SfN Abstracts

MBF Bioscience will be presenting 3 posters at the Society for Neuroscience Annual Meeting, November 15-19, in Washington, D.C.

For information on the presentation times and locations, please click any of the following links:

“INTERACTIVE AUTOMATIC RECONSTRUCTION OF MULTIPLE NEURONS STAINED WITH THE GOLGI METHOD”
“AUTOMATED DESIGN-BASED STEREOLOGY USING THE OPTICAL FRACTIONATOR METHOD”
“A METHODOLOGY FOR EMPIRICAL DETERMINATION OF GUARD ZONE HEIGHT FOR THE OPTICAL FRACTIONATOR”

INTERACTIVE AUTOMATIC RECONSTRUCTION OF MULTIPLE NEURONS STAINED WITH THE GOLGI METHOD (398.16)

Location

V V 18

Presentation Date/Time

Monday, Nov 17, 2008, 11:00 AM -12:00 PM

Authors

Julie Simpson, Ph.D., Muhammad Abdul Karim, Ph.D., Susan Hendricks, Ph.D., Paul Angstman, Jack Glaser

Abstract

Unlike other staining methods, Golgi impregnation is widely used in many studies involving neuron morphology because it completely stains individual cells and is relatively straightforward to perform. However, the complexity of the forest of neuronal branches makes automatic neuron tracing extremely challenging, and currently manual computer-assisted neuron tracing is the preferred method for neuron reconstruction of Golgi stained tissue.

While automated neuron tracing has been successfully applied to other single and multi-cell visualization modalities, (e.g., cell fills, GFP transgenes, DAB) in multiple imaging techniques (brightfield, widefield, confocal, two photon confocal fluorescence), automated reconstruction of neurons within Golgi stained tissue has proven thus far to be an unsuitable candidate for this technology. Current algorithms are sensitive to a number of factors common to Golgi impregnation, including poor signal-to-noise ratio, discontinuous staining, presences of spines, dense overlapping branches and the sheer number of neurons labeled within a section. The results of automated tracing algorithms, even in 3D stacks where branches can be visually resolved, typically require significant effort to correct branching errors and produce correct tree structures.

Our new algorithm uses an interactive approach in which approximate traces are displayed and updated in real-time as the user selects points on the 3D image stack across multiple fields-of-view. The final reconstruction, including thickness and XYZ mapping, is performed after the user approves an approximate trace. Several points can be specified between the tree root and endings to add more constraint to the trace to prevent the automatic trace from following neighboring processes. This way, the user enforces the correct anatomical tree structure while tracing in 3D is performed by the computer. This user input is especially important for Golgi-impregnated cells since accurate centerlines of traced branches are needed for spine location analysis. Accurate detection in crowded locations require that image stacks need to be acquired at high magnification, resulting in multiple image stacks that extend across multiple fields-of-view.

This new approach is aimed at improving the efficiency of complex reconstruction tasks in Golgi-stained tissue, balancing the trade-off between efforts required to edit fully-automated traces and manual tracing, ultimately reducing the reconstruction time between image acquisition and spine analyses.

Disclosures

J. Simpson , MBF Bioscience, A. Employment (full or part-time); M. Abdul Karim, MBF Bioscience, A. Employment (full or part-time); S.J. Hendricks, MBF Bioscience, A. Employment (full or part-time); P. Angstman, MBF Bioscience, A. Employment (full or part-time); J. Glaser, MBF Bioscience, A. Employment (full or part-time).

AUTOMATED DESIGN-BASED STEREOLOGY USING THE OPTICAL FRACTIONATOR METHOD (398.17)

Location

V V 19

Presentation Date/Time

Monday, Nov 17, 2008, 8:00 AM – 9:00 AM

Authors

Dan Peruzzi, Ph.D., Susan Hendricks, Ph.D., Nicolas Roussel, Ph.D., Paul Angstman, Jack Glaser

Abstract

Unbiased stereological methods can be labor-intensive even using computer-controlled microscope equipment. In an ongoing effort to increase the efficiency of data analysis, we are investigating a new methodology for automated cell recognition. Previously we demonstrated an improvement in data collection rate by integrating automated cell detection with the unbiased 3D counting frame of the Optical Fractionator in the stereology program, Stereo Investigator. Adaptive local thresholding was used to detect cells based on minimum diameter. However, tightly packed cells were inaccurately detected as large single cells.

To improve cell detection accuracy, we present a model-based approach with non-overlap constraints for handling aggregated cells, where cells are modeled as deformable 3D ellipsoids. In addition to improving cell counts, the current approach more accurately models the shapes of detected cells to provide accurate volume rendering.

The improved algorithm is more robust and can split falsely merged cells based on a number of criteria (size, eccentricity, etc). The software features interactive user inputs for adjusting incorrect computer detection or improving model fitting of identified cells. Adjustments to the models are accumulated from counting site to counting site, decreasing the rate of user intervention with successive counts. Automated detection results are inspected and edited by the investigator to ensure accuracy.

In addition, multiple cell types can be quantified with one pass using multichannel analysis. Subsequent multichannel analysis will be able to detect cohorts of cells with colocalization filters and nearest-neighbor analysis. The current automated cell recognition software improves the efficiency for accurate analyses with the Optical Fractionator while providing more detailed morphological analyses.

Disclosures

D. Peruzzi , MBF Bioscience, A. Employment (full or part-time); S.J. Hendricks, MBF Bioscience, A. Employment (full or part-time); N. Roussel, MBF Bioscience, A. Employment (full or part-time); P. Angstman, MBF Bioscience, A. Employment (full or part-time); J. Glaser, MBF Bioscience, A. Employment (full or part-time).

Support

NIH Grant MH076188

A METHODOLOGY FOR EMPIRICAL DETERMINATION OF GUARD ZONE HEIGHT FOR THE OPTICAL FRACTIONATOR (398.18)

Location

V V 20

Presentation Date/Time

Monday, Nov 17, 2008, 9:00 AM -10:00 AM

Authors

Susan Hendricks, Ph.D., Alissa Wilson, Jack Glaser

Abstract

Proper implementation of the unbiased Optical Fractionator stereology method requires that each object has the opportunity to be counted once and only once. One aspect of this method is the use of guard zones, which are placed at the top and bottom of the histological section to be sampled. These help ensure that the tissue included within the disector does not contain any artifacts due to sectioning or processing.

During the pilot study, configuration of the sampling parameters can include empirical determination of the guard zones; this enables unbiased and efficient data collection. We demonstrate here a straightforward method for calculating guard zone heights by examining depth-of-field, focus sensitivity, post-processing thickness, and the size of the objects to be counted. Our preliminary data suggest that by systematically counting sample tissue without guard zones and creating a histogram that plots the cell depth, the investigator can empirically determine where to position the disector. Further, we found that the size of the histogram bins in relation to the average post-processing thickness of the tissue was integral to interpreting the data in a meaningful way.

Optimizing the height sampling fraction is the most efficient means to generate an adequate volume fraction for quantification using the Optical Fractionator. Therefore, understanding the factors that influence the guard zone height is an important consideration when maximizing the disector height without the introduction of sampling bias. By comparing estimates obtained with the empirically determined guard zone height versus estimates obtained with smaller and larger guard zones, it is possible to measure the effect of guard zones on the precision of the estimate obtained with the Optical Fractionator.

Disclosures

S. Hendricks , MBF Bioscience, A. Employment (full or part-time); A. Wilson, MBF Bioscience, A. Employment (full or part-time); J. Glaser, MBF Bioscience, A. Employment (full or part-time).