Evaluation of multiatlas label fusion forin vivomagnetic resonance imaging orbital segmentation

Abstract

Multi-atlas methods have been successful for brain segmentation, but their application to smaller anatomies remains relatively unexplored. We evaluate 7 statistical and voting-based label fusion algorithms (and 6 additional variants) to segment the optic nerves, eye globes and chiasm. For non-local STAPLE, we evaluate different intensity similarity measures (including mean square difference, locally normalized cross correlation, and a hybrid approach). Each algorithm is evaluated in terms of the Dice overlap and symmetric surface distance metrics. Finally, we evaluate refinement of label fusion results using a learning based correction method for consistent bias correction and Markov random field regularization. The multi-atlas labeling pipelines were evaluated on a cohort of 35 subjects including both healthy controls and patients. Across all three structures, NLSS with a mixed weighting type provided the most consistent results; for the optic nerve NLSS resulted in a median Dice similarity coefficient of 0.81, mean surface distance of 0.41 mm and Hausdorff distance 2.18 mm for the optic nerves. Joint label fusion resulted in slightly superior median performance for the optic nerves (0.82, 0.39 mm and 2.15 mm), but slightly worse on the globes. The fully automated multi-atlas labeling approach provides robust segmentations of orbital structures on MRI even in patients for whom significant atrophy (optic nerve head drusen) or inflammation (multiple sclerosis) is present.

Multi-atlas segmentation of biomedical images: A survey

“…Complex, more advanced methods can pay off and should be something we continue to work on. However, this endeavor critically depends on a proper evaluation of the methods, as demonstrated in some recent efforts (Rueda et al, 2014; Menze et al, 2014; Panda et al, 2014; Goksel et al, 2014). Going forward, a grand challenge of biomedical image segmentation will be to establish standardized datasets and performance evaluation metrics to be used to objectively compare various segmentation algorithms, including MAS-based techniques.…”

Structural-functional relationships between eye orbital imaging biomarkers and clinical visual assessments

“…Segmentation of the MRI data for computation of image-derived anatomical metrics was based off a previously described multi-atlas segmentation method [2, 3], which automatically segments the optic nerves (including the CSF sheaths), extraocular rectus muscles, eye globes, and orbital fat. This method uses 20 manually labeled atlas images, which include healthy controls as well as ON head drusen, optic neuritis and multiple sclerosis (MS) patients.…”

“…An experienced undergraduate manually labeled approximately 20 subjects of each MRI and CT imaging modality. Then, multi-atlas segmentation was performed to segment the extraocular rectus muscles, eye globes, optic nerves, and orbital fat [2,3]. Twenty-one different structural metrics were then calculated from the segmentation pipelines.…”

Development of a reliable method for orbit segmentation & measuring

“…Lamecker et al present in [13] a Statistical Shape Models (SSM) to model shape variety and to allow the robust division of the orbit into six predefined parts. Another approach is using an atlas [14]. Measuring of the orbital volume is possible with all segmentation approaches, an automated determination of more parameters has not been conducted.…”