Christopher S. Bjornsson

A multi‐model approach to simultaneous segmentation and classification of heterogeneous populations of cell nuclei in 3D confocal microscope images

Automated segmentation and morphometry of fluorescently labeled cell nuclei in batches of 3D confocal stacks is essential for quantitative studies. Model‐based segmentation algorithms are attractive due to their robustness. Previous methods incorporated a single nuclear model. This is a limitation for tissues containing multiple cell types with different nuclear features. Improved segmentation for such tissues requires algorithms that permit multiple models to be used simultaneously. This requires a tight integration of classification and segmentation algorithms. Two or more nuclear models are constructed semiautomatically from user‐provided training examples. Starting with an initial over‐segmentation produced by a gradient‐weighted watershed algorithm, a hierarchical fragment merging tree rooted at each object is built. Linear discriminant analysis is used to classify each candidate using multiple object models. On the basis of the selected class, a Bayesian score is computed. Fragment merging decisions are made by comparing the score with that of other candidates, and the scores of constituent fragments of each candidate. The overall segmentation accuracy was 93.7% and classification accuracy was 93.5%, respectively, on a diverse collection of images drawn from five different regions of the rat brain. The multi‐model method was found to achieve high accuracy on nuclear segmentation and classification by correctly resolving ambiguities in clustered regions containing heterogeneous cell populations.

Model neural prostheses with integrated microfluidics: a potential intervention strategy for controlling reactive cell and tissue responses

Model silicon intracortical probes with microfluidic channels were fabricated and tested to examine the feasibility of using diffusion-mediated delivery to deliver therapeutic agents into the volume of tissue exhibiting reactive responses to implanted devices. Three-dimensional probe structures with microfluidic channels were fabricated using surface micromachining and deep reactive ion etching (DRIE) techniques. In vitro functional tests of devices were performed using fluorescence microscopy to record the transient release of Texas Red labeled transferrin (TR-transferrin) and dextran (TR-dextran) from the microchannels into 1% w/v agarose gel. In vivo performance was characterized by inserting devices loaded with TR-transferrin into the premotor cortex of adult male rats. Brain sections were imaged using confocal microscopy. Diffusion of TR-transferrin into the extracellular space and uptake by cells up to 400 /spl mu/m from the implantation site was observed in brain slices taken 1 h postinsertion. The reactive tissue volume, as indicated by the presence of phosphorylated mitogen-activated protein kinases (MAPKs), was characterized using immunohistochemistry and confocal microscopy. The reactive tissue volume extended 600, 800, and 400 /spl mu/m radially from the implantation site at 1 h, 24 h, and 6 weeks following insertion, respectively. These results indicate that diffusion-mediated delivery can be part of an effective intervention strategy for the treatment of reactive tissue responses around chronically implanted intracortical probes.

The Choroid Plexus and Cerebrospinal Fluid: Emerging Roles in Development, Disease, and Therapy

Although universally recognized as the source of cerebrospinal fluid (CSF), the choroid plexus (ChP) has been one of the most understudied tissues in neuroscience. The reasons for this are multiple and varied, including historical perceptions about passive and permissive roles for the ChP, experimental issues, and lack of clinical salience. However, recent work on the ChP and instructive signals in the CSF have sparked new hypotheses about how the ChP and CSF provide unexpected means for regulating nervous system structure and function in health and disease, as well as new ChP-based therapeutic approaches using pluripotent stem cell technology. This minisymposium combines new and established investigators to capture some of the newfound excitement surrounding the ChP-CSF system.

It Takes a Village: Constructing the Neurogenic Niche

Although many features of neurogenesis during development and in the adult are intrinsic to the neurogenic cells themselves, the role of the microenvironment is irrefutable. The neurogenic niche is a melting pot of cells and factors that influence CNS development. How do the diverse elements assemble and when? How does the niche change structurally and functionally during embryogenesis and in adulthood? In this review, we focus on the impact of non-neural cells that participate in the neurogenic niche, highlighting how cells of different embryonic origins influence this critical germinal space.

Automatic Summarization of Changes in Biological Image Sequences Using Algorithmic Information Theory

An algorithmic information-theoretic method is presented for object-level summarization of meaningful changes in image sequences. Object extraction and tracking data are represented as an attributed tracking graph (ATG). Time courses of object states are compared using an adaptive information distance measure, aided by a closed-form multidimensional quantization. The notion of meaningful summarization is captured by using the gap statistic to estimate the randomness deficiency from algorithmic statistics. The summary is the clustering result and feature subset that maximize the gap statistic. This approach was validated on four bioimaging applications: 1) It was applied to a synthetic data set containing two populations of cells differing in the rate of growth, for which it correctly identified the two populations and the single feature out of 23 that separated them; 2) it was applied to 59 movies of three types of neuroprosthetic devices being inserted in the brain tissue at three speeds each, for which it correctly identified insertion speed as the primary factor affecting tissue strain; 3) when applied to movies of cultured neural progenitor cells, it correctly distinguished neurons from progenitors without requiring the use of a fixative stain; and 4) when analyzing intracellular molecular transport in cultured neurons undergoing axon specification, it automatically confirmed the role of kinesins in axon specification.

Robust Adaptive 3-D Segmentation of Vessel Laminae From Fluorescence Confocal Microscope Images and Parallel GPU Implementation

This paper presents robust 3-D algorithms to segment vasculature that is imaged by labeling laminae, rather than the lumenal volume. The signal is weak, sparse, noisy, nonuniform, low-contrast, and exhibits gaps and spectral artifacts, so adaptive thresholding and Hessian filtering based methods are not effective. The structure deviates from a tubular geometry, so tracing algorithms are not effective. We propose a four step approach. The first step detects candidate voxels using a robust hypothesis test based on a model that assumes Poisson noise and locally planar geometry. The second step performs an adaptive region growth to extract weakly labeled and fine vessels while rejecting spectral artifacts. To enable interactive visualization and estimation of features such as statistical confidence, local curvature, local thickness, and local normal, we perform the third step. In the third step, we construct an accurate mesh representation using marching tetrahedra, volume-preserving smoothing, and adaptive decimation algorithms. To enable topological analysis and efficient validation, we describe a method to estimate vessel centerlines using a ray casting and vote accumulation algorithm which forms the final step of our algorithm. Our algorithm lends itself to parallel processing, and yielded an 8× speedup on a graphics processor (GPU). On synthetic data, our meshes had average error per face (EPF) values of (0.1-1.6) voxels per mesh face for peak signal-to-noise ratios from (110-28 dB). Separately, the error from decimating the mesh to less than 1% of its original size, the EPF was less than 1 voxel/face. When validated on real datasets, the average recall and precision values were found to be 94.66% and 94.84%, respectively.

Visualization and correction of automated segmentation, tracking and lineaging from 5-D stem cell image sequences

BackgroundNeural stem cells are motile and proliferative cells that undergo mitosis, dividing to produce daughter cells and ultimately generating differentiated neurons and glia. Understanding the mechanisms controlling neural stem cell proliferation and differentiation will play a key role in the emerging fields of regenerative medicine and cancer therapeutics. Stem cell studies in vitro from 2-D image data are well established. Visualizing and analyzing large three dimensional images of intact tissue is a challenging task. It becomes more difficult as the dimensionality of the image data increases to include time and additional fluorescence channels. There is a pressing need for 5-D image analysis and visualization tools to study cellular dynamics in the intact niche and to quantify the role that environmental factors play in determining cell fate.ResultsWe present an application that integrates visualization and quantitative analysis of 5-D (x,y,z,t,channel) and large montage confocal fluorescence microscopy images. The image sequences show stem cells together with blood vessels, enabling quantification of the dynamic behaviors of stem cells in relation to their vascular niche, with applications in developmental and cancer biology. Our application automatically segments, tracks, and lineages the image sequence data and then allows the user to view and edit the results of automated algorithms in a stereoscopic 3-D window while simultaneously viewing the stem cell lineage tree in a 2-D window. Using the GPU to store and render the image sequence data enables a hybrid computational approach. An inference-based approach utilizing user-provided edits to automatically correct related mistakes executes interactively on the system CPU while the GPU handles 3-D visualization tasks.ConclusionsBy exploiting commodity computer gaming hardware, we have developed an application that can be run in the laboratory to facilitate rapid iteration through biological experiments. We combine unsupervised image analysis algorithms with an interactive visualization of the results. Our validation interface allows for each data set to be corrected to 100% accuracy, ensuring that downstream data analysis is accurate and verifiable. Our tool is the first to combine all of these aspects, leveraging the synergies obtained by utilizing validation information from stereo visualization to improve the low level image processing tasks.

NetMets: software for quantifying and visualizing errors in biological network segmentation

One of the major goals in biomedical image processing is accurate segmentation of networks embedded in volumetric data sets. Biological networks are composed of a meshwork of thin filaments that span large volumes of tissue. Examples of these structures include neurons and microvasculature, which can take the form of both hierarchical trees and fully connected networks, depending on the imaging modality and resolution. Network function depends on both the geometric structure and connectivity. Therefore, there is considerable demand for algorithms that segment biological networks embedded in three-dimensional data. While a large number of tracking and segmentation algorithms have been published, most of these do not generalize well across data sets. One of the major reasons for the lack of general-purpose algorithms is the limited availability of metrics that can be used to quantitatively compare their effectiveness against a pre-constructed ground-truth. In this paper, we propose a robust metric for measuring and visualizing the differences between network models. Our algorithm takes into account both geometry and connectivity to measure network similarity. These metrics are then mapped back onto an explicit model for visualization.

Constant pressure fluid infusion into rat neocortex from implantable microfluidic devices.

Implantable electrode arrays capable of recording and stimulating neural activity with high spatial and temporal resolution will provide a foundation for future brain computer interface technology. Currently, their clinical impact has been curtailed by a general lack of functional stability, which can be attributed to the acute and chronic reactive tissue responses to devices implanted in the brain. Control of the tissue environment surrounding implanted devices through local drug delivery could significantly alter both the acute and chronic reactive responses, and thus enhance device stability. Here, we characterize pressure-mediated release of test compounds into rat cortex using an implantable microfluidic platform. A fixed volume of fluorescent cell marker cocktail was delivered using constant pressure infusion at reservoir backpressures of 0, 5 and 10 psi. Affected tissue volumes were imaged and analyzed using epifluorescence and confocal microscropies and quantitative image analysis techniques. The addressable tissue volume for the 5 and 10 psi infusions, defined by fluorescent staining with Hoescht 33342 dye, was significantly larger than the tissue volume addressed by simple diffusion (0 psi) and the tissue volume exhibiting insertion-related cell damage (stained by propidium iodide). The results demonstrate the potential for using constant pressure infusion to address relevant tissue volumes with appropriate pharmacologies to alleviate reactive biological responses around inserted neuroprosthetic devices.

Metrics for comparing explicit representations of interconnected biological networks

One of the major goals in biomedical image processing is accurate segmentation of networks embedded in volumetric data sets. Biological networks are composed of a meshwork of thin filaments that span large volumes of tissue. Examples of these structures include neurons and microvasculature, which can take the form of both hierarchical trees and fully connected networks, depending on the imaging modality and resolution. Network function depends on both the geometric structure and connectivity. Therefore, there is considerable demand for algorithms that segment biological networks embedded in three-dimensional data. While a large number of tracking and segmentation algorithms have been published, most of these do not generalize well across data sets. One of the major reasons for the lack of general-purpose algorithms is the limited availability of metrics that can be used to quantitiatively compare their effectiveness against a pre-constructed ground-truth. In this paper, we propose a robust metric for measuring and visualizing the differences between network models. Our algorithm takes into account both geometry and connectivity to measure network similarity. These metrics are then mapped back onto an explicit model for visualization.