Fast and robust segmentation of spherical particles in volumetric data sets from brightfield microscopy

Abstract

In this article we present an approach for a precise segmentation of spherical particles in transmitted light image stacks. A main goal was its fast operation and a high robustness to occlusions and agglomerations of the particles. The system is based on a voting procedure that finds the centers and radii of the particles and a subsequent precise segmentation with an active contour approach. To meet the demands of an online pollenmonitor for high speed and low memory consumption a multi-scale approach was applied. The proposed techniques successfully segmented the pollen grains in a vast amount of different air samples (about 2.7TB of raw data). The results on one of the most cluttered samples are presented in this paper.

A 3D Active Surface Model for the Accurate Segmentation of Drosophila Schneider Cell Nuclei and Nucleoli

“…every surface point can be reached from the detected center c. As proposed in [2] for the 2D case in pollen segmentation, we then used a projection of the dataset gradients onto radial vectors pointing away from the detected center (∇I radial )(x) = (∇I)(x) , x−c ||x−c|| , thus reducing the influence of vectors pointing in other directions. Additionally, as done in [2], those vectors originating from darker inner structures and thus pointing outwards were set to zero length. The resulting gradient image contains by far less gradients corresponding to structures other than the nucleus, but the vectors set to zero length still cause problems in the next step.…”

“…The resulting gradient image contains by far less gradients corresponding to structures other than the nucleus, but the vectors set to zero length still cause problems in the next step. Instead of applying the Canny edge detector as it was done in [2], we directly use the resulting gradient magnitude as edge information. We compute the gradients of this edge image and, to get rid of the gradients now caused by the zeromagnitude regions, we use the radial projection of these gradients.…”

“…First, the nuclei are detected by the generalized Hough-Transform [1] for the detection of spherical objects, as it has been used e.g. in [3] for the 3D detection of Arabidopsis thaliana root nuclei in confocal microscopic data and in [2] for the 2D detection of pollen grains in brightfield data. The detection method will be presented in section 3.…”

Detection of bubbles as concentric circular arrangements

“…The Randomized and Probabilistic HTs utilize the random sampling techniques to sample the hypothesis of the circles from the edge pixels and remove verified hypotheses from the image to accelerate the process. HT and its modifications have been utilized in several applications, e.g., segmentation of spherical particles in transmitted light image stacks [23], segmentation of bubbles and drops in complex dispersions in bioreactors [28], and detection of circular objects in pulsative medical video [22].…”

Image analysis of Arabidopsis trichome patterning in 4D confocal datasets

“…This filter serves as an edge detector that is robust to noise, since also the neighbours are required to be local maxima. A similar technique was already successfully applied in [4]. Robustness to intra-as well as inter-subject variations is achieved using a scale-space approach by applying the filter to image X convolved with Gaussian kernels of widths in the range of [6μm, .…”