The wide availability, low radiation dose and short acquisition time of ConeBeam CT (CBCT) scans make them an attractive source of data for compiling databases of anatomical structures. However CBCT has higher noise and lower contrast than helical slice CT, which makes segmentation more challenging and the optimal methods are not yet known. This paper evaluates several methods of segmenting airway geometries (nares, nasal cavities and pharynx) from typical dental quality head and neck CBCT data. The nasal cavity has narrow and intricate passages and is separated from the paranasal sinuses by thin walls, making it is susceptible to either over-or under-segmentation. The upper airway was split into two: the nasal cavity and the pharyngeal region (nasopharynx to larynx). Each part was segmented using global thresholding, multi-step level-set, and region competition methods (the latter using thresholding, clustering and classification initialisation and edge attraction techniques). The segmented 3D surfaces were evaluated against a reference manual segmentation using distance-, overlap-and volume-based metrics. Global thresholding, multi-step level-set, and region competition all gave satisfactory results for the lower part of the airway (nasopharynx to larynx). Edge attraction failed completely. A semi-automatic region-growing segmentation with multi-thresholding (or classification) initialization offered the best quality segmentation. With some minimal manual editing, it resulted in an accurate upper airway model, as judged by the similarity and volumetric indices, while being the least time consuming of the semi-automatic methods, and relying the least on the operator's expertise.

BackgroundUnderstanding airflow through human airways is of importance in drug delivery and development of assisted breathing methods. In this work, we focus on development of a new method to obtain an averaged upper airway geometry from computed tomography (CT) scans of many individuals. This geometry can be used for air flow simulation. We examine the geometry resulting from a data set consisting of 26 airway scans. The methods used to achieve this include nasal cavity segmentation and a deformable template matching procedure.MethodsThe method uses CT scans of the nasal cavity of individuals to obtain a segmented mesh, and coronal cross-sections of this segmented mesh are taken. The cross-sections are processed to extract the nasal cavity, and then thinned (‘skeletonized’) representations of the airways are computed. A reference template is then deformed such that it lies on this thinned representation. The average of these deformations is used to obtain the average geometry. Our procedure tolerates a wider variety of nasal cavity geometries than earlier methods.ResultsTo assess the averaging method, key landmark points on each of the input scans as well as the output average geometry are located and compared with one another, showing good agreement. In addition, the cross-sectional area (CSA) profile of the nasal cavities of the input scans and average geometry are also computed, showing that the CSA of the average model falls within the variation of the population.ConclusionsThe use of a deformable template method for aligning and averaging the nasal cavity provides an improved, detailed geometry that is unavailable without using deformation.Electronic supplementary materialThe online version of this article (doi:10.1186/s12880-016-0154-8) contains supplementary material, which is available to authorized users.

Previously we developed an active contour method for segmenting and tracking cells in phase-contrast microscopy images. Our method is capable of fine-grained segmentation on noisy image sequences. In this paper, we improve the active contour segmentation model to provide better accuracy, by selectively identifying areas of the contour with low confidence and removing them. The method is applied to HMEC-1 cells (human microvascular endothelial cells). The segmentation provided by the method is quantitavely compared with manually-drawn contours, showing close fit and capability to 'lock' on to cell boundaries for hundreds of frames.

No abstract

The problem of automated segmenting and tracking of the outlines of cells in microscope images is the subject of active research. While great progress has been made on recognizing cells that are of high contrast and of predictable shape, many situations arise in practice where these properties do not exist and thus many interesting potential studies - such as the migration patterns of astrocytes to scratch wounds - have been relegated to being largely qualitative in nature. Here we analyse a select number of recent developments in this area, and offer an algorithm based on parametric active contours and formulated by taking into account cell movement dynamics. This Cell-Derived Active Contour (CDAC) method is compared with two state-of-the-art segmentation methods for phase-contrast microscopy. Specifically, we tackle a very difficult segmentation problem: human astrocytes that are very large, thin, and irregularly-shaped. We demonstrate quantitatively better results for CDAC as compared to similar segmentation methods, and we also demonstrate the reliable segmentation of qualitatively different data sets that were not possible using existing methods. We believe this new method will enable new and improved automatic cell migration and movement studies to be made.