Russell J. Hamilton

Fast iterative algorithms for three‐dimensional inverse treatment planning

Three types of iterative algorithms, algebraic inverse treatment planning (AITP), simultaneous iterative inverse treatment planning (SIITP), and iterative least-square inverse treatment planning (ILSITP), differentiated according to their updating sequences, were generalized to three dimension with true beam geometry and dose model. A rapid ray-tracing approach was developed to optimize the primary beam components. Instead of recalculating the dose matrix at each iteration, the dose distribution was generated by scaling up or down the dose matrix elements of the previous iteration. This significantly increased the calculation speed. The iterative algorithms started with an initial intensity profile for each beam, specified by a two-dimensional pixel beam map of M elements. The calculation volume was divided into N voxels, and the calculation was done by repeatedly comparing the calculated and desired doses and adjusting the values of the beam map elements to minimize an objective function. In AITP, the iteration is performed voxel by voxel. For each voxel, the dose discrepancy was evaluated and the contributing pencil beams were updated. In ILSITP and SIITP, the iteration proceeded pencil beam by pencil beam instead of voxel by voxel. In all cases, the iteration procedure was repeated until the best possible dose distribution was achieved. The algorithms were applied to two examples and the results showed that the iterative techniques were able to produce superior isodose distributions.

A comparison of four patient immobilization devices in the treatment of prostate cancer patients with three dimensional conformal radiotherapy.

PURPOSE To determine the variability of patient positioning during three-dimensional conformal radiotherapy (3D-CRT) for prostate cancer treated with no immobilization or one of four immunobilization devices, and to determine the effects of patient body habitus and pelvic circumference on patient movement with each individual inmobilization technique. METHODS AND MATERIALS To see whether our immobilization techniques have improved day-to-day patient movement, a retrospective analysis was carried out. A total of 62 patients treated at one facility on a single machine with 3D-CRT via a four-field box technique (anterior-posterior and opposed laterals) in the supine position with either no immobilization or one of four immobilization devices. Five groups of patients were compared: (a) group 1-no immobilization; (b) group 2-alpha cradle from the waist to upper thigh; (c) group 3-alpha cradle from waist to below the knees; (d) group 4-styrofoam leg immobilizer (below knees); and (e) group 5-aquaplast cast encompassing the entire abdomen and pelvis to midthigh with alpha cradle immobilization to their lower legs and feet. Prior to starting radiotherapy, portal films of all four treatment fields were obtained 1 day before treatment. Subsequently, portal films were then obtained at least once a week. Portal films were compared with the simulation films and appropriate changes were made and verified on the next day prior to treatment. A deviation of greater than 0.5 cm or greater was considered to be clincally significant in our analysis. We studied the difference among the types of immobilization and no immobilization by looking at the frequency of movements (overall, and on each of the three axes) that a patient had during the course of his treatment. Using a logistic regression model, the probability of overall and individual directional movement for each group was obtained. In addition, the effects of patient body habitus and pelvic circumference on movement were analyzed. RESULTS The maximum deviation was 2 cm and the median deviation was 1.2 cm. For each patient, the probability of movement ranged from 0 to 76%, with a mean of 39%. There was no significant difference seen in overall movement with any of the immobilzation devices compared to no immobilization, but there was less vertical (9 vs. 18%; p = 0.03) and AP (6 vs. 15%; p = 0.14) movement with the aquaplast than any other group. However, when examining the lateral direction, the aquaplast had significantly more movement (32 vs. 9%; p < 0.001). When accounting for body habitus and pelvic circumference, no immobilization device was effective in reducing movement in obese patients or in patients with pelvic circumference greater than 105 cm. The aquaplast group had a significantly increased amount of lateral movement with obesity (42 vs. 23%; p < 0.05), and with pelvic circumference >105 cm (33 vs. 29%; p < 0.05). CONCLUSIONS There was no significant reduction in overall patient movement noted with any of the immobilization devices compared to no immobilization. The aquaplast group had reduced vertical and AP movement of greater than 0.5 cm. There was significantly more lateral movement with aquaplast appreciated in obese patients or patients with pelvic circumferences greater than 105 cm. The aquaplast immobilization appears to be useful in reducing movement in two very clinicaly important dimensions (AP and vertical). Despite our findings, other immobilization may still be useful especially in the treatment of nonobese patients. It is clear that the optimal immobilization technique and patient positioning are yet to be determined.

Comparison of static conformal field with multiple noncoplanar arc techniques for stereotactic radiosurgery or stereotactic radiotherapy

PURPOSE Compare the use of static conformal fields with the use of multiple noncoplanar arcs for stereotactic radiosurgery or stereotactic radiotherapy treatment of intracranial lesions. Evaluate the efficacy of these treatment techniques to deliver dose distributions comparable to those considered acceptable in current radiotherapy practice. METHODS AND MATERIALS A previously treated radiosurgery case of a patient presenting with an irregularly shaped intracranial lesion was selected. Using a three-dimensional (3D) treatment-planning system, treatment plans using a single isocenter multiple noncoplanar arc technique and multiple noncoplanar conformal static fields were generated. Isodose distributions and dose volume histograms (DVHs) were computed for each treatment plan. We required that the 80% (of maximum dose) isodose surface enclose the target volume for all treatment plans. The prescription isodose was set equal to the minimum target isodose. The DVHs were analyzed to evaluate and compare the different treatment plans. RESULTS The dose distribution in the target volume becomes more uniform as the number of conformal fields increases. The volume of normal tissue receiving low doses (> 10% of prescription isodose) increases as the number of static fields increases. The single isocenter multiple arc plan treats the greatest volume of normal tissue to low doses, approximately 1.6 times more volume than that treated by four static fields. The volume of normal tissue receiving high (> 90% of prescription isodose) and intermediate (> 50% of prescription isodose) doses decreases by 29 and 22%, respectively, as the number of static fields is increased from four to eight. Increasing the number of static fields to 12 only further reduces the high and intermediate dose volumes by 10 and 6%, respectively. The volume receiving the prescription dose is more than 3.5 times larger than the target volume for all treatment plans. CONCLUSIONS Use of a multiple noncoplanar conformal static field treatment technique can significantly reduce the volume of normal tissue receiving high and intermediate doses compared with a single isocenter multiple arc treatment technique, while providing a more uniform dose in the target volume. Close conformation of the prescription isodose to the target volume is not possible using static uniform conformal fields for target shapes lacking an axis of rotational symmetry or plane of mirror symmetry.

The potential for normal tissue dose reduction with neoadjuvant hormonal therapy in conformal treatment planning for stage C prostate cancer

PURPOSE Preirradiation hormonal cytoreduction of prostate cancer has been proven to reduce exposure of normal structures by decreasing the size of the target volume. Dose-volume histogram (DVH) analysis, however, does not always appear to demonstrate a strong positive benefit with the use of neoadjuvant hormone therapy. This study analyzes various other factors influencing dose to normal organs, which may determine the success or failure of neoadjuvant hormonal therapy in achieving its goals. METHODS AND MATERIALS Patients with bulky clinical Stage C adenocarcinoma of the prostate were given 3 months of hormone treatment consisting of oral Flutamide and monthly Zoladex injections prior to irradiation. Computerized tomography (CT) scans of the pelvis were obtained both prior to and following hormonal treatment. Treatment plans were generated by three-dimensional (3D) conformal treatment planning. The change in the volume of the prostate was assessed along with the percentage of prescribed dose delivered to the rectum and bladder. Various factors such as prostate size, bladder/rectum size, and organ shape were studied. Both dose-volume histograms (DVH) and dose-surface area histograms (DSH) were used for analysis. RESULTS Six of seven patients had reduction in the size of their prostates. The mean volumes of the prostate before and after hormonal manipulation were 129.1 +/- 32.9 standard deviation (SD) cm3 and 73.0 +/- 29.5 SD cm3, respectively (p = 0.0059). The volume of rectum receiving 80% of the prescribed dose was reduced in five of seven patients from a mean of 83.2 to 59.9 cm3 (p = 0.045). The volume of bladder receiving 80% of the prescribed dose was also reduced in five out of seven patients from a mean of 74.5 to 40.2 cm3 (p = 0.098). Correlation between the size of the prostate and volume of rectum and bladder treated was not always consistent: greater reduction in prostate size did not necessarily result in large decreases in dose to bladder or rectum. The total size of the bladder and rectum were found to be important factors in normal tissue radiation exposure; the benefits of hormone therapy may be lost if the bladder and rectum are allowed to decrease in size. Also, the bladder may be prone to sagging into the pelvis of some patients following hormone therapy, resulting in a less optimal therapeutic ratio. CONCLUSION Reduction in prostate size by neoadjuvant hormonal manipulation does decrease the amount of normal tissue irradiated in most patients. However, the correlation between the reduction in prostate size and amount of rectum or bladder treated is not linear if other variables are not controlled. Factors such as the shape of the organs, as well as the distensible nature of the bladder and rectum, play major roles in dose to normal tissues. These facts may mask the benefits of cytoreduction and could be obstacles in realizing consistent benefits from preirradiation hormonal treatment in the clinical setting if they are ignored.

Repositioning accuracy of a noninvasive head fixation system for stereotactic radiotherapy

We report on the repositioning accuracy of patient setup achieved with a noninvasive head fixation device for stereotactic radiotherapy. A custom head mask which attaches to our stereotactic radiosurgery head ring assembly is fabricated for each patient. The position and orientation of a patient in the stereotatic space at the time of treatment are determined from analyzing portal films containing images of radio-opaque spheres embedded in a custom mouthpiece. From analysis of 104 setups of 12 patients, we find that the average distance between the treated isocenter and its mean position is 1.8 mm, and that the standard deviations of the position of the treated isocenter in stereotactic coordinate space about its mean position are less than 1.4 mm in translation in any direction and less than 1 degree of rotation about any axis.

Performance of a video-image-subtraction-based patient positioning system

PURPOSE We have developed and tested an interactive video system that utilizes image subtraction techniques to enable high precision patient repositioning using surface features. We report quantitative measurements of system performance characteristics. METHODS AND MATERIALS Video images can provide a high precision, low cost measure of patient position. Image subtraction techniques enable one to incorporate detailed information contained in the image of a carefully verified reference position into real-time images. We have developed a system using video cameras providing orthogonal images of the treatment setup. The images are acquired, processed and viewed using an inexpensive frame grabber and a PC. The subtraction images provide the interactive guidance needed to quickly and accurately place a patient in the same position for each treatment session. We describe the design and implementation of our system, and its quantitative performance, using images both to measure changes in position, and to achieve accurate setup reproducibility. RESULTS Under clinical conditions (60 cm field of view, 3.6 m object distance), the position of static, high contrast objects could be measured with a resolution of 0.04 mm (rms) in each of two dimensions. The two-dimensional position could be reproduced using the real-time image display with a resolution of 0.15 mm (rms). Two-dimensional measurement resolution of the head of a patient undergoing treatment for head and neck cancer was 0.1 mm (rms), using a lateral view, measuring the variation in position of the nose and the ear over the course of a single radiation treatment. Three-dimensional repositioning accuracy of the head of a healthy volunteer using orthogonal camera views was less than 0.7 mm (systematic error) with an rms variation of 1.2 mm. Setup adjustments based on the video images were typically performed within a few minutes. The higher precision achieved using the system to measure objects than to reposition them suggests that the variability in repositioning is dominated by the ability of the therapist to make small, controlled changes in the position of the patient. CONCLUSION Using affordable, off-the-shelf technology, we have developed a patient positioning system that achieves repositioning accuracy normally associated with fractionated stereotactic systems. The technique provides real-time guidance and can be used to easily and quickly correct patient setup before every treatment, thus significantly reducing overall random positioning error. This improved positioning capability provides the precision required to realize the potential gains of conformal radiotherapy.

Matching photon and electron fields with dynamic intensity modulation

A technique was developed to reduce the size and magnitude of the hot and cold spots in the abutting regions of photon and electron fields. The photon and electron fields were set up such that the photon field extended approximately 2 cm into the electron field in the abutting region. The region of the photon beam that overlapped the electron field was modulated using a multileaf collimator, effectively broadening the photon penumbra to make it complimentary to the electron penumbra. The computer calculations were verified using film measurements for abutting a 6 MV photon beam with a 9 MeV electron beam. A uniform dose was achieved at a prespecified depth of 2 cm, and dose uniformity was improved at the specified depth and beyond compared with unmodulated photon beams. A slight increase in dose inhomogeneity was seen at shallower depths. The overall areas of the hot and cold spots were significantly reduced. The technique also reduced the sensitivity of dose homogeneity to setup errors such that the magnitudes of the hot and cold spots were about half of those produced with unmodulated photon beam when an overlap or gap of 4 mm was introduced. The technique was applied to the treatment of a head and neck cancer and a lymphoma involving the right pleura with markedly reduced dose inhomogeneity in the abutting regions.

A three-dimensional algorithm for optimizing beam weights and wedge filters.

An essential step towards optimizing and automating radiation therapy treatment planning is to develop an effective algorithm to find the optimal beam weights and wedge filters for a given set of beam directions and modalities. This problem is solved by introducing a variable transformation based on the universal and omni wedge principles. Instead of directly optimizing an objective function with respect to wedge angles and orientations, each field is first decomposed into a superposition of an open field and two orthogonal wedged fields. This transforms the problem of finding J beam weights, wedge angles, and orientations to that of optimizing a system with 3J beam weights (J open beams and 2J nominal wedged beams), where J is the total number of incident beam directions. An iterative algorithm based on a method originally developed for image reconstruction is used to find the 3J beam weights. The technique is applied to a few clinical cases. Treatment plans are improved compared to those obtained through the conventional manual trial and error planning process. In addition, planning time and effort are greatly reduced.

Verification of the omni wedge technique.

The optimal field shape achieved using a multileaf collimator (MLC) often requires collimator rotation to minimize the adverse effects of the scalloped dose distribution the leaf steps produce. However, treatment machines are designed to deliver wedged fields parallel or perpendicular to the direction of the leaves. An analysis of cases from our clinic showed that for 25% of the wedged fields used to treat brain and lung tumors, the wedge direction and optimal MLC orientation differed by 20 degrees or more. The recently published omni wedge technique provides the capability of producing a wedged field with orientation independent of the orientation of the collimator. This paper presents a comparison of the three-dimensional (3D) dose distributions of the omni wedged field with distributions of wedged fields produced using both the universal and dynamic wedge techniques. All measurements were performed using film dosimetry techniques. The omni wedge generated fields closely matched the conventional wedged fields. Throughout 95% of the irradiated volume (excluding the penubra), the dose distribution of the omni wedged field ranged from +5.5 to -3.5 +/- 1.5% of that of the conventionally wedged fields. Calculation of the omni wedged field is as accurate as conventional wedged field calculation when using a 3D treatment planning systems. For two-dimensional treatment planning systems, where one must assume that the omni wedged field is identical to a conventional field, the calculated field and the delivered field differs by a small amount.

The omni wedge: A method to produce wedged fields at arbitrary orientations

A method to produce wedged fields at any orientation relative to the collimator is described. The wedged field is generated by combining two appropriately weighted orthogonal wedged field segments at fixed collimator and gantry positions. The method requires only that wedged fields can be produced in orthogonal directions without rotating the collimator, such as is commonly provided on most radiation therapy accelerators by sets of standard and rotated wedges. Expressions are derived relating the effective wedge angle and orientation to the weighting and wedge angles of the orthogonal wedged field segments. This technique will be important when using multileaf collimator field shaping for which collimator rotation is dictated by target or critical structure shape and orientation. The term omni wedge is introduced to describe this technique.