Johan Degerman

Combined segmentation and tracking of neural stem-cells

In this paper we analyze neural stem/progenitor cells in an time-lapse image sequence. By using information about the previous positions of the cells, we are able to make a better selection of possible cells out of a collection of blob-like objects. As a blob detector we use Laplacian of Gaussian (LoG) filters at multiple scales, and the cell contours of the selected cells are segmented using dynamic programming. After the segmentation process the cells are tracked in the sequence using a combined nearest-neighbor and correlation matching technique. An evaluation of the system show that 95% of the cells were correctly segmented and tracked between consecutive frames.

An automatic system for in vitro cell migration studies

This paper describes a system for in vitro cell migration analysis. Adult neural stem/progenitor cells are studied using time‐lapse bright‐field microscopy and thereafter stained immunohistochemically to find and distinguish undifferentiated glial progenitor cells and cells having differentiated into type‐1 or type‐2 astrocytes. The cells are automatically segmented and tracked through the time‐lapse sequence. An extension to the Chan‐Vese Level Set segmentation algorithm, including two new terms for specialized growing and pruning, made it possible to resolve clustered cells, and reduced the tracking error by 65%. We used a custom‐built manual correction module to form a ground truth used as a reference for tracked cells that could be identified from the fluorescence staining. On average, the tracks were correct 95% of the time, using our new segmentation. The tracking, or association of segmented cells, was performed using a 2‐state Hidden Markov Model describing the random behaviour of the cells. By re‐estimating the motion model to conform with the segmented data we managed to reduce the number of tracking parameters to essentially only one. Upon characterization of the cell migration by the HMM state occupation function, it was found that glial progenitor cells were moving randomly 2/3 of the time, while the type‐2 astrocytes showed a directed movement 2/3 of the time. This finding indicates possibilities for cell‐type specific identification and cell sorting of live cells based on specific movement patterns in individual cell populations, which would have valuable applications in neurobiological research.

Time-lapse microscopy and image processing for stem cell research modeling cell migration

This paper presents hardware and software procedures for automated cell tracking and migration modeling. A time-lapse microscopy system equipped with a computer controllable motorized stage was developed. The performance of this stage was improved by incorporating software algorithms for stage motion displacement compensation and auto focus. The microscope is suitable for in-vitro stem cell studies and allows for multiple cell culture image sequence acquisition. This enables comparative studies concerning rate of cell splits, average cell motion velocity, cell motion as a function of cell sample density and many more. Several cell segmentation procedures are described as well as a cell tracking algorithm. Statistical methods for describing cell migration patterns are presented. In particular, the Hidden Markov Model (HMM) was investigated. Results indicate that if the cell motion can be described as a non-stationary stochastic process, then the HMM can adequately model aspects of its dynamic behavior.

Tracking neural stem cells in time-lapse microscopy image sequences

This paper describes an algorithm for tracking neural stem/progenitor cells in a time-lapse microscopy image sequence. The cells were segmented in a semiautomatic way using dynamic programming. Since the interesting cells were identified by fluorescent staining at the end of the sequence, the tracking was performed backwards. The number of detected cells varied throughout the sequence: cells could appear or disappear at the image boundaries or at cell clusters, some cells split, and the segmentation was not always correct. To solve this asymmetric assignment problem, a modified version of the auction algorithm by Bertsekas was used. The assignment weights were calculated based on distance, correlation and size between possible matching cells. Cell splits are of special interest, therefore tracks without a matching cell were divided into two groups: 1. Merging cells (splitting cells, moving forward in time) and 2. Non-merging cells. These groups were separated based on difference in size of the involved cells, and difference in image intensity of the contour and interior of the possibly merged cell. The tracking algorithm was evaluated using a sequence consisting of 57 images, each image containing approximately 50 cells. The evaluation showed that 99% of the cell-to-cell associations were correct. In most cases, only one association per track was incorrect so in total 55 out of 78 different tracks in the sequence were tracked correctly. Further improvements will be to apply interleaved segmentation and tracking to produce a more reliable segmentation as well as better tracking results.

A computational 3D model for reconstruction of neural stem cells in bright-field time-lapse microscopy

This paper describes a computational model for image formation of in-vitro adult hippocampal progenitor (AHP) cells, in bright-field time-lapse microscopy. Although this microscopymodality barely generates sufficient contrast for imaging translucent cells, we show that by using a stack of defocused image slices it is possible to extract position and shape of spherically shaped specimens, such as the AHP cells. This inverse problem was solved by modeling the physical objects and image formation system, and using an iterative nonlinear optimization algorithm to minimize the difference between the reconstructed and measured image stack. By assuming that the position and shape of the cells do not change significantly between two time instances, we can optimize these parameters using the previous time instance in a Bayesian estimation approach. The 3D reconstruction algorithm settings, such as focal sampling distance, and PSF, were calibrated using latex spheres of known size and refractive index. By using the residual between reconstructed and measured image intensities, we computed a peak signal-to-noise ratio (PSNR) to 28 dB for the sphere stack. A biological specimen analysis was done using an AHP cell, where reconstruction PSNR was 28 dB as well. The cell was immuno-histochemically stained and scanned in a confocal microscope, in order to compare our cell model to a ground truth. After convergence the modelled cell volume had an error of less than one percent.

Comparing a supervised and an unsupervised classification method for burst detection in neonatal EEG

Hidden Markov Models (HMM) and Support Vector Machines (SVM) using unsupervised and supervised training, respectively, were compared with respect to their ability to correctly classify burst and suppression in neonatal EEG. Each classifier was fed five feature signals extracted from EEG signals from six full term infants who had suffered from perinatal asphyxia. Visual inspection of the EEG by an experienced electroencephalographer was used as the gold standard when training the SVM, and for evaluating the performance of both methods. The results are presented as receiver operating characteristic (ROC) curves and quantified by the area under the curve (AUC). Our study show that the SVM and the HMM exhibit similar performance, despite their fundamental differences.

Time-Lapse Microscopy and Classification of in Vitro Cell Migration Using Hidden Markov Modeling

This paper describes a system for in vitro cell migration analysis. Adult neural stem/progenitor cells are studied using time-lapse microscopy and thereafter stained immunohistochemically to find and distinguish between undifferentiated glial progenitor cells and cells having differentiated into type-1 or type-2 astrocytes. The cells are automatically segmented and tracked throughout the time-lapse sequence. The evaluation showed that 88% of the cells were correctly segmented and tracked by the automatic system. Upon characterization of the cell migration by hidden Markov modeling, it was found that the motion of glial progenitor cells was random 2/3 of the time, while the type-2 astrocytes showed a directed movement 2/3 of the time. This finding indicates possibilities for cell-type specific identification and cell sorting of live cells based on specific movement patterns in individual cell populations, which will have valuable applications in neurobiological research

3D occupancy grid mapping using statistical radar models

We have developed a numerically efficient occupancy grid mapping method in three dimensions for automotive radar, where we take into account the radar measurement signal-to-noise ratio. The mapping performance, i.e. to estimate length, height, and in-between spacing of parked cars, is demonstrated as we use acquired data from a radar prototype developed in collaboration with Qamcom Research and Technology3. The radar has a unique antenna providing unambiguous azimuth and elevation for a wide field of view radar, covering -50° in both dimensions, making mapping in three dimensions feasible. Employing self-developed off-line radar signal processing on raw data, we extract SNR which is used together with a Swerling 1 model to compute the probability of detection for grid map update. Moreover, we present a novel very simplistic way of updating the grid as we use fast trilinear interpolation in the measurement domain, in which the grid spacing is uniform. Having mounted the radar in forward direction the EGO-vehicle drive parallel to four parked cars with different inter-spacing, and we manage to measure the distances within the error of the grid spacing, 0.2 m.

Statistical analysis of multi-channel detection using data from airborne AESA radar

We investigate the ground clutter homogeneity and target detection performance using airborne multi-channel AESA (Active Electronically Scanned Array) radar data from flight trials over the southern part of Sweden. The data sets consisted of clutter returns from both urban and rural areas. Multivariate normality tests indicate a significant difference between these two environments, and particularly the urban clutter does not fit the Gaussian signal model. Furthermore, we expose differences in terms of homogeneity as well. As homogeneity measure we employed the commonly used generalized inner product (GIP), and compared this to a homogeneity measure based on projection in the eigenspace of the covariance matrix. The focus of our interest was to examine the importance of screening the secondary data from non-homogeneities, when using adaptive target detection in AESA radar systems. To evaluate the non-homogeneity detection (NHD) performance we employed a synthetic target scheme. Our evaluation showed negligible improvement from NHD preprocessing, regardless of method, and hence we conclude that the clutter in our data sets is structurally homogeneous. However, in dense target scenarios it is crucial to screen for target contamination in secondary data, as we demonstrate on real data.