Sensitivity study of an automated system for daily patient QA using EPID exit dose images.

The dosimetric consequences of errors in patient setup or beam delivery and anatomical changes are not readily known. A new product, PerFRACTION (Sun Nuclear Corporation), is designed to identify these errors by comparing the exit dose image measured on an electronic portal imaging device (EPID) from each field of each fraction to those from baseline fraction images. This work investigates the sensitivity of PerFRACTION to detect the deviation caused by these errors in a variety of realistic scenarios. Integrated EPID images were acquired in clinical mode and saved in ARIA. PerFRACTION automatically pulled the images into its database and performed the user-defined comparison. We induced errors of 1 mm and greater in jaw, multileaf collimator (MLC), and couch position, 1° and greater in collimation rotation (patient yaw), 0.5-1.5% in machine output, rail position, and setup errors of 1-2 mm shifts and 0.5-1° roll rotation. The planning techniques included static, intensity modulated radiation therapy (IMRT) and VMAT fields. Rectangular solid water phantom or anthropomorphic head phantom were used in the beam path in the delivery of some fields. PerFRACTION detected position errors of the jaws, MLC, and couch with an accuracy of better than 0.4 mm, and 0.5° for collimator rotation error and detected the machine output error within 0.2%. The rail position error resulted in PerFRACTION detected dose deviations up to 8% and 3% in open field and VMAT field delivery, respectively. PerFRACTION detected induced errors in IMRT fields within 2.2% of the gamma passing rate using an independent conventional analysis. Using an anthropomorphic phantom, setup errors as small as 1 mm and 0.5° were detected. Our work demonstrates that PerFRACTION, using integrated EPID image, is sensitive enough to identify positional, angular, and dosimetric errors.

A practical method for the reuse of nanoDot OSLDs and predicting sensitivities up to at least 7000 cGy.

PURPOSE: Optically stimulated luminescence dosimeter (OSLDs) are often used to make in vivo dose measurements. Most users calibrate the OSLDs using the software provided by the vendor which typically is intended for doses up to about 300 cGy with an uncertainty of ±5.5%. OSLDs that reach that dose are then discarded, and new ones are purchased and calibrated. A method has been developed for reusing OSLDs and predicting dose sensitivity up to at least 7000 cGy. MATERIALS AND METHODS: The nanoDot OSLDs used in this study were routinely used to do in vivo measurements for TBI patients. Instead of using the calibration program provided by the vendor, each nanoDot was bleached to about 100 counts (~0.1 cGy), then calibrated with 50 cGy to produce a sensitivity specific to each nanoDot prior to the patient measurement. NanoDots were read in the hardware mode and the sensitivity factor was applied manually to subsequent patient in-vivo TBI measurements. This was followed by bleaching prior to the next use. The changes of nanoDot sensitivity relative to accumulated dose were analyzed among nine nanoDots. In addition, a method to predict a nanoDot's sensitivity was investigated which aims to reduce the number of sensitivity calibrations while retaining dosimetric accuracy. RESULTS: Individual per-use nanodot calibrations were performed up to 7000 cGy for 37 clinical TBI patients. Among the nine nanoDots analyzed in this paper, the sensitivity vs accumulated dose decreased linearly up to about 3000 cGy, with linear fitting curve R2 values above 0.93. After 3000 cGy of accumulated dose, the sensitivity started to plateau and tended to increase by 6000 cGy, with 2nd order polynomial curve R2 values above 0.94. With this finding, an efficient and accurate method to predict nanoDots' sensitivities was developed. With the method applied to the nine OSLDs, a total of 127 sensitivities were predicted and retrospectively compared with measured sensitivities. The predicted sensitivities agreed with measured sensitivities within ±4.0% with an average of -0.8%. CONCLUSIONS: This study is the first to demonstrate the reuse of nanoDot OSLDs on numerous patients with accumulated dose up to 7000 cGy. Our nanoDot re-usage methodology is accurate, cost-saving and feasible. A time-saving method is also provided that allows a user to reuse a nanoDot with sensitivities predicted with better accuracy than the 5.5% value provided by the conventional batch calibration method.

Performance measure characterization for evaluating neuroimage segmentation algorithms.

Characterizing the performance of segmentation algorithms in brain images has been a persistent challenge due to the complexity of neuroanatomical structures, the quality of imagery and the requirement of accurate segmentation. There has been much interest in using the Jaccard and Dice similarity coefficients associated with Sensitivity and Specificity for evaluating the performance of segmentation algorithms. This paper addresses the essential characteristics of the fundamental performance measure coefficients adopted in evaluation frameworks. While exploring the properties of the Jaccard, Dice and Specificity coefficients, we propose new measure coefficients Conformity and Sensibility for evaluating image segmentation techniques. It is indicated that Conformity is more sensitive and rigorous than Jaccard and Dice in that it has better discrimination capabilities in detecting small variations in segmented images. Comparing to Specificity, Sensibility provides consistent and reliable evaluation scores without the incorporation of image background properties. The merits of the proposed coefficients are illustrated by extracting neuroanatomical structures in a wide variety of brain images using various segmentation techniques.

Statistical validation of a new helical tomotherapy patient transfer station.

The purpose of this work is to evaluate statistically the accuracy of a patient transfer station (PTS, TomoTherapy Inc., Madison, WI) that automatically converts one planning-station-generated treatment plan to another one with a different beam model. In our department we have installed 2 HI*ART tomotherapy systems, and patients often need to be transferred from one tomotherapy unit to the other. Thirty patients who underwent patient transfer between the two systems were evaluated. For each patient, dose differences between his/her original plan and PTS-transferred plan were evaluated by comparing doses at 10 randomly selected positions in his/her CT images. The Pearson indexes were calculated to analyze the relationship of the deviations to other parameters, which include absolute dose levels, sites (targets or normal tissues), dose accuracy of original plans and that of transferred plans. The dose accuracy of a treatment plan was determined by comparing delivered doses at the center of a 30 cm x 30 cm x 12 cm solid water phantom to planned doses at the same position. The calculated dose difference between original and transferred plans was, on average, 0.8% +/- 0.5%; the maximum deviation in absolute values was 1.9% in target volumes and 2.5% in normal organs. The errors generated during PTS-based transferring process were random and did not show correlation with other parameters. The PTS took less than 10 minutes to generate a backup plan that is much less than 2 hours that is approximately needed to create a duplicate plan manually. The results show that a PTS-transferred plan is an acceptable match to the original plan. With a physician's approval, a transferred plan is acceptable for treatment without the necessity of being revalidated in phantom. Thus far, all of our PTS plans have been approved by the treating physician without further optimization.

Validation of OSLD and a treatment planning system for surface dose determination in IMRT treatments.

PURPOSE: To evaluate the accuracy of skin dose determination for composite multibeam 3D conformal radiation therapy (3DCRT) and intensity modulated radiation therapy (IMRT) treatments using optically stimulated luminescent dosimeters (OSLDs) and Eclipse treatment planning system. METHODS: Surface doses measured by OSLDs in the buildup region for open field 6 MV beams, either perpendicular or oblique to the surface, were evaluated by comparing against dose measured by Markus Parallel Plate (PP) chamber, surface diodes, and calculated by Monte Carlo simulations. The accuracy of percent depth dose (PDD) calculation in the buildup region from the authors' Eclipse system (Version 10), which was precisely commissioned in the buildup region and was used with 1 mm calculation grid, was also evaluated by comparing to PP chamber measurements and Monte Carlo simulations. Finally, an anthropomorphic pelvic phantom was CT scanned with OSLDs in place at three locations. A planning target volume (PTV) was defined that extended close to the surface. Both an 8 beam 3DCRT and IMRT plan were generated in Eclipse. OSLDs were placed at the CT scanned reference locations to measure the skin doses and were compared to diode measurements and Eclipse calculations. Efforts were made to ensure that the dose comparison was done at the effective measurement points of each detector and corresponding locations in CT images. RESULTS: The depth of the effective measurement point is 0.8 mm for OSLD when used in the buildup region in a 6 MV beam and is 0.7 mm for the authors' surface diode. OSLDs and Eclipse system both agree well with Monte Carlo and/or Markus PP ion chamber and/or diode in buildup regions in 6 MV beams with normal or oblique incidence and across different field sizes. For the multiple beam 3DCRT plan and IMRT plans, the differences between OSLDs and Eclipse calculations on the surface of the anthropomorphic phantom were within 3% and distance-to-agreement less than 0.3 mm. CONCLUSIONS: The authors' experiment showed that OSLD is an accurate dosimeter for skin dose measurements in complex 3DCRT or IMRT plans. It also showed that an Eclipse system with accurate commissioning of the data in the buildup region and 1 mm calculation grid can calculate surface doses with high accuracy and has a potential to replace in vivo measurements.

Dosimetric study and verification of total body irradiation using helical tomotherapy and its comparison to extended SSD technique.

The American College of Radiology practice guideline for total body irradiation (TBI) requires a back-up treatment delivery system. This study investigates the development of helical tomotherapy (HT) for delivering TBI and compares it with conventional extended source-to-surface distance (X-SSD) technique. Four patients' head-to-thigh computed tomographic images were used in this study, with the target defined as the body volume without the left and right lungs. HT treatment plans with the standard TBI prescription (1.2 Gy/fx, 10 fractions) were generated and verified on phantoms. To compare HT plans with X-SSD treatment, the dose distribution of X-SSD technique was simulated using the Eclipse software. The average dose received by 90% of the target volume was 12.3 Gy (range, 12.2-12.4 Gy) for HT plans and 10.3 Gy (range, 10.08-10.58 Gy) for X-SSD plans (p < 0.001). The left and right lung median doses were 5.44 Gy and 5.40 Gy, respectively, for HT plans and 8.34 Gy and 8.95 Gy, respectively, for X-SSD treatment. The treatment planning time was comparable between the two methods. The beam delivery time of HT treatment was longer than X-SSD treatment. In conclusion, HT-based TBI plans have better dose coverage to the target and better dose sparing to the lungs compared with X-SSD technique, which applies dose compensators, lung blocks, and electron boosts. This study demonstrates that HT is possible for delivering TBI. Clinical validation of the feasibility of this approach would be of interest in the future.

Skull-stripping magnetic resonance brain images using a model-based level set.

The segmentation of brain tissue from nonbrain tissue in magnetic resonance (MR) images, commonly referred to as skull stripping, is an important image processing step in many neuroimage studies. A new mathematical algorithm, a model-based level set (MLS), was developed for controlling the evolution of the zero level curve that is implicitly embedded in the level set function. The evolution of the curve was controlled using two terms in the level set equation, whose values represented the forces that determined the speed of the evolving curve. The first force was derived from the mean curvature of the curve, and the second was designed to model the intensity characteristics of the cortex in MR images. The combination of these forces in a level set framework pushed or pulled the curve toward the brain surface. Quantitative evaluation of the MLS algorithm was performed by comparing the results of the MLS algorithm to those obtained using expert segmentation in 29 sets of pediatric brain MR images and 20 sets of young adult MR images. Another 48 sets of elderly adult MR images were used for qualitatively evaluating the algorithm. The MLS algorithm was also compared to two existing methods, the brain extraction tool (BET) and the brain surface extractor (BSE), using the data from the Internet brain segmentation repository (IBSR). The MLS algorithm provides robust skull-stripping results, making it a promising tool for use in large, multi-institutional, population-based neuroimaging studies.