CATFISH FUNCTIONAL BRAIN IMAGING: VALIDATION OF AN AUTOMATED SOFTWARE TOOL FOR MAPPING FUNCTIONAL NEURAL NETWORKS AND CIRCUITS

K. Olson¹; G. Lin²; M.K. Chawla¹; A. Vazdarjanova¹; S.N. Burke¹; B.L. McNaughton¹; P.F. Worley³*; J.F. Guzowski⁴; B. Roysam²; C.A. Barnes1

1. NSMA, Univ Arizona, Tucson, AZ, USA
2. ECSE, Rensselaer Polytechnic Inst, Troy, NY, USA
3. Neurosci, Johns Hopkins, Baltimore, MD, USA
4. Neurosci, Univ New Mexico, Albuquerque, NM, USA

Cellular compartment analysis of temporal activity by fluorescence in situ hybridization (catFISH) provides both temporal and cellular resolution of brain activity that agrees well with electrophysiological recordings. While the goal of the catFISH method to visualize large-scale, behavior-driven cellular activity maps of the brain has been achieved, its application is limited by manual image analysis methods currently available. A comprehensive image analysis system supported by a graphical user interface has been developed that can process multi-spectral confocal image stacks. Segmentation accuracy was greatly improved by a transform combining intensity gradients and geometric distance for the 3-D watershed step, followed by a model-based merging procedure. Accurate algorithms were developed for FISH signal quantification, cell nucleus classification, and automated montaging. Three independent observers manually segmented nuclei, and decided whether the nuclei or cytoplasm of cells exhibited fluorescence or not. Two brain areas were examined: CA1 region of the hippocampus, and the parietal cortex. The software-based segmentation, and intranuclear and cytoplasmic FISH signal classification was compared with a consensus of the segmentations and classifications generated by the three observers. The average mismatches between manual and automated segmentation were, respectively for CA1 and parietal cortex, 5% and 16%; for intranuclear classification 1% and 1%; for cytoplasmic classification 8% and 4%. This type of automated analysis will enable large-scale catFISH studies in behavioral neuroscience.

Support Contributed By: AG18230 & MH01565

technique, in situ hybridization, immediate early gene, hippocampus