Robust and automated three-dimensional segmentation of densely packed cell nuclei in different biological specimens with Lines-of-Sight decomposition

• DOI: 10.1186/s12859-015-0617-x
• Publication Date: 2015-06-06T22:03:48Z
• Authors: B. Mathew; A. Schmitz; S. Muñoz-Descalzo; N. Ansari; F. Pampaloni; E.H.K. Stelzer; S.C. Fischer
• Keywords: Ground truth; Robustness; Biological specimen

Due to the large amount of data produced by advanced microscopy, automated image analysis is crucial in modern biology. Most applications require reliable cell nuclei segmentation. However, many biological specimens are densely packed and appear touch one another images. Therefore, a major difficulty three-dimensional segmentation decomposition that apparently each other. Current methods highly adapted certain specimen or specific microscope. They do not ensure similarly accurate performance, i.e. their robustness for different datasets guaranteed. Hence, these elaborate adjustments dataset. We present an algorithm robust. Our approach combines local adaptive pre-processing with based on Lines-of-Sight (LoS) separate touching into approximately convex parts. demonstrate superior performance our using from recorded microscopes. The images were confocal light sheet-based fluorescence early mouse embryo two cellular spheroids. compared accuracy ground truth test results state-of-the-art methods. shows method throughout all (mean F-measure: 91 %) whereas other failed at least dataset (F-measure ≤ 69 %). Furthermore, volume measurements improved LoS decomposition. required laborious parameter values achieve results. did value adjustments. was achieved fixed set values. developed novel fully incorporating easily accessible features correct splitting independent shape, size intensity. showed methods, performing accurately variety can be readily applied quantitative evaluation drug testing, developmental