I Rusu

3D CT-based volumetric dose assessment of 2D plans using GEC-ESTRO guidelines for cervical cancer brachytherapy

PURPOSE To investigate two-dimensional (2D) radiograph-based plans using three-dimensional (3D) dose-volume histogram (DVH) parameters following guidelines from Gynecologic GEC-ESTRO Working Group (GEC-ESTRO). METHODS AND MATERIALS Nineteen high-dose-rate (HDR) fractions from 8 patients were studied. Prescription was 45 Gy from external beam radiation therapy plus 30 Gy in five fractions from HDR using tandem and ring/ovoids. Both radiographs and CT scan were obtained. Treatment was planned using radiographs following American Brachytherapy Society (ABS) guidelines. Retrospective evaluation of above 2D plans on a 3D volumetric basis was achieved by generating CT image-based 3D plans using same dwell times. RESULTS In 2D plans, International Commission on Radiation Units and Measurement (ICRU) bladder and rectal point doses were 3.8+/-0.4 and 3.0+/-0.5 Gy, respectively. In 3D plans, rectum D(2 cc) is 4.0+/-1.0 Gy and bladder D(2 cc) is 5.4+/-0.9 Gy. Position of actual hottest spot in 3D rectum volume was close to the position of ICRU rectal point. ICRU bladder point did not match with the actual hottest spot in 3D bladder volume. In 2D plans, H-point dose was 5.8+/-0.2 Gy. In 3D plans, dose to CT-based cervix (D(90)) reduced from 7.1 to 4.2 Gy as the cervical volume increased from 12 to 39 cc. Average D(2 cc)/ICRU dose ratio was calculated to be 1.36/1.01 for bladder/rectum, respectively. CONCLUSIONS The DVH analysis of 2D plans revealed a suboptimal coverage of CT-based cervix and a negative correlation between coverage and cervical size. Rectum dose to 2 cc weakly correlated with ICRU point dose. Currently published constraint for bladder in 3D planning is tighter than ABS guidelines in past 2D planning.

TU‐FF‐A2‐02: Dual‐Fields Rotational Total Skin Electron Irradiation/therapy

Purpose: We have developed a new technique for rotational total skin electron therapy (RTSET). The technique combines the rotational method of McGill University with the Stanford University angled dual‐fields method. We report dosimetry characteristics and in‐vivodosimetry.Method and Materials: Patients stand on a rotational (0.9 rpm) platform at an extended SSD (332cm) with field‐size 133‐cm by 133‐cm. The gantry is angled 70° and 110° so that beam points above patients head and below patients feet, respectively, in order to minimize contaminant x‐ray dose. A “high‐dose‐rate” mode (600 MU/min) on a Varian‐21EX delivers a 6 MeV‐electron beam. Daily treatments require only about 0.5 h, one‐half of which is used for patient set‐up. Shields are used for eyes, nails, and toes. Dosimetrymeasurements include static dose‐rate at extended SSD, rotational dose‐rate for a rotating patient/phantom, and a power law correction for SSD variation. In‐vivo dose measurements are performed using XV‐films and MOSFET. Films strips are attached firmly on patients skin to avoid any air gaps. Results: We obtained MU = Dose/(0.0751*0.39 *(332/SSD)2.25). Therefore, a prescription dose of 125cGy at 325cm SSD required 4068 MU or 7.5 revolutions. The dose distribution along the vertical direction was measured by a parallel‐plate ionization chamber in a phantom. For a patient prescribed 116cGy daily, the average measuredsurface dose with film was 110cGy, within 5% of prescribed dose. In‐vivoMOSFET on a patient prescribed 125cGy daily, measured doses were distributed from 99% to 102% from the prescribed dose of 125cGy. Conclusion: The dual‐field RTSET offers combined advantages of shorter beam‐on time, uniform dose distribution, larger treatment fields, and less x‐ray contamination than other techniques. Our in‐vivomeasurements revealed that delivered dose matched prescribed dose to < 5% and dose uniformity was within 6% variation.

Stereotactic Radiation for Palliation of Skull Base Recurrences of Salivary Gland Carcinomas: Implications for Tumor Targeting

Background Approximately 3 to 13% of salivary carcinomas recur at the skull base. We report our experience treating these recurrences with stereotactic radiation. Methods In total, 14 patients with skull base recurrence of salivary gland carcinoma were identified. Patient characteristics, treatment parameters, response to treatment, local recurrence‐free/overall survival, and patterns of failure were studied. Results All 12 symptomatic patients experienced palliation of symptoms. Two grade 3 toxicities were observed. Local recurrence‐free survival after skull base treatment was 28 months (74 months after allowing for additional course of salvage radiotherapy). Overall survival was 153 months from primary diagnosis and 67 months from first skull base failure. Of 13 treatment failures, 8 occurred at margins; the rest were infield. All intracranial failures occurred along meningeal surfaces. Conclusions Stereotactic radiation provides well‐tolerated palliation for the majority of patients, but with a high rate of local failure. Due to the propensity for meningeal failures, we suggest increasing margins along the meningeal surfaces when treating these patients.

SU-F-T-638: Is There A Need For Immobilization in SRS?

PURPOSE Frameless Stereotactic radiosurgery (SRS) is increasingly used in the clinic. Cone-Beam CT (CBCT) to simulation-CT match has replaced the 3-dimensional coordinate based set up using a stereotactic localizing frame. The SRS frame however served as both a localizing and immobilizing device. We seek to measure the quality of frameless (mask based) and frame based immobilization and evaluate its impact on target dose. METHODS Each SRS patient was set up by kV on-board imaging (OBI) and then fine-tuned with CBCT. A second CBCT was done at treatment-end to ascertain intrafraction motion. We compared pre- vs post-treatment CBCT shifts for both frameless and frame based SRS patients. CBCT to sim-CT fusion was repeated for each patient off-line to assess systematic residual image registration error. Each patient was re-planned with measured shifts to assess effects on target dose. RESULTS We analyzed 11 patients (12 lesions) treated with frameless SRS and 6 patients (11 lesions) with a fixed frame system. Average intra-fraction iso-center positioning errors for frameless and frame-based treatments were 1.24 ± 0.57 mm and 0.28 ± 0.08 mm (mean ± s.d.) respectively. Residual error in CBCT registration was 0.24 mm. The frameless positioning uncertainties led to target dose errors in Dmin and D95 of 15.5 ± 18.4% and 6.6 ± 9.1% respectively. The corresponding errors in fixed frame SRS were much lower with Dmin and D95 reduced by 4.2 ± 6.5% and D95 2.5 ± 3.8% respectively. CONCLUSION Frameless mask provides good immobilization with average patient motion of 1.2 mm during treatment. This exceeds MRI voxel dimensions (∼0.43mm) used for target delineation. Frame-based SRS provides superior patient immobilization with measureable movement no greater than the background noise of the CBCT registration. Small lesions requiring submm precision are better served with a frame based SRS.

Evaluation of RTOG Guidelines for Monte Carlo-based Lung SBRT Planning

Twelve lung SBRT plans calculated with heterogeneity corrected PB algorithm on the Brainlab iPlan were normalized to deliver 50Gy PTV dose in 5 fractions using 8-12 non-coplanar, conformal 6MV fields. (PB) Treatment plans were then recalculated using MC algorithm without changing any beam parameters (MC Non-Optimized). Next, MUs in MC plans were optimized (MC Optimized) to satisfy PTV prescription dose at 50Gy. PTV dose parameters Dmin, Dmax, Dmean, D90, D2cm, R50% and conformality index (CI), were evaluated and compared per RTOG 0813 guidelines for PB, MC Non-Optimized and MC Optimized plans.

SU-E-T-699: Evaluation of a Monte Carlo Based TPS for SBRT

Purpose: Hypo‐fractionated dose regimens used in stereotactic body radiotherapy(SBRT) demand superior dose calculation and delivery accuracy. We evaluate the impact of using Monte Carlo (MC)treatment planning system (TPS) in SBRT of lung patients.Methods: A commercial MC TPS (iPlan 4.1.2, BrainLAB AG, Feldkirchen, Germany) validated in heterogeneous phantom geometry was employed. For lung and bone heterogeneities, iPlan has shown excellent agreement with measurements to within 2–3%/2mm for dose/distance‐to‐agreement. Five lungSBRT patient plans (PTV: 10 – 250 cc, mean = 67.5cc, S.D. = 85.9cc) originally computed with pencil beam (PB) algorithm were retrospectively analyzed using MC. Prescription dose was 50Gy to >99% of PTV in 5 fractions using 10–12 non‐coplanar 6MV beams. For each patient, four delivery modes were considered: 3DCRT, IMRT, dynamic‐conformal arc (DCA), and hybrid‐arc (HA). Monitor units obtained from PB plans were used to recalculate MCdose. All patient plans were compared using PTV and OAR dose indices (Dmin, Dmax, Dmean and D95).Results: MC calculated doses were significantly lower compared to PB plans for PTV surrounded by lung medium. Owing to electron disequilibrium, largest dose discrepancy was seen in superficial (<5mm) regions of target. For a 22mm target (PTV = 38.2cc): PTV Dmean and Dmin were 14.3% and 27.6% lower respectively. Corresponding differences were smaller for ITV (Dmean and Dmin 14% and 7.6% lower respectively) showing better agreement between PB and MC plans. For PTV surrounded by soft‐tissue or adjacent to chest‐wall, MCdoses were closer to PB (<5% agreement). Results were relatively insensitive to target size and mode of delivery Conclusions: This study highlights the need for MC based dose calculations in lungSBRT. Closer agreement between PB and MC results for ITV calls for robust and reliable image‐guidance.

SU‐GG‐T‐190: In‐Vivo Surface Dosimetry With An Optically Stimulated Luminescence Dosimeter

Purpose: We investigate the use of a new optically stimulated luminescencedosimeter (OSLD) for surface dosimetry.Method and Materials: OSLDs are small plastic disks (5mm diameter × 0.2mm thick) infused with aluminum oxide doped with carbon ( Al 2 O 3 : C ) . Their operation is similar to TLDs; the irradiated crystals yield a signal proportional to absorbed dose upon optical stimulation. The detectors are housed in a light‐tight plastic casing measuring 24 × 10 × 1‐mm. A dose‐response curve for 6MV photons from Varian 21Ex was established. OSLDs were next used to measure surface/buildup doses in a plastic water phantom. Three field sizes were considered: 10×10cm, 5×5cm and 2×2cm and results compared with a parallel plate ion chamber. Clinical performance of OSLD was next evaluated by measuring surface doses in head and neck patients undergoing 3DCRT/IMRT. All measurements were repeated with a MOSFETdosimeter currently used in our clinic. Results: The detectors are easy to use, require no preparation/annealing and can be read 10 minutes post‐irradiation. The effective depth of measurement for OSLD was found to be 0.4 mm inside the plastic casing. Surface/buildup region doses with OSLD were in excellent agreement with ion chamber data. When these detectors were used to measure patient surface dose, they compared favorably with MOSFET detectors. Due to smaller intrinsic buildup, OSLD measured surface doses were slightly lower with a mean dose ratio of 0.968 ± 0.011. Conclusion: A new commercially available Al 2 O 3 detector from Landauer is a convenient tool for measuring patient surface dosimetry. Detector performance compares well with existing dosimeters in radiotherapy. Unlike other dosimeters, however, the new detector shows no field size, energy or angular dependence.

WE‐E‐332‐07: Workloads for High Dose Rate Brachytherapy Facilities

Purpose: Neither NCRP Report 147, Structural Shielding Design… Facilities nor NCRP Report 151, Structural Shielding Design… Facilities address high dose rate (HDR) brachytherapy facility shielding. From a review of HDR patients treated at a major medical center, we report yearly patient census, treatment times, and workloads, compare them to literature data, and observe the consequences for HDR facility design. Method and Materials: For shielding calculations for an HDR vault, the yearly workload is defined as the product of the air kerma strength (Gy m2/h) at 1 m and the time (h) the source is used yearly. For a conventional 10 Ci HDR Ir-192 source, 0.04 Gy m2/h is commonly used. Limited literature data on HDR workloads varies between 4 Gy/Yr to 15 Gy/Yr. From a two year (2006, 2007) review of patient treated with HDR, we identify patient treatment dates, anatomic sites, prescribed absorbed doses, fractions, and total treatment times. From these data we extract yearly patient census, workloads by anatomic site and total yearly workloads. Results: By anatomic site, of 41 patients treated yearly, 14 were vagina, 13 uterus, 7 breast, 3 base of tongue, 2 sarcoma, and 2 head and neck. The average total number of fractions yearly was 232. Excluding quality assurance procedures, the average total yearly treatment time was 16.4 h and the average yearly workload at 1 m was 0.655 Gy/Yr, about 0.013 Gy/Week. These data are well below suggested literature values. This workload is less than 20% of the yearly workload of 4 Gy assuming 200 patients per year and a total yearly treatment time of 100 h used in this HDR facilitys design. Conclusion: Using current popular HDR treatment methodologies, workloads of 1 to 2 Gy/Yr likely are sufficient for HDR facility design, potentially reducing shielding costs.

SU‐GG‐T‐182: Effects of Block Location On Dosimetry in the Total Body Irradiation (TBI)

Purpose: Blocks are frequently used in TBI to protect critical organs such as lung. Among different cancer centers, the location of block varies significantly; for example, in the block tray, between the machine and patient, or directly in front of the patient. In this study, we investigate effects of block location on the dosimetry of blocked region. Method and Materials: Blocks at four source‐block distances were studied, i.e. 65cm (block tray), 120cm, 200cm and 350cm (in front of patient). All blocks were 5 HVL thick and were made based on the same film to protect an area of 12.5×12.5 cm2. Dosimetry of blocked area was measured using an ion chamber and film in a phantom at 10cm depth. All results were normalized to the open field dose. Results: Doses at the center of blocked area for source‐blockdistances of 65, 120, 200, 350 cm are 18.1%, 15.7%, 13.9% and 13.5% respectively. The penumbra of blocked area decreases from 0.9 cm to 0.4 cm as the block moves from the block tray closer to the phantom. For a desired transmission dose of 20% (as a percentage of open field), the effective size of blocked area (i.e. area whose dose is under the desired transmission) is 7.2cm, 10.1cm, 10.9cm and 11.4cm respectively for blocks at 65, 120, 200, 350 cm source‐block distances respectively. For a desired transmission of 30%, the effective size of blocked area is 11.7cm, 11.9cm, 12.1cm and 12.2cm respectively. For a 50 % level, it is 12.3cm, 12.5cm, 12.7cm and 12.7cm respectively. Conclusion: Dose under TBI blocks is significantly higher and the effective size of blocking is significantly smaller when block is located in block tray than directly in front of the patient. Hence, proper positioning of TBI blocks is critical to achieve the desired dose reduction.

SU‐FF‐T‐228: Factors Affecting IMRT QA

Purpose: We evaluate the impact of plan parameters and setup errors on IMRT QA results. Methods and Materials: Prostate (23MV, 5 field) and H& N (6 MV, 7 field) IMRT plans were exported to a cubic QA phantom. First, the magnitude of random errors in IMRT QA was estimated by repeating each measurement ten‐times. Next, the following parameters were independently adjusted: intensity levels (5 to 20), minimum MLC segments (0.5 to 1.5cm), phantom setup error (0.5cm), and their impact on IMRT QA was evaluated via ionization chamber and EDR film. Coronal plane film results were analyzed using γ‐index, dose difference, and distance to agreement. Results: Based on repeat QA measurements, the central axis ion chamber dose was found to be highly reproducible and in good agreement with planned dose (mean deviation, prostate: 1.79±0.10%; and HN H&N: 0.60±0.09). Varying the number of intensity levels and MLC segment size had a small effect on the central axis ion chamber dose (<1.2%). However, film results showed that mean γ‐index changed by 10% when minimum MLC segment size was increased from 0.5 to1.5cm. The number of intensity levels had a smaller effect on γ‐index. For a 0.5cm setup error in H&N plan, 3.8% deviation was noted in the measured dose, however this coincidentally improved QA results. No corresponding dose discrepancy was seen in a fiducial‐less film. Conclusions: For an accurately modeled linac,IMRT QA results show remarkable consistency and agreement with planned dose. Plan complexity tended to worsen QA results for H&N patients. Sub‐optimal IMRT plans with incorrect plan parameters and setup errors can have detrimental effect on patient treatment. However, IMRT QA does not always flag these errors, as their impact on QA results may not be large enough to be noticed.