Purpose: To implement variable treatment planning margins to account for regional variations of setup uncertainties in Head‐and‐Neck cancer radiotherapy. Method and Materials: Five bony landmarks (C2, mandible, C5, caudal C7, jugular notch) were identified from previous studies. At each point on the CTV, a variable margin was calculated as the weighted average of margins at the landmark points, with the weight determined by a Gaussian falloff function of the distance between the current location and landmark point. Ten CT images of a Head‐and‐Neck cancer patient were used to test this variable margin strategy, in comparison with the standard global margin expansion method. We examined the overlapping of CTV and PTV in these 10 actual setup positions to evaluate the effectiveness of the design strategy. Results: For the variable margin approach using a margin of 2.5mm at the reference landmark C2, an averaged 99.20% of CTV is enclosed within the PTV, while a constant 2.5mm margin expansion results to 97.88% coverage. With reference margin reduced to 2.0mm, the variable margin approach has an averaged coverage of 97.84%, similar to that of constant 2.5mm margin expansion; however, it has a smaller PTV volume than the constant 2.5mm margin design. Paired t‐test on samples from 10 treatment fractions shows no significant difference on CTV coverage between variable 2.0mm margin and constant 2.5 margin approaches (p=0.054), but the non‐overlapped PTV portion is significantly smaller for the variable 2.0mm margin approach than the constant 2.5mm margin approach (p<0.0001). Our result also shows a better CTV coverage when using the variable margin approach in the lower neck area where a larger setup error normally occurs. Conclusion: We implemented a variable margin approach to account for regional variations of setup errors for Head‐and‐Neck cancer radiotherapy, and demonstrated its superiority over the global constant margin expansion approach.
Purpose: To develop and test two independent algorithms that automatically create the photon treatment fields for a four‐field box beam arrangement, a common treatment technique for cervical cancer in low‐ and middle‐income countries. Methods: Two algorithms were developed and integrated into Eclipse using its Advanced Programming Interface:3D Method: We automatically segment bony anatomy on CT using an in‐house multi‐atlas contouring tool and project the structures into the beam's‐eye‐view. We identify anatomical landmarks on the projections to define the field apertures. 2D Method: We generate DRRs for all four beams. An atlas of DRRs for six standard patients with corresponding field apertures are deformably registered to the test patient DRRs. The set of deformed atlas apertures are fitted to an expected shape to define the final apertures. Both algorithms were tested on 39 patient CTs, and the resulting treatment fields were scored by a radiation oncologist. We also investigated the feasibility of using one algorithm as an independent check of the other algorithm. Results: 96% of the 3D‐Method‐generated fields and 79% of the 2D‐method‐generated fields were scored acceptable for treatment (“Per Protocol” or “Acceptable Variation”). The 3D Method generated more fields scored “Per Protocol” than the 2D Method (62% versus 17%). The 4% of the 3D‐Method‐generated fields that were scored “Unacceptable Deviation” were all due to an improper L5 vertebra contour resulting in an unacceptable superior jaw position. When these same patients were planned with the 2D method, the superior jaw was acceptable, suggesting that the 2D method can be used to independently check the 3D method. Conclusion: Our results show that our 3D Method is feasible for automatically generating cervical treatment fields. Furthermore, the 2D Method can serve as an automatic, independent check of the automatically‐generated treatment fields. These algorithms will be implemented for fully automated cervical treatment planning.
Purpose: To investigate and validate the use of an independent deformable‐based contouring algorithm for automatic verification of auto‐contoured structures in the head and neck towards fully automated treatment planning. Methods: Two independent automatic contouring algorithms [(1) Eclipse's Smart Segmentation followed by pixel‐wise majority voting, (2) an in‐house multi‐atlas based method] were used to create contours of 6 normal structures of 10 head‐and‐neck patients. After rating by a radiation oncologist, the higher performing algorithm was selected as the primary contouring method, the other used for automatic verification of the primary. To determine the ability of the verification algorithm to detect incorrect contours, contours from the primary method were shifted from 0.5 to 2cm. Using a logit model the structure‐specific minimum detectable shift was identified. The models were then applied to a set of twenty different patients and the sensitivity and specificity of the models verified. Results: Per physician rating, the multi‐atlas method (4.8/5 point scale, with 3 rated as generally acceptable for planning purposes) was selected as primary and the Eclipse‐based method (3.5/5) for verification. Mean distance to agreement and true positive rate were selected as covariates in an optimized logit model. These models, when applied to a group of twenty different patients, indicated that shifts could be detected at 0.5cm (brain), 0.75cm (mandible, cord), 1cm (brainstem, cochlea), or 1.25cm (parotid), with sensitivity and specificity greater than 0.95. If sensitivity and specificity constraints are reduced to 0.9, detectable shifts of mandible and brainstem were reduced by 0.25cm. These shifts represent additional safety margins which might be considered if auto‐contours are used for automatic treatment planning without physician review. Conclusion: Automatically contoured structures can be automatically verified. This fully automated process could be used to flag auto‐contours for special review or used with safety margins in a fully automatic treatment planning system.
Purpose: To implement soft‐tissue image‐guided proton therapy using inroom mobile CT. Methods: Anthropomorphic phantom was first used to determine the setup accuracy using in‐ room mobile CT. Laser and bbs were used for the initial setup (marked isocenter). CT data was then acquired with in‐room mobile CT (daily CT). The shift between the marked isocenter and the planned isocenter (final isocenter) was determined from the daily CT using in‐house Computer Assisted Targeting (CAT) software. Orthogonal DRRs of the day was also generated from the daily CT. The phantom was then transferred on the treatment couch top to the treatment machine using a transportation system, and again aligned to the marked isocenter. Couch shifts were made to align the phantom to the final isocenter using the shifts as determined using the CAT software, and verified using orthogonal X‐ray images with the daily DRRs. Results: Phantom data suggests that following the setup procedure as described above, targeting accuracy could be within 1 mm. Patient data are being acquired and analyzed. Conclusion: In‐room mobile CT is capable of providing soft‐tissue image‐guided proton therapy.
Purpose: To develop and validate a prediction model using radiomics features extracted from MR images to distinguish radiation necrosis from tumor progression for brain metastases treated with Gamma knife radiosurgery. Methods: The images used to develop the model were T1 post‐contrast MR scans from 71 patients who had had pathologic confirmation of necrosis or progression; 1 lesion was identified per patient (17 necrosis and 54 progression). Radiomics features were extracted from 2 images at 2 time points per patient, both obtained prior to resection. Each lesion was manually contoured on each image, and 282 radiomics features were calculated for each lesion. The correlation for each radiomics feature between two time points was calculated within each group to identify a subset of features with distinct values between two groups. The delta of this subset of radiomics features, characterizing changes from the earlier time to the later one, was included as a covariate to build a prediction model using support vector machines with a cubic polynomial kernel function. The model was evaluated with a 10‐fold cross‐validation. Results: Forty radiomics features were selected based on consistent correlation values of approximately 0 for the necrosis group and >0.2 for the progression group. In performing the 10‐fold cross‐validation, we narrowed this number down to 11 delta radiomics features for the model. This 11‐delta‐feature model showed an overall prediction accuracy of 83.1%, with a true positive rate of 58.8% in predicting necrosis and 90.7% for predicting tumor progression. The area under the curve for the prediction model was 0.79. Conclusion: These delta radiomics features extracted from MR scans showed potential for distinguishing radiation necrosis from tumor progression. This tool may be a useful, noninvasive means of determining the status of an enlarging lesion after radiosurgery, aiding decision‐making regarding surgical resection versus conservative medical management.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.