lung, cervical, ovarian, renal and nasopharyngeal cancers) while showing low expression in normal tissue. We used the Cy7dye as the optical contrast agent that was coupled to folate. A number of tests to determine the linearity, stability, and thesaturation limit of the system were performed prior to in vivo imaging. The Cy7-folate conjugate was administered intrave-nously into the tail vein of tumor bearing mice and images were recorded at several different time points before and after i.v.injection of the contrast agent.

Results: The testing of the imaging system using a surrogate imaging phantom containing Cy7 with known concentrationshowed a linear correlation between emission intensity and Cy7 concentration. The animal study revealed that the folatereceptor-expressing tumors can be distinguished from the surrounding healthy tissue without further image processing orenhancements. In animals, it was found that the system can detect and image tumors as small as 0.1 millimeter, which is farbeyond the detection limit of conventional imaging techniques. In figure 1 we show the acquired images at three different timepoints after i.v. injection of the conjugate.

Conclusions: We have shown that receptor targeted fluorescent imaging is an ultra-sensitive and reliable technique fortumor-specific imaging. Coupled with the low cost and ionizing radiation-free environment of the imaging modality, it isbelieved that this imaging system and method may have a wide range of applications in radiation oncology related research e.g.it allows for the study of biological processes and the monitoring of the therapeutic response.

2423 Surface Dose Prediction and Verification for IMRT Plans Using Line Dose Profiles

R. Berg,1 S. Klash,2 M. Gossman1

1Radiation Oncology, Erlanger Medical Center, Chattanooga, TN, 2SJK Physics, Radiation Oncology, Dallas, TX

Purpose/Objective: MOSFETs are used to verify the entrance dose for a given conformal field arrangement as expected froma three dimensional treatment plan. For sliding window IMRT treatments, a combination of fluence patterns are used to providean integral dose to the target tissue. This renders predicting measurements at the skin surface more difficult because ofsubstantial dose gradients in each beam. The objective was to find a method of predicting MOSFET skin dose and test theapplicability in the clinic.

Materials/Methods: The technique was developed using the Helios IMRT computerized treatment planning system. Using itsability to provide line dose profiles, a calculation of the anticipated MOSFET measurement can be made by extrapolating theprofile depth to the last 1 to 4 millimeters of the patient skin. It is here where the computer dose calculation model may notalways be dependable. Sagittal and coronal views also provide useful information to aid in the placement of the MOSFET onthe patient. The approved IMRT plan has a mini-bolus of appropriate thickness added to a selected location at the patient skinsurface. The location is judiciously selected to provide a region of large dose values and low dose gradients. The MOSFETremains on the patient during the delivery of all the fields. The resulting dose to the MOSFET is then compared to predictedvalues for treatment dose verification.

Results: 61 measurements were made on 59 patients. 57 of the measurements (93%) were within 10 centigray of the predictedvalue. The remaining 4 measurements had large predicted values ranging from 157 to 224 centigray and still fell within 10%of the expected value. The distribution of predicted minus measured values was approximately centered about zero.

Conclusions: The method of extrapolating a line dose to the skin surface is a useful technique for predicting in vivo MOSFETdose measurements. This serves as additional verification of the IMRT plan in addition to the quality assurance prior tocommencement of treatment.

2424 Dosimetry Validation of IMRT Treatment Plans Using MLC Log Files

W. Luo, J. Li, J. Yang, S. McNeeley, C. C. Ma

Radiation Oncology, Fox Chase Cancer Center, Philadelphia, PA

Purpose/Objective: In intensity-modulated radiation therapy (IMRT), a number of beams of nonuniform fluences are used tooptimize the composite dose distribution. The fluence maps for individual beams are derived through a treatment optimizationsystem and are delivered using computer-controlled devices such as multileaf collimators (MLC). Potential sources of error inIMRT treatment include dose calculation inaccuracies, beam delivery errors and target localization uncertainties (due to setuperrors and organ motion). Dosimetric verifications of IMRT treatment are often performed using ion chambers, film, TLDs, etc.

S590 I. J. Radiation Oncology ● Biology ● Physics Volume 60, Number 1, Supplement, 2004

on anatomical patient phantoms, which do not give the dose in the actual patient. Monte Carlo simulations have been used tovalidate IMRT plans using patient CT and fluence maps derived from MLC leaf sequences, which do not account for beamdelivery errors. The purpose of this work is to implement Monte Carlo based IMRT quality assurance (QA) using actualtreatment fluence maps rebuilt from the MLC dynalog files to validate both treatment planning and beam delivery accuracy.

Materials/Methods: The IMRT plans were generated on the CORVUS treatment optimization system (NOMOS Corp.,Sewickley, PA) and the treatments were delivered on Varian 21EX accelerators with 60 pairs of MLC leaves. The MLC dynologfiles record the actual MUs (fractional doses), gantry angle and MLC leaf positions for each MLC segment. The data are takennominally every 50 ms by the MLC controller during a dynamic or step-and-shoot beam delivery. Two dynolog files aremanually saved for each field. A program has been written to process the data and create 400x400 fluence maps for individualfields based on the MLC leaf positions and MUs for the open field segments, leakage, and scatter for the blocked areas. Thesefluence maps were then used in our home-grown Monte Carlo software to calculate the dose distribution in the patient (orphantom). The dose distribution obtained this way can be compared with that from the treatment plan and measurements forIMRT QA.

Results: The Monte Carlo simulation results based on 20 prostate IMRT patients were compared with ion chamber, film andTLD measurements. The fluence maps for individual fields agreed well between Monte Carlo and measurements, indicating theaccuracy and reliability of the MLC dynolog files. These results are consistent with the previous findings (Li et al Med Phys2003 30 799–805). We have compared dose distributions in a pelvic phantom recalculated by CORVUS as the hybrid QA plansfor these prostate treatments with film and ion chamber measurements. The relative dose distributions agreed generally within3%/3mm for the high dose regions, and the absolute dose values at the selected measurement positions were consistent within2.5% for more than 70% of the cases and within 4% for all the cases. Our Monte Carlo results also revealed a 7% differencein the IMRT dose distribution for non-coplanar IMRT plans with and without applying heterogeneity corrections, which wascaused by the excessive beam attenuation by the femoral bones.

Conclusions: The MLC dynolog files record accurately the leaf positions and MUs for each MLC segment during IMRT beamdelivery. Monte Carlo simulations of exact patient geometry rebuilt from CT using actual fluence maps reconstructed fromMLC dynolog files provide a unique solution for IMRT QA to validate the dose distribution delivered to a patient.

2425 Factors Influencing the Dose to Rectum During the Treatment of Prostate Cancer with IMRT

N. M. Reddy, B. M. Sood, D. Nori

Radiation Oncology, The NY Hospital Queens, Flushing, NY

Purpose/Objective: Patients with small rectal wall volumes have been shown to be at a higher risk for bleeding than patientswith large rectal wall volumes (Jackson et al. IJROBP 2001; 49:685–698) and dose received by 10 cc of rectum correlated withrectal toxicity (Huang et al. IJROBP 2002; 54:1314–1321). The purpose of this study was to observe the variation in thevolumes of whole rectum (WR) and rectal wall (RW), and to evaluate the relationship between dose to 30, 50 or 70% volumesor dose per cc of WR or RW vs. WR or RW volumes.

Materials/Methods: Prostate, SV, whole rectum (with fillings � WR), rectal wall only (WR-fillings � RW) and bladder werecontoured using the CAT scan images with 5mm separation, for 21 patients. The caudal limit of rectum was the first slice abovethe anal verge and the cranial limit was first slice below the sigmoid flexure. The prescription was 45 Gy to prostate and SVin 25 fractions (Prostate � SV � CTV, CTV � margin � PTV) and 36 Gy to prostate in 20 fractions, with 5F IMRT plans.Superior and posterior margins for CTV or prostate only was 0.5 cm and 1 cm on all other dimensions. Beam angles,dose-volume constraints and CTV margins were nearly the same for all patients. This analysis was limited to 45 Gy coursebecause some of the patients had 45 Gy IMRT � brachytherapy. Dose to 30, 50 and 70% of WH and RW were estimated fromthe DVH. In addition, dose to 1 cc at 30, 50 and 70% volumes of WR or RW were estimated by factoring out the variationsin WR and RW volumes between patients. At a given % volume level, the dose to that % volume was divided by the % volumein cc of the WR or RW. For example, the dose received by 30% of WR was divided with 7.5 in the case of a patient with aWR volume of 25 cc and the dose received by 30% of WR was divided with 53 in the case of WR with a volume of 177. Theresulting value is expressed as dose per cc at 30, 50 and 70% volumes. These data were plotted against the total volumes ofWR or RW. Pearsons correlation coefficient r and the two-tailed P values were estimated to measure the degree of associationbetween WR vs. RW and dose to a given % volume or per cc of WR or RW vs. WR or RW volumes.

Results: The volume of WR varied from 25 to 177 cc between patients. Larger the volume of WR, larger was also the volumeof RW (18–68cc, P � 0.01). Dose to RW was less in 10 patients by 6–21% at 50% volume, compared to doses to WR. Thisdifference in doses seen for large WR volumes decreased beyond 20% volume and disappeared below 10% volume. Dose to30%, 50% or 70% volumes of WR or RW was not correlated to the volume of WR or RW (p � 0.2). This was because thesame dose-volume constraints were placed on WR independent of variations in WR volumes between patients. However,smaller the volume of rectum, higher was the dose per cc of WR or RW and vice versa (p � 0.001). This was due to the factthat dose per cc takes into account of the variations in the volume of WR or RW between patients. The variation in dose receivedby 30% of WR varied by a factor 1.3 (29–38 Gy), whereas the volume of WR in cc varied by a factor of 7 (25–177 cc). Thismeans that while 30% of prescription dose was delivered to 7.5 cc of a rectum with 25 cc volume, the same dose was deliveredto 53 cc of rectum with a volume of 177 cc. Therefore, the dose delivered to 1 cc of rectum was related to the volume of wholerectum or rectal wall when expressed as dose per cc vs. volume.

Conclusions: Rectal volumes vary from patient to patient. Larger the volume of whole rectum, larger was also the volume ofRW. In IMRT, as in the case of 3D, patients with smaller rectal volumes would be at a higher risk for rectal toxicity becausedose delivered to per cc of rectum increases with the decrease in the volume of rectum (Fig. 2). Therefore, it is suggested that:1) dose per cc of rectum in addition to the doses to 30, 50 and 70% volumes may be used to predict rectal toxicity more reliablyand, 2) that the use of rectal wall volumes to express dose to rectum would be more accurate and reliable because rectal wallvolume changes due to rectal fillings would be minimal.

S591Proceedings of the 46th Annual ASTRO Meeting