Contact & Support

Fully Automated Tracking of Chloroplasts in Elodea Leaf Cells from 4D Image Data

Fully automated tracking of moving 3D structures in living cells represents several challenges that are typical to the analysis of biological image data. Taking 4D image data from chloroplasts in living Elodea leaf cells as an example we demonstrate new approaches to meet these challenges (Figure 1). Chloroplasts are surprisingly dynamic organelles: They move extensively throughout cells, they regularly split or fuse and their shape changes constantly. A more detailed understanding of chloroplast dynamic requires reliable quantifications of chloroplast trafficking, division and fusion. We use the SP5 confocal microscope (Leica Microsystems CMS GmbH) with two channels to image living Elodea leaf cells. Auto fluorescence is used to track chloroplasts and the cell wall is detected from reflections.


Topics & Tags

Table of Content

Fig. 1: Representative section of the original 4D data stack (left panel) and 3D reconstruction of a single time point (right panel). Choloroplasts are shown in green, the cell wall in red or dark green.


A first challenge is the reliable separation between signal originating from chloroplasts and background noise. Autoadaptive strategies are used to compute suitable thresholds. A second challenge is the detection of individual chloroplasts and their separation from other chloroplasts in the immediate vicinity. We detect potential chloroplast seeds and employ local thresholds and shape criteria to successively grow these seeds while keeping them separated from adjacent chloroplast seeds (Figure 2).

Fig 2: Intermediate steps of the analyis are represented to illustrate main strategies.

Chloroplasts are detected in 3D to allow volume measurements and the assessment of shape parameters. The third challenge is the tracking of chloroplasts over time. Individual chloroplasts need to be followed throughout the 4D data set in order to from the close spatial proximity of chloroplasts, morphological similarities between neighbor in the next time frame and successively increase the search range to ensure optimal linking (Figure 3).

Fig 3: Illustration of the approach for choroplast tracking. Chloroplasts are linked to their closest neigbour in the next time frame. The search range is successively increased.


We demonstrate that fully automated tracking of chloroplasts in Elodea leaf cells from 4D image data is possible using Definiens Cognition Network Technology. We are confident that the strategies we employed are readily applicable to many other biological image analysis tasks.