The Image Analysis Group is concerned with advanced analysis of biological imagery. The main thrust of research is to develop novel techniques for automatically extracting high-level information from the full range of images generated by the biologists. Current research is at the Steven Fisher's Lab. focuses on the effects of retinal detachment. The retina image analysis group is working on algorithms for automated cell counting, retinal layer segmentation, texture analysis, and a host of other quantifiers. The microtubule group is working on techniques for tracking microtubules as well as analyzing microtubule dynamics.
Cell addition and loss are important biological events in development and pathology. As a result, counts of cells and nuclei from histological sections provide quantitative information central to studying changes in cells, tissues, and organs. In retinal images, the number of photoreceptors, layer thickness and other structural changes are common measurement of visual function of retina (for details see "Cellular Remodeling in Mammalian Retina Induced by Retinal Detachment" in Webvision). While progress in understanding changes in such parameters as cell structure or protein expression has been rapid during the past few decades, methods for automatically determining cell number, layer thickness and structural changes have remained limited. In this project, image analysis methods have been developed for microtubule tracing, retina nucleus detection and retinal layer segmentation.
Evaluating the Performance of Microtubule Tracing in Live Cell Images: Methods and Ground Truth
We propose an evaluation method to compare the tracing results to ground truth data. The ground truth manually obtained from different experts is available for downloading. Read more ...
Automated detection of cell nuclei in retinal images
We develop a new method for automatically detecting nuclei, and therefore cell bodies, from a 2-D digital micrograph. The proposed approach provides an accurate, simple, and reliable method to count cells, nuclei, or other objects in sectioned materials. Read more...
Multispectral Image Segmentation
We quantitatively analyze the utility of multispectral imagery for classification and segmentation tasks in histopathology imagery and breast cancer cell detection. Read more...
Probabilistic Analysis of Retinal Neuron Morphology
Due to noise from the imaging process and imperfect cell labeling techniques, bioimages provide inherent uncertainty in their data, making it difficult to give precise answers to many quantifications. Due to these constraints, we introduce probabilistic methods to analyze the morphology of ganglion cells with minimal human interaction. Read more..
Automated Segmentation of the ONL in retinal images
We developed a new variational technique for automatic segmentation of layers in retinal images. Prior information in the form of a reference image is exploited to obtain a reliable segmentation of all the images in a dataset composed of 50 images for four different experimental conditions. Qualitative and quantitative results are presented. Read more...
Automated Layer Boundary Segmentation in retinal images
The algorithm uses parametric active contours or snakes to locate the boundaries between layers in an image, e.g. retinal layer boundaries. For locating a particular boundary in an image, a snake is initialized near that boundary and is iteratively deformed to find the boundary. Read more...