Cheng, J., Xiong, W., Chia, S. C., Lim, J. H., Sankaran, S., & Ahmed, S. (2014). Neurosphere segmentation in brightfield images. In S. Ourselin & M. A. Styner (Eds.), Medical Imaging 2014: Image Processing (Vol. 9034, p. 90344D). SPIE. https://doi.org/10.1117/12.2043365
Abstract:
The challenge of segmenting neurospheres (NSPs) from bright eld images includes uneven background illumination (vignetting), low contrast and shadow-casting appearance near the well wall. We propose a pipeline for neurosphere segmentation in bright eld images, focusing on shadow-casting removal. Firstly, we remove vignetting by creating a synthetic blank eld image from a set of bright eld images of the whole well. Then, radial line integration is proposed to remove the shadow-casting and therefore facilitate automatic segmentation. Furthermore, a weighted bi-directional decay function is introduced to prevent undesired gradient effect of line integration on NSPs without shadow-casting. Afterward, multiscale Laplacian of Gaussian (LoG) and localized
region-based level set are used to detect the NSP boundaries. Experimental results show that our proposed radial line integration method (RLI) achieves higher detection accuracy over existing methods in terms of precision, recall and F-score with less computational time.
License type:
Publisher Copyright
Funding Info:
This research / project is supported by the Agency for Science, Technology and Research - Joint Council Office
Grant Reference no. : JCOAG03_FG01_2009
Description:
Copyright 2014 Society of Photo‑Optical Instrumentation Engineers (SPIE). One print or electronic copy may be made for personal use only. Systematic reproduction and distribution, duplication of any material in this publication for a fee or for commercial purposes, and modification of the contents of the publication are prohibited.