No abstract

The purpose of this work was to characterize expert variation in segmentation of intracranial structures pertinent to radiation therapy, and to assess a registration-driven atlas-based segmentation algorithm in that context. Eight experts were recruited to segment the brainstem, optic chiasm, optic nerves, and eyes, of 20 patients who underwent therapy for large space-occupying tumors. Performance variability was assessed through three geometric measures: volume, Dice similarity coefficient, and Euclidean distance. In addition, two simulated ground truth segmentations were calculated via the simultaneous truth and performance level estimation (STAPLE) algorithm and a novel application of probability maps. The experts and automatic system were found to generate structures of similar volume, though the experts exhibited higher variation with respect to tubular structures. No difference was found between the mean Dice coefficient (DSC) of the automatic and expert delineations as a group at a 5% significance level over all cases and organs. The larger structures of the brainstem and eyes exhibited mean DSC of approximately 0.8–0.9, whereas the tubular chiasm and nerves were lower, approximately 0.4–0.5. Similarly low DSC have been reported previously without the context of several experts and patient volumes. This study, however, provides evidence that experts are similarly challenged. The average maximum distances (maximum inside, maximum outside) from a simulated ground truth ranged from (−4.3, +5.4) mm for the automatic system to (−3.9, +7.5) mm for the experts considered as a group. Over all the structures in a rank of true positive rates at a 2 mm threshold from the simulated ground truth, the automatic system ranked second of the nine raters. This work underscores the need for large scale studies utilizing statistically robust numbers of patients and experts in evaluating quality of automatic algorithms.

The purpose of this study is to assess the real target dose coverage when radiation treatments were delivered to lung cancer patients based on treatment planning according to the RTOG-0236 Protocol. We compare calculated dosimetric results between the more accurate anisotropic analytical algorithm (AAA) and the pencil beam algorithm for stereotactic body radiation therapy treatment planning in lung cancer. Ten patients with non-small cell lung cancer were given 60 Gy in three fractions using 6 and 10 MV beams with 8-10 fields. The patients were chosen in accordance with the lung RTOG-0236 protocol. The dose calculations were performed using the pencil beam algorithm with no heterogeneity corrections (PB-NC) and then recalculated with the pencil beam with modified Batho heterogeneity corrections (PB-MB) and the AAA using an identical beam setup and monitor units. The differences in calculated dose to 95% or 99% of the PTV, between using the PB-NC and the AAA, were within 10% of prescribed dose (60 Gy). However, the minimum dose to 95% and 99% of PTV calculated using the PB-MB were consistently overestimated by up to 40% and 36% of the prescribed dose, respectively, compared to that calculated by the AAA. Using the AAA as reference, the calculated maximum doses were underestimated by up to 27% using the PB-NC and overestimated by 19% using the PB-MB. The calculations of dose to lung from PB-NC generally agree with that of AAA except in the small high-dose region where PB-NC underestimates. The calculated dose distributions near the interface using the AAA agree with those from Monte Carlo calculations as well as measured values. This study indicates that the real minimum PTV dose coverage cannot be guaranteed when the PB-NC is used to calculate the monitor unit settings in dose prescriptions.

Segmenting the thyroid gland in head and neck CT images is of vital clinical significance in designing intensity-modulated radiation therapy (IMRT) treatment plans. In this work, we evaluate and compare several multiple-atlas-based methods to segment this structure. Using the most robust method, we generate automatic segmentations for the thyroid gland and study their clinical applicability. The various methods we evaluate range from selecting one single atlas based on one of three similarity measures, to combining the segmentation results obtained with several atlases and weighting their contribution using techniques including a simple majority vote rule, a technique called STAPLE that is widely used in the medical imaging literature, and the similarity between the atlas and the volume to be segmented. We show that the best results are obtained when several atlases are combined and their contributions are weighted with a measure of similarity between each atlas and the volume to be segmented. We also show that with our data set, STAPLE does not always lead to the best results. Automatic segmentations generated by the combination method using the correlation coefficient (CC) between the deformed atlas and the patient volume, which is the most accurate and robust method we evaluated, are presented to a physician as 2D contours and modified to meet clinical requirements. It is shown that about 40% of the contours of the left thyroid and about 42% of the right thyroid can be used directly. An additional 21% on the left and 24% on the right require only minimal modification. The amount and the location of the modifications are qualitatively and quantitatively assessed. We demonstrate that, although challenged by large inter-subject anatomical discrepancy, atlas-based segmentation of the thyroid gland in IMRT CT images is feasible by involving multiple atlases. The results show that a weighted combination of segmentations by atlases using the CC as the similarity measure slightly outperforms standard combination methods, e.g., the majority vote rule and STAPLE, as well as methods selecting one single most similar atlas. Results we have obtained suggest that using our contours as initial contours to be edited has clinical value.

Purpose: Intensity-modulated radiation therapy ͑IMRT͒ is the state of the art technique for head and neck cancer treatment. It requires precise delineation of the target to be treated and structures to be spared, which is currently done manually. The process is a time-consuming task of which the delineation of lymph node regions is often the longest step. Atlas-based delineation has been proposed as an alternative, but, in the authors' experience, this approach is not accurate enough for routine clinical use. Here, the authors improve atlas-based segmentation results obtained for level II-IV lymph node regions using an active shape model ͑ASM͒ approach. Methods: An average image volume was first created from a set of head and neck patient images with minimally enlarged nodes. The average image volume was then registered using affine, global, and local nonrigid transformations to the other volumes to establish a correspondence between surface points in the atlas and surface points in each of the other volumes. Once the correspondence was established, the ASMs were created for each node level. The models were then used to first constrain the results obtained with an atlas-based approach and then to iteratively refine the solution. Results: The method was evaluated through a leave-one-out experiment. The ASM-and atlas-based segmentations were compared to manual delineations via the Dice similarity coefficient ͑DSC͒ for volume overlap and the Euclidean distance between manual and automatic 3D surfaces. The mean DSC value obtained with the ASM-based approach is 10.7% higher than with the atlas-based approach; the mean and median surface errors were decreased by 13.6% and 12.0%, respectively. Conclusions: The ASM approach is effective in reducing segmentation errors in areas of low CT contrast where purely atlas-based methods are challenged. Statistical analysis shows that the improvements brought by this approach are significant.