good success rates in treating certain cancers. In an effort to improve the curative rates in cancer patients, ongoing research is being conducted to enhance the amount of radiation dose that is absorbed within the tumour volume while minimizing the dose absorbed by the surrounding normal tissue. One method for enhancing tumour dose absorption with 125I brachytherapy seeds is to increase the number of photoelectric atomic interactions within the tumour volume by introducing small quantities of a silver compound, taking advantage of the K-edge effect. Because low-energy electrons and Auger electrons are the primary sources of brachytherapy dose enhancement, acquiring accurate experimental measurements of absorbed dose increases is a major challenge. To circumvent this problem, an x ray fluorescence excitation spectroscopy dosimetry technique supplemented with clinically accepted dosimetry calculations was developed to estimate relative absorbed dose increases in a water phantom containing up to 7.5 mM of silver. Excellent agreement was observed between theoretically derived Monte Carlo dosimetric predictions and experimental measurements. These results successfully demonstrated that K-edge enhanced 125I brachytherapy is indeed possible with future development of a non-toxic silver chelate.

A macropencil beam model: clinical implementation for conformal and intensity modulated radiation therapy.

The increasing use of irregularly shaped, off-centre fields in advanced treatment techniques, particularly intensity modulated radiation therapy, has strained the limits of conventional, broad-beam dose calculation algorithms. More recent models, such as kernel-based pencil beams and Monte Carlo methods, are accurate but suffer from the time needed for calculations and from the lack of clearly established methods for determining the parameters needed to match calculations with the particular dosimetric characteristics of an individual machine. This paper presents the implementation of a model that uses an extended source model to calculate the variation of fluence at the patient surface for any arbitrarily shaped field. It uses a macropencil beam model to calculate phantom scatter. Both head scatter and phantom scatter models use exponential functions fit to a series of measurements to determine the model's parameters. The means by which the model can be implemented in a clinical setting using standard dosimetric equipment is presented. Results for two separate machines and three energies are presented. Comparisons with measurements for a set of regular and irregular fields demonstrate the accuracy of the model for conventional, conformal and intensity modulated treatments. For rectangular and irregular fields at depths up to 20 cm, the accuracy was better than < or =1.5%, compared with errors of up to 7.5% with a standard algorithm. For a 20-step intensity modulated field, the accuracy was 3.4% compared with 18% with the conventional algorithm. The advantages of this model for IMRT are discussed.

Noninvasive surgery of prostate tissue by high intensity focused ultrasound: an updated report.

OBJECTIVE: To establish clinical efficacy and safety of High Intensity Focused Ultrasound (HIFU) for the treatment of benign prostatic hyperplasia (BPH) in a multiple site clinical study. METHODS: Seven clinical sites were set up for the studies, five in the USA, one in Canada and one in Japan respectively. Sixty two patients were enrolled in these three studies. Transrectal ultrasound probes made to produce sufficient acoustic power required for focused ultrasound surgery of the prostate as well as to perform imaging of the prostate, were employed in the study. The probes ware made of 2.5, 3.0, 3.5, 4.0 and 4.5 cm focal length transducers to treat varying prostate sizes and shapes and operated at 4 MHz frequency for both imaging and treatment. The employed ultrasound device produced both transverse and longitudinal images of the prostate on the same display. The images were used for selection of tissue volume, treatment planning and monitoring of tissue during the HIFU treatment cycle. The patients in the USA and Canada were followed for two years and those in Japan were followed for one year on a regular interval. The results were evaluated for changes in the peak flow rate (Qmax in ml/s), quality of life (QOL) and International Prostate Symptom Score (IPSS). RESULTS: The average pre / post treatment results at 180 days were significantly different for Qmax, QOL and IPSS 8.5/14.2 (ml/s), 4.7/2.1 and 22/10 respectively. CONCLUSION: Under this protocol, HIFU was found safe and efficacious for the treatment of BPH. The HIFU treatment produced statistically significant results for the parameters measured with least complications. Additionally, the HIFU treatment was found to be durable.

Changes in cardiorespiratory fitness, psychological wellbeing, quality of life, and vocational status following a 12 month cardiac exercise rehabilitation programme.

OBJECTIVE: To examine and evaluate improvements in cardiorespiratory fitness, psychological wellbeing, quality of life, and vocational status in postmyocardial infarction patients during and after a comprehensive 12 month exercise rehabilitation programme. SUBJECTS: The sample population comprised 124 patients with a clinical diagnosis of myocardial infarction (122 men and two women). INTERVENTIONS: 62 patients were randomly allocated to a regular weekly aerobic training programme, three times a week for 12 months, and compared with 62 matched controls who did not receive any formal exercise training. A five year follow up questionnaire/interview was subsequently conducted on this population to determine selected vocational/lifestyle changes. RESULTS: Significant improvements in cardiorespiratory fitness (p < 0.01-0.001), psychological profiles (p < 0.05-0.001), and quality of life scores (p < 0.001) were recorded in the treatment population when compared with their matched controls. Although there were no significant differences in mortality, a larger percentage of the regular exercisers resumed full time employment and they returned to work earlier than the controls. Controls took lighter jobs, lost more time from work, and suffered more non-fatal reinfarctions (p < 0.05-0.01). CONCLUSIONS: Regularly supervised and prolonged aerobic exercise training improves cardiorespiratory fitness, psychological status, and quality of life. The trained population also had a reduction in morbidity following myocardial infarction, and significant improvement in vocational status over a five year follow up period.

Optimization of intensity modulated beams with volume constraints using two methods: cost function minimization and projections onto convex sets.

For accurate prediction of normal tissue tolerance, it is important that the volumetric information of dose distribution be considered. However, in dosimetric optimization of intensity modulated beams, the dose-volume factor is usually neglected. In this paper we describe two methods of volume-dependent optimization for intensity modulated beams such as those generated by computer-controlled multileaf collimators. The first method uses a volume sensitive penalty function in which fast simulated annealing is used for cost function minimization (CFM). The second technique is based on the theory of projections onto convex sets (POCS) in which the dose-volume constraint is replaced by a limit on integral dose. The ability of the methods to respect the dose-volume relationship was demonstrated by using a prostate example involving partial volume constraints to the bladder and the rectum. The volume sensitive penalty function used in the CFM method can be easily adopted by existing optimization programs. The convex projection method can find solutions in much shorter time with minimal user interaction.

Regulation of nuclear factor-kappa B and its inhibitor I kappa B-alpha/MAD-3 in monocytes by Mycobacterium tuberculosis and during human tuberculosis.

Blood monocytes from patients with active tuberculosis are activated in vivo, as evidenced by an increase in the stimulated release of proinflammatory cytokines, such as TNF-alpha, and the spontaneous expression of IL-2R. Further, monocytes from patients demonstrate an augmented susceptibility to a productive infection with HIV-1 in vitro. Mycobacterium tuberculosis and its components are strong signals to activate monocytes to production of cytokines. In this study we examined the basis of activation of monocytes during active tuberculosis and by M. tuberculosis. We found a constitutive degradation of I kappa B-alpha, the major cytoplasmic inhibitor of nuclear factor kappa B (NF-kappa B), in freshly isolated PBMC and monocytes from patients with tuberculosis. In contrast, I kappa B-alpha levels in PBMC and monocytes from healthy subjects or from patients with nontuberculous pulmonary conditions were intact. Further, by electrophoretic mobility shift assay, NF-kappa B was activated in monocytes from tuberculous patients. The expression of I kappa B-alpha gene, which is responsive to activation by NF-kappa B, was up-regulated in PBMC and monocytes from patients, but not in mononuclear cells from healthy subjects or those with nontuberculous lung diseases. By contrast, the expression of other adherence-associated early genes, such as IL-8 and IL-1 beta, was not up-regulated in PBMC of tuberculous patients. Further, M. tuberculosis and its tuberculin, purified protein derivative, induced the degradation of I kappa B-alpha and the expression of I kappa B-alpha mRNA, and purified protein derivative induced the activation of NF-kappa B in monocytes.

Effects of irradiation geometry on treatment plan optimization with linac-based radiosurgery.

A comparison was made of different treatment plans to determine the effect on the three-dimensional dose distributions of varying the allowed parameters in linac-based stereotactic radiosurgery with circular collimators; these parameters are arc position, length, and weighting, and collimator diameter. For the class of eccentrically shaped target volumes that are not so irregular as to require several separate isocenters, it was found that superior dose distributions could be achieved by varying arc length, arc position, arc weighting, and collimator diameter. An analysis of the results achieved with an automated planning program indicates that, in general, the variables of arc position and arc length are of greater importance than collimator size or beam weighting. However, there are cases where varying these latter two parameters does result in markedly better dose distributions. A deeper investigation into the effects of multiple collimators on the dose distribution in the area of steepest gradient demonstrated that multiple collimator sizes do not significantly degrade the dose falloff, which is in fact mostly determined by the effects of intersecting arcs.

Stereotactic radiosurgery: a review and comparison of methods.

PURPOSE: Stereotactic radiosurgery (SRS) is an evolving modality for treating well-circumscribed intracranial lesions. Different physical methods have been developed to deliver highly localized dose distributions accurately. We review the different methods and the documented clinical results to present a coherent view of radiosurgery, and to aid physicians and physicists in the appropriate use of this modality. DESIGN: A review of the medical physics and clinical literature was conducted. The physical aspects of the different methods and their impact on treatment were summarized. Results were compiled from those individual clinical series with adequate follow-up data to compare the various modalities with respect to treatment outcome for benign tumors, metastases, and vascular malformations. RESULTS: The physical accuracy was comparable between radiosurgical methods. Differences between gamma radiation and linear accelerator methods had little effect on the dose distribution for single isocenter treatments. Charged particle methods could produce better dose localization for large lesions (> 25 cm3) than was possible with photon methods. Clinical results indicate similar lesion control rates between all radiosurgical methods. There was a progressive increase in the median size of treated lesions for gamma radiation, linear accelerator, and charged particle methods. CONCLUSION: For small lesions (< 5 cm3), physical dose distributions are similar for the photon methods, but linear accelerator methods offer more flexibility for the treatment of intermediate-sized (5 to 25 cm3) lesions in applying future technical developments. More clinical results are needed before firm conclusions can be drawn on the type of lesions to be treated, and the dose-volume parameters to be used.

Endovascular treatment of cerebral arteriovenous malformations following radiosurgery.

PURPOSE: Previous reports of embolization of cerebral arteriovenous malformations (AVMs) have evaluated the technique as adjunctive therapy prior to surgery or radiosurgery; our aim is to assess the role of embolization following radiosurgery. PATIENTS: Six patients previously treated with radiosurgery and showing no response as judged by cerebral angiography were embolized 24 to 55 months (mean 34.3 months) after initial radiosurgery. RESULTS: In five of six, a significant volume reduction was achieved ranging from 60%-100% (mean 74%). One patient was treated with embolization alone and the AVM has remained fully thrombosed 2 years after treatment. Three patients underwent surgical resection for cure after embolization, and two patients had repeat radiosurgery to a significantly smaller AVM volume. One patient had an asymptomatic carotid dissection at embolization; however, no clinically apparent complications occurred in the treatment group. CONCLUSION: Embolization can be used after radiosurgery to assist in the management of those AVMs that have not responded to initial treatment.

Effects of respiratory motion on dose uniformity with a charged particle scanning method.

A three-dimensional spot-scanning technique for radiotherapy with protons is being developed at the Paul Scherrer Institute. As part of the effort to optimize the design and ensure clinically useful dose distributions, a computer simulation of the dose deposition in the presence of respiratory motion was performed. Preliminary experiments have characterized the proton beam and the scanning procedure. Using these parameters, the computer program calculated the dose within a uniform volume of water in the presence of respiratory motion. Respiration amplitude, respiration period, respiration direction, number of fractions, size and position of the beamspots and rescanning multiplicity were systematically varied and the effect on the dose distribution determined. The dose uniformity is very dependent on the direction of the respiration relative to the three independent beam scanning directions. The dose uniformity decreases with increasing respiration amplitude, but has little response to changes in respiration frequency. Rescanning the volume, such as with fractionation, improves the dose uniformity roughly as the square root of the number of fractions. Broad, Gaussian beams result in better dose uniformity than narrow, sharply delineated ones, but produce slower dose fall-off at the edges of the scanned volume. Results of this work are being incorporated into the design of the system.