Donald H. Szarowski

Brain responses to micro-machined silicon devices

Micro-machined neural prosthetic devices can be designed and fabricated to permit recording and stimulation of specific sites in the nervous system. Unfortunately, the long-term use of these devices is compromised by cellular encapsulation. The goals of this study were to determine if device size, surface characteristics, or insertion method affected this response. Devices with two general designs were used. One group had chisel-shaped tips, sharp angular corners, and surface irregularities on the micrometer size scale. The second group had rounded corners, and smooth surfaces. Devices of the first group were inserted using a microprocessor-controlled inserter. Devices of the second group were inserted by hand. Comparisons were made of responses to the larger devices in the first group with devices from the second group. Responses were assessed 1 day and 1, 2, 4, 6, and 12 weeks after insertions. Tissues were immunochemically labeled for glial fibrillary acidic protein (GFAP) or vimentin to identify astrocytes, or for ED1 to identify microglia. For the second comparison devices from the first group with different cross-sectional areas were analyzed. Similar reactive responses were observed following insertion of all devices; however, the volume of tissue involved at early times, <1 week, was proportional to the cross-sectional area of the devices. Responses observed after 4 weeks were similar for all devices. Thus, the continued presence of devices promotes formation of a sheath composed partly of reactive astrocytes and microglia. Both GFAP-positive and -negative cells were adherent to all devices. These data indicate that device insertion promotes two responses-an early response that is proportional to device size and a sustained response that is independent of device size, geometry, and surface roughness. The early response may be associated with the amount of damage generated during insertion. The sustained response is more likely due to tissue-device interactions.

Cerebral Astrocyte Response to Micromachined Silicon Implants

The treatment of neurologic disorders and the restoration of lost function due to trauma by neuroprosthetic devices has been pursued for over 20 years. The methodology for fabricating miniature devices with sophisticated electronic functions to interface with nervous system tissue is available, having been well established by the integrated circuit industry. Unfortunately, the effectiveness of these devices is severely limited by the tissue reaction to the insertion and continuous presence of the implant, a foreign object. This study was designed to document the response of reactive astrocytes in the hope that this information will be useful in specifying new fabrication technologies and devices capable of prolonged functioning in the brain. Model probes fabricated from single crystal silicon wafers were implanted into the cerebral cortices of rats. The probes had a 1 x 1-mm tab, for handling, and a 2-mm-long shaft with a trapezoidal cross-section (200-microm base, 60microm width at the top, and 130 microm height). The tissue response was studied by light and scanning electron microscopy at postinsertion times ranging from 2 to 12 weeks. A continuous sheath of cells was found to surround the insertion site in all tissue studied and was well developed but loosely organized at 2 weeks. By 6 and 12 weeks, the sheath was highly compacted and continuous, isolating the probe from the brain. At 2 and 4 weeks, the sheath was disrupted when the probe was removed from the fixed tissue, indicating that cells attached more strongly to the surface of the probe than to the nearby tissue. The later times showed much less disruption. Scanning electron microscopy of the probes showed adherent cells or cell fragments at all time points. Thus, as the sheath became compact, the cells on the probe and the cells in the sheath had decreased adhesion to each other. Immunocytochemistry demonstrated that the sheath was labeled with antibodies to glial fibrillary acidic protein (GFAP), an indicator for reactive gliosis. The tissue surrounding the insertion site showed an increased number of GFAP-positive cells which tended to return to control levels as a function of time after probe insertion. It was concluded that reactive gliosis is an important part of the process forming the cellular sheath. Further, the continuous presence of the probe appears to result in a sustained response that produces and maintains a compact sheath, at least partially composed of reactive glia, which isolates the probe from the brain.

Rapid automated three-dimensional tracing of neurons from confocal image stacks

Algorithms are presented for fully automatic three-dimensional (3D) tracing of neurons that are imaged by fluorescence confocal microscopy. Unlike previous voxel-based skeletonization methods, the present approach works by recursively following the neuronal topology, using a set of 4 /spl times/ N/sup 2/ directional kernels (e.g., N = 32), guided by a generalized 3D cylinder model. This method extends our prior work on exploratory tracing of retinal vasculature to 3D space. Since the centerlines are of primary interest, the 3D extension can be accomplished by four rather than six sets of kernels. Additional modifications, such as dynamic adaptation of the correlation kernels, and adaptive step size estimation, were introduced for achieving robustness to photon noise, varying contrast, and apparent discontinuity and/or hollowness of structures. The end product is a labeling of all somas present, graph-theoretic representations of all dendritic/axonal structures, and image statistics such as soma volume and centroid, soma interconnectivity, the longest branch, and lengths of all graph branches originating from a soma. This method is able to work directly with unprocessed confocal images, without expensive deconvolution or other preprocessing. It is much faster that skeletonization, typically consuming less than a minute to trace a 70 MB image on a 500 MHz computer. These properties make it attractive for large-scale automated tissue studies that require rapid on-line image analysis, such as high-throughput neurobiology/angiogenesis assays, and initiatives such as the Human Brain Project.

Development of a rare cell fractionation device: application for cancer detection

Isolating rare cells from biological fluids including whole blood or bone marrow is an interesting biological problem. Characterization of a few metastatic cells from cancer patients for further study is desirable for prognosis/diagnosis. Traditional methods have not proven adequate, due to the compositional complexity of blood, with its large numbers of cell types. To separate individual cells based on their mechanical characteristics, we have developed a series of massively parallel microfabricated sieving device. These devices were constructed with four successively narrower regions of channels numbering /spl sim/1800 per region. As cells traversed the device, they encountered each region and stopped at a gap width that prohibited passage due to their size. Cultured neuroblastoma cells, when mixed with whole blood and applied to the device, were retained in the 10-/spl mu/m-wide by 20-/spl mu/m-deep channels. All other cells migrated to the output. A derivative of the same device was utilized to characterize migration of whole blood. Adult white blood cells were retained at the 2.5-/spl mu/m-wide by 5-/spl mu/m-deep channels, while red blood cells passed through these channels. Devices designed to capture rare cells in peripheral circulation for downstream analysis will provide an important tool for diagnosis and treatment.

Light Microscopic Images Reconstructed by Maximum Likelihood Deconvolution

The main purpose of this chapter is to introduce the reader to the methodology of maximum likelihood (ML)-based deblurring algorithms. It is aimed at the interdisciplinary scientist, who may not be concerned about the underlying mathematical foundations of the methodology but who needs to understand the main principles behind the algorithms used. Some mathematical principles are explained, but the interested reader may find more details in the numerous publications cited in Holmes (1989, 1992), Krishnamurthi et al (1992), and Shaw and Rawlins (1991). A sample image reconstruction is presented from each of three microscope modalities, including the wide-field epifluorescence (WFF) microscope, the confocal pinhole laser-scanned epifluorescence microscope (CLSM), and the transmitted light brightfleld (BF) microscope.

Median-based robust algorithms for tracing neurons from noisy confocal microscope images

This paper presents a method to exploit rank statistics to improve fully automatic tracing of neurons from noisy digital confocal microscope images. Previously proposed exploratory tracing (vectorization) algorithms work by recursively following the neuronal topology, guided by responses of multiple directional correlation kernels. These algorithms were found to fail when the data was of lower quality (noisier, less contrast, weak signal, or more discontinuous structures). This type of data is commonly encountered in the study of neuronal growth on microfabricated surfaces. We show that by partitioning the correlation kernels in the tracing algorithm into multiple subkernels, and using the median of their responses as the guiding criterion improves the tracing precision from 41% to 89% for low-quality data, with a 5% improvement in recall. Improved handling was observed for artifacts such as discontinuities and/or hollowness of structures. The new algorithms require slightly higher amounts of computation, but are still acceptably fast, typically consuming less than 2 seconds on a personal computer (Pentium III, 500 MHz, 128 MB). They produce labeling for all somas present in the field, and a graph-theoretic representation of all dendritic/axonal structures that can be edited. Topological and size measurements such as area, length, and tortuosity are derived readily. The efficiency, accuracy, and fully-automated nature of the proposed method makes it attractive for large-scale applications such as high-throughput assays in the pharmaceutical industry, and study of neuron growth on nano/micro-fabricated structures. A careful quantitative validation of the proposed algorithms is provided against manually derived tracing, using a performance measure that combines the precision and recall metrics.

Automated Three-Dimensional Tracing of Neurons in Confocal and Brightfield Images

Automated three-dimensional (3-D) image analysis methods are presented for tracing of dye-injected neurons imaged by fluorescence confocal microscopy and HRP-stained neurons imaged by transmitted-light brightfield microscopy. An improved algorithm for adaptive 3-D skeletonization of noisy images enables the tracing. This algorithm operates by performing connectivity testing over large N x N x N voxel neighborhoods exploiting the sparseness of the structures of interest, robust surface detection that improves upon classical vacant neighbor schemes, improved handling of process ends or tips based on shape collapse prevention, and thickness-adaptive thinning. The confocal image stacks were skeletonized directly. The brightfield stacks required 3-D deconvolution. The results of skeletonization were analyzed to extract a graph representation. Topological and metric analyses can be carried out using this representation. A semiautomatic method was developed for reconnection of dendritic fragments that are disconnected due to insufficient dye penetration, an imaging deficiency, or skeletonization errors.

Advances in automated 3-D image analysis of cell populations imaged by confocal microscopy

Automated three-dimensional (3-D) image analysis methods are presented for rapid and effective analysis of populations of fluorescently labeled cells or nuclei in thick tissue sections that have been imaged three dimensionally using a confocal microscope. The methods presented here greatly improve upon our earlier work (Roysam et al.:J Microsc 173: 115-126, 1994). The principal advances reported are: algorithms for efficient data pre-processing and adaptive segmentation, effective handling of image anisotrophy, and fast 3-D morphological algorithms for separating overlapping or connected clusters utilizing image gradient information whenever available. A particular feature of this method is its ability to separate densely packed and connected clusters of cell nuclei. Some of the challenges overcome in this work include the efficient and effective handling of imaging noise, anisotrophy, and large variations in image parameters such as intensity, object size, and shape. The method is able to handle significant inter-cell, intra-cell, inter-image, and intra-image variations. Studies indicate that this method is rapid, robust, and adaptable. Examples were presented to illustrate the applicability of this approach to analyzing images of nuclei from densely packed regions in thick sections of rat liver, and brain that were labeled with a fluorescent Schiff reagent.

Attenuation correction in confocal laser microscopes: a novel two‐view approach

Confocal microscopy is a three‐dimensional (3D) imaging modality, but the specimen thickness that can be imaged is limited by depth‐dependent signal attenuation. Both software and hardware methods have been used to correct the attenuation in reconstructed images, but previous methods do not increase the image signal‐to‐noise ratio (SNR) using conventional specimen preparation and imaging. We present a practical two‐view method that increases the overall imaging depth, corrects signal attenuation and improves the SNR. This is achieved by a combination of slightly modified but conventional specimen preparation, image registration, montage synthesis and signal reconstruction methods. The specimen is mounted in a symmetrical manner between a pair of cover slips, rather than between a slide and a cover slip. It is imaged sequentially from both sides to generate two 3D image stacks from perspectives separated by approximately 180° with respect to the optical axis. An automated image registration algorithm performs a precise 3D alignment, and a model‐based minimum mean squared algorithm synthesizes a montage, combining the content of both the 3D views. Experiments with images of individual neurones contrasted with a space‐filling fluorescent dye in thick brain tissue slices produced precise 3D montages that are corrected for depth‐dependent signal attenuation. The SNR of the reconstructed image is maximized by the method, and it is significantly higher than in the single views after applying our attenuation model. We also compare our method with simpler two‐view reconstruction methods and quantify the SNR improvement. The reconstructed images are a more faithful qualitative visualization of the specimens structure and are quantitatively more accurate, providing a more rigorous basis for automated image analysis.

Algorithms for accurate 3D registration of neuronal images acquired by confocal scanning laser microscopy

This paper presents automated and accurate algorithms based on high‐order transformation models for registering three‐dimensional (3D) confocal images of dye‐injected neurons. The algorithms improve upon prior methods in several ways, and meet the more stringent image registration needs of applications such as two‐view attenuation correction recently developed by us. First, they achieve high accuracy (≈ 1.2 voxels, equivalent to 0.4 µm) by using landmarks, rather than intensity correlations, and by using a high‐dimensional affine and quadratic transformation model that accounts for 3D translation, rotation, non‐isotropic scaling, modest curvature of field, distortions and mechanical inconsistencies introduced by the imaging system. Second, they use a hierarchy of models and iterative algorithms to eliminate potential instabilities. Third, they incorporate robust statistical methods to achieve accurate registration in the face of inaccurate and missing landmarks. Fourth, they are fully automated, even estimating the initial registration from the extracted landmarks. Finally, they are computationally efficient, taking less than a minute on a 900‐MHz Pentium III computer for registering two images roughly 70 MB in size. The registration errors represent a combination of modelling, estimation, discretization and neuron tracing errors. Accurate 3D montaging is described; the algorithms have broader applicability to images of vasculature, and other structures with distinctive point, line and surface landmarks.