A generic classification-based method for segmentation of nuclei in 3D images of early embryos

Abstract

Background: Studying how individual cells spatially and temporally organize within the embryo is a fundamental issue in modern developmental biology to better understand the first stages of embryogenesis. In order to perform high-throughput analyses in three-dimensional microscopic images, it is essential to be able to automatically segment, classify and track cell nuclei. Many 3D/4D segmentation and tracking algorithms have been reported in the literature. Most of them are specific to particular models or acquisition systems and often require the fine tuning of parameters.

Results: We present a new automatic algorithm to segment and simultaneously classify cell nuclei in 3D/4D images. Segmentation relies on training samples that are interactively provided by the user and on an iterative thresholding process. This algorithm can correctly segment nuclei even when they are touching, and remains effective under temporal and spatial intensity variations. The segmentation is coupled to a classification of nuclei according to cell cycle phases, allowing biologists to quantify the effect of genetic perturbations and drug treatments. Robust 3D geometrical shape descriptors are used as training features for classification. Segmentation and classification results of three complete datasets are presented. In our working dataset of the Caenorhabditis elegans embryo, only 21 nuclei out of 3,585 were not detected, the overall F-score for segmentation reached 0.99, and more than 95% of the nuclei were classified in the correct cell cycle phase. No merging of nuclei was found.

Conclusion: We developed a novel generic algorithm for segmentation and classification in 3D images. The method, referred to as Adaptive Generic Iterative Thresholding Algorithm (AGITA), is freely available as an ImageJ plug-in.

Figure 1

Flowchart of the overall procedure. The first step consists in training the program by learning sample nuclei. The user clicks on the approximate position of the nuclei in the image. Using an iterative thresholding algorithm, nuclei are automatically extracted around each clicked position. The user has to validate the segmented nuclei that will serve as training samples for the classifier. The second step is completely automatic and will segment, as well as classify, objects in the entire dataset.

Figure 2

Flowchart for building the hierarchy of objects used by the segmentation and classification procedure. When an object is well classified its information is saved and the object is inserted into the hierarchy structure.

Figure 3

Hierarchy structure of classified objects. Starting with a low threshold value, the whole embryo is segmented but cannot be classified as a valid object. When the threshold value increases, objects can be classified as valid and put in the hierarchy with the saved information. In the case of object separation, the associated object at the previous threshold, if any, is computed in order to put the objects in their right place in the hierarchy. The best object inside each branch is computed by maximizing the classification, and the segmented object is reconstructed.

Figure 4

Results of segmentation and classification on the C. elegans dataset. Top row: 3D view of segmented data for two different time points (T = 86,122). Bottom row: 3D view of classified data for the same time points. Note that all objects are well separated.

Figure 5

Segmentation contours overlaid on different slices (Z = 8,9,21) of raw data for the C. elegans dataset. Left and right columns correspond to left and right columns of Figure 4. Note that even hard to distinguish nuclei could be detected.

Figure 6

Results of segmentation and classification on the Drosophila dataset for different time points (T = 5,9,25). Top row : 3D view of original raw data. Middle row : 3D view of segmented data. Bottom row : classification results (Red = interphase, Green = prophase, Magenta = metaphase, Cyan = anaphase, Yellow = telophase). Note that in bottom middle, almost all of the nuclei have just divided; they are in anaphase or telophase stages and are correctly segmented and classified.

Figure 7

Results of segmentation on the C. elegans dataset with the HK-Means method for the same time points and the same z slices as Figures 5and 6. Top row: 3D surface view of segmented objects. Next rows: some slices with contours overlaid. (*) indicates merged nuclei; (N) indicates noisy detected object, (O) indicates missing object; and (S) indicates inaccurate segmentation, especially for metaphase nuclei.