Event Abstract

Scale-space based segmentation of cells in functional two-photon in vivo images

  • 1 Imec, Life Science Technologies, Belgium
  • 2 Imec, EHS, Belgium
  • 3 Neuro-Electronics Research Flanders, Belgium

In order to better measure a cell’s response to different chemical or mechanical stimuli robust quantification of the cell’s spatiotemporal organization is required. Accurate identification of individual cellular phenotypes can be obtained by combining automated microscopic acquisition with extensive morphological feature extraction and data mining strategies. Here we evaluate two scale-space approaches for segmentation of microscopic images as opposed to simple intensity thresholding. The first approach is based on the scale-space generated by the Laplacian of Gaussian operator (i.e. Mexican Hat filter, LoG (Prodanov 2012)). While the second approach is based on a deep scale representation generated by the bi-Laplacian of Gaussian (bLoG). Algorithms were implemented as ImageJ plugins.

To test the algorithm, we obtained two-photon images with a chronic preparation, involving implantation of a titanium headpost and a 5 mm glass window that gives optical access to the mouse cortex (Andermann 2010). The two photon setup used consists of a custom designed two-photon microscope (Neurolabware LLC (cheng 2011)) and a Spectra-Physics MaiTai HP DeepSee tunable ultrafast infrared laser (690 nm - 1040 nm tunable range, 70 fs pulses at the sample). The microscope features a fully movable frame, a rotatable objective mount, an 8 kHz resonant scanner and a high-speed frame-grabber based digitization system (30 to 120 Hz frame rates, >1000 pixels per line, 512 to 128 lines per frame), optimized 2 inch collection optics, and 40% quantum efficiency GaAsP photodetectors. The goal of the experiment was to image the glial response and neural activity in vivo, using transgenic mice expressing green fluorescent protein (GFP) astrocytes (Aldh1l1-eGFP+/-) or microglia (CX3CR1-eGFP+/-) or using GFP calcium indicators (GCaMP-family).

We could successfully apply this technique to these functional in vivo two-photon images, whereas thresholding approaches did not perform adequately. In contrast to histogram-based approaches, where a global or a regional threshold level for binarization is computed, the feature-based approaches take into consideration the neighborhood of any given image pixel. The cell segmentation is based on detection of the zero crossings in the filter response, which correspond to edges or ridges in the input images (Lindberg 1994).


Prodanov, D. (2012). Mexican Hat Filter. http://rsbweb.nih.gov/ij/plugins/mexican-hat/index.html

Andermann, M.L., Kerlin, A.M., Reid, R.C. (2010). Chronic cellular imaging of mouse visual cortex during operant behavior and passive viewing. Front Cell Neurosci. 4:3. doi. 10.3389/fncel.2010.00003

Cheng, A., Gonçalves, J., Golshani, P., Arisaka, K., Portera-Cailliau, C. (2011). Simultaneous 2-photon calcium imaging at different cortical depths in vivo with spatiotemporal multiplexing. Nat methods. 8(2):139-142.

Lindeberg, T. (1994). Scale-space theory: A basic tool for analysing structures at different scales. J App Stat. 21 (2): 224–270.

Keywords: Cell segmentation, two-photon imaging, scale-space, ImageJ, image processing, Mexican hat filter

Conference: Imaging the brain at different scales: How to integrate multi-scale structural information?, Antwerp, Belgium, 2 Sep - 6 Sep, 2013.

Presentation Type: Poster presentation

Topic: Poster session

Citation: Mols K, Prodanov D and Bonin V (2013). Scale-space based segmentation of cells in functional two-photon in vivo images. Front. Neuroinform. Conference Abstract: Imaging the brain at different scales: How to integrate multi-scale structural information?. doi: 10.3389/conf.fninf.2013.10.00033

Received: 31 Jul 2013; Published Online: 31 Aug 2013.

* Correspondence: Miss. Katrien Mols, Imec, Life Science Technologies, Leuven, 3001, Belgium, molsk@imec.be

© 2007 - 2018 Frontiers Media S.A. All Rights Reserved