Automatic Segmentation of Head and Neck CT Images by GPU-Accelerated Multi-atlas Fusion

Han, Xiao; Han, Xiaodong; Hibbard, Lyndon S.; O’Connell, Nicolette; Willcut, Virgil

doi:10.54294/5oitxb

Cited by 17 publications

(3 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Based on the encouraging results which we have obtained with this approach for the thyroid and with a model-based approach for the lymph node regions (Chen et al 2010) we are planning a multi-rater validation study for the thyroid, the lymph node regions and the parotid. Early results which we have obtained with the parotid indicate that a model-based approach as the one we have used for the lymph node regions may be better than multi-atlas-based approaches as proposed by Ramus and Malandain (2010), Yang et al (2010) or Han et al (2010).…”

Section: Discussionmentioning

confidence: 91%

Evaluation of multiple-atlas-based strategies for segmentation of the thyroid gland in head and neck CT images for IMRT

Chen

Niermann²,

Deeley

et al. 2011

Phys. Med. Biol.

View full text Add to dashboard Cite

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.

show abstract

Section: Discussionmentioning

confidence: 91%

Evaluation of multiple-atlas-based strategies for segmentation of the thyroid gland in head and neck CT images for IMRT

Chen

Niermann²,

Deeley

et al. 2011

Phys. Med. Biol.

View full text Add to dashboard Cite

show abstract

“…Briefly, the intra-patient algorithm of ADMIRE performs a block-wise non-linear registration to get a robust initial alignment, followed by a dense localcorrelation-coefficient (LCC) based deformable registration to get the final deformable vector field (DVF). This tool has been reported and evaluated in several international challenges of head & neck and lung patients DIR with highranking results (18)(19)(20)(21). Our institution also comprehensively evaluated for HNC patients with expert-delineated contours as ground truth, including seven OARs [brain stem, cord, left and right (L/R) parotids, L/R submandibular gland and mandible] ( 22).…”

Section: Sct Generation From Cbctmentioning

confidence: 99%

Establish a comprehensive plan robustness evaluation platform for intensity modulated proton therapy for bilateral head and neck cancer based on the daily cone-beam computed tomography: a dose accumulation study between the PTV based IMPT (PTV-IMPT) and CTV based robust optimized IMPT (ro-IMPT) planning strategies

Wang¹,

An²,

Li³

et al. 2022

Ther Radiol Oncol

View full text Add to dashboard Cite

Background: A comprehensive plan evaluation platform was established based on the daily cone-beam computed tomography (CBCT) to assess the treatment robustness quality between planning target volumebased intensity modulated proton therapy (PTV-IMPT) and clinical target volume (CTV)-based robust optimized IMPT (ro-IMPT) planning strategies in bilateral head and neck cancer (HNC) treatment.Methods: Nine bilateral HNC patients' CT structure sets were used in this study. Daily CBCTs were converted into synthetic-CT (sCT) for dose reconstruction. The accuracy of the proton dose calculation in sCT is cross-validated via the same day's verification-CT sim (vCT) with 3D gamma index comparison.PTV-IMPT and ro-IMPT were generated on the initial planning CT (pCT). CTV high-risk volume (CTV_ high) received 70 Gy and CTV low/intermediate-risk (CTV_low) received 60 Gy. For PTV-IMPT, the PTVs were expanded 3 mm from the CTV; For ro-IMPT, robust optimization used a 3 mm setup and 3.5% range uncertainties. Dose accumulations were then calculated on the 35 sets of daily sCT, and the target coverages were compared to the initial plans.Results: The 3D gamma index dose comparison (3 mm/3%) showed an average pass rate of 98.2%±1.5% comparing the same day's pair of sCT and vCT with both plans (total 38 pairs). Through the dose accumulation of 35 treatment fractions, the PTV-IMPT plan group's mean V100 of CTV_high/CTV_low coverage degraded to 80.70%/85.73% compared to 96.72%/96.13% of the ro-IMPT group (P<0.002). One patient did have suboptimal coverage (CTV_low <90%) even with ro-IMPT. Significant weight loss was noted for this patient during the treatment course (>5 lbs).Conclusions: A comprehensive plan robustness evaluation platform based on the CBCT is established in our clinical workflow and enables dose accumulation and plan robustness evaluation on a daily basis. ro-IMPT demonstrated an optimal planning strategy over PTV-IMPT for bilateral HNC treatment. However, special cautions are needed for patients with significant weight or geometry changes.

show abstract

“…Most applications of the multi‐atlas segmentation (MAS) approach 10 lie in automatic structural segmentation in brain magnetic resonance imaging (MRI) data, 17–23 while MAS has also shown usefulness in the segmentation of objects in images of different modalities 24 such as MRI, computed tomography (CT), ultrasonography, and in different body regions such as head and neck (H&N), 25–30 thorax, 31 abdomen, 32,33 and multiple body regions 34,35 . An implicit assumption in the MAS method is that there should be a large enough atlas set, which has complete and perfect segmentations of target objects and which covers the object shape and geographic layout patterns and image intensity appearance patterns of the whole population of subjects under study.…”

Section: Introductionmentioning

confidence: 99%

SOMA: Subject‐, object‐, and modality‐adapted precision atlas approach for automatic anatomy recognition and delineation in medical images

Udupa

Odhner

et al. 2021

Medical Physics

View full text Add to dashboard Cite

In the multi-atlas segmentation (MAS) method, a large enough atlas set, which can cover the complete spectrum of the whole population pattern of the target object will benefit the segmentation quality. However, the difficulty in obtaining and generating such a large set of atlases and the computational burden required in the segmentation procedure make this approach impractical. In this paper, we propose a method called SOMA to select subject-, object-, and modality-adapted precision atlases for automatic anatomy recognition in medical images with pathology, following the idea that different regions of the target object in a novel image can be recognized by different atlases with regionally best similarity, so that effective atlases have no need to be globally similar to the target subject and also have no need to be overall similar to the target object. Methods: The SOMA method consists of three main components: atlas building, object recognition, and object delineation. Considering the computational complexity, we utilize an all-to-template strategy to align all images to the same image space belonging to the root image determined by the minimum spanning tree (MST) strategy among a subset of radiologically near-normal images. The object recognition process is composed of two stages: rough recognition and refined recognition. In rough recognition, subimage matching is conducted between the test image and each image of the whole atlas set, and only the atlas corresponding to the best-matched subimage contributes to the recognition map regionally. The frequency of best match for each atlas is recorded by a counter, and the atlases with the highest frequencies are selected as the precision atlases. In refined recognition, only the precision atlases are examined, and the subimage matching is conducted in a nonlocal manner of searching to further increase the accuracy of boundary matching. Delineation is based on a U-net-based deep learning network, where the original gray scale image together with the fuzzy map from refined recognition compose a two-channel input to the network, and the output is a segmentation map of the target object. Results: Experiments are conducted on computed tomography (CT) images with different qualities in two body regions -head and neck (H&N) and thorax, from 298 subjects with nine objects and 241 subjects with six objects, respectively. Most objects achieve a localization error within two voxels after refined recognition, with marked improvement in localization accuracy from rough to refined recognition of 0.6-3 mm in H&N and 0.8-4.9 mm in thorax, and also in delineation accuracy (Dice coefficient) from refined recognition to delineation of 0.01-0.11 in H&N and 0.01-0.18 in thorax. Conclusions:The SOMA method shows high accuracy and robustness in anatomy recognition and delineation. The improvements from rough to refined 7806

show abstract

Automatic Segmentation of Head and Neck CT Images by GPU-Accelerated Multi-atlas Fusion

Cited by 17 publications

References 12 publications

Evaluation of multiple-atlas-based strategies for segmentation of the thyroid gland in head and neck CT images for IMRT

Evaluation of multiple-atlas-based strategies for segmentation of the thyroid gland in head and neck CT images for IMRT

SOMA: Subject‐, object‐, and modality‐adapted precision atlas approach for automatic anatomy recognition and delineation in medical images

Contact Info

Product

Resources

About