• Corpus ID: 214795163

Cell Segmentation by Combining Marker-Controlled Watershed and Deep Learning

  title={Cell Segmentation by Combining Marker-Controlled Watershed and Deep Learning},
  author={Filip Lux and Petr Matula},
We propose a cell segmentation method for analyzing images of densely clustered cells. The method combines the strengths of marker-controlled watershed transformation and a convolutional neural network (CNN). We demonstrate the method universality and high performance on three Cell Tracking Challenge (CTC) datasets of clustered cells captured by different acquisition techniques. For all tested datasets, our method reached the top performance in both cell detection and segmentation. Based on a… 

Figures and Tables from this paper

Marker-controlled watershed with deep edge emphasis and optimized H-minima transform for automatic segmentation of densely cultivated 3D cell nuclei
The use of edge emphasizing U-Nets and optimized H-minima transform can improve the marker-controlled watershed transform for segmentation of densely cultivated 3D cell nuclei.
Improvement of cell image analysis system based on CNN
A CNN-based segmentation technique for cell images to improve the segmentation accuracy in the analysis system and demonstrates that the proposed technique gives a satisfactory accuracy in terms both of the geometrical shape and division timing in the cell segmentation task.
Assessment of deep learning algorithms for 3D instance segmentation of confocal image datasets
This paper implemented and quantitatively compared a number of representative DL pipelines for 3D segmentation, alongside a highly efficient non-DL method named MARS, and shows that the DL pipelines have very different levels of accuracy.
Self-supervised pretraining for transferable quantitative phase image cell segmentation
A novel U-Net-based method for robust adherent cell segmentation for quantitative phase microscopy image is designed and optimised and a self-supervised pretraining technique using nonlabelled data is proposed.
Self-Supervised Pretraining for Transferable Quantitative Phase Image Cell Segmentation
A U-Net-based method for robust adherent cell segmentation for quantitative phase microscopy image is designed and optimised and a self-supervised pretraining technique using nonlabelled data is proposed which improved both segmentation performance and transferability to different cell types.
UnMICST: Deep learning with real augmentation for robust segmentation of highly multiplexed images of human tissues
Two generally applicable approaches to improving segmentation accuracy are demonstrated as scored using new human-labelled segmentation masks spanning multiple human tissues, yielding a set of high accuracy segmentation models.
Enhanced segmentation of label‐free cells for automated migration and interaction tracking
The substantially improved quantitative analysis of cell migration by the revised version of AMIT is more than two orders of magnitude faster than the previous implementation, which makes it feasible to process video data at higher spatial and temporal resolutions.
DeepImageJ: Bridging Deep Learning to ImageJ
DeepImageJ bridges the gap between Deep Learning (DL) model developers and bioimage analysts with a user-friendly plugin of ImageJ/Fiji to run trained DL models in a one-click, and contributes to the democratization of DL in the bioimage analysis.
An adaptively optimized algorithm for counting nuclei in X-ray micro-CT scans of whole organisms
A method to count cells using nuclei in an X-ray microtomographic scan of a zebrafish is proposed and adaptively calibrated using parameters approximated from the manifold of manually selected and optimized special cases.


DIC Image Segmentation of Dense Cell Populations by Combining Deep Learning and Watershed
  • F. Lux, P. Matula
  • Computer Science
    2019 IEEE 16th International Symposium on Biomedical Imaging (ISBI 2019)
  • 2019
The main idea of the approach is to train a convolutional neural network to detect both cellular markers and cellular areas and, based on these predictions, to split the individual cells using the watershed transform.
A deep learning-based algorithm for 2-D cell segmentation in microscopy images
This work presents a single-channel whole cell segmentation algorithm that uses a deep learning approach to learn and predict locations of the cells and their nuclei, and combines that with thresholding and watershed-based segmentation.
Attentive neural cell instance segmentation
Seeded Watersheds for Combined Segmentation and Tracking of Cells
If cells are imaged over time, segmentation results from a previous time frame can be used as seeds for watershed segmentation of the current time frame, and over-segmentation is reduced by rule-based merging, propagating labels from one time-frame to the next.
DCAN: Deep contour‐aware networks for object instance segmentation from histology images
Segmenting and tracking cell instances with cosine embeddings and recurrent hourglass networks
Nuclei Segmentation Using Marker-Controlled Watershed, Tracking Using Mean-Shift, and Kalman Filter in Time-Lapse Microscopy
A novel marker-controlled watershed based on mathematical morphology is proposed, which can effectively segment clustered cells with less oversegmentation and design a tracking method based on modified mean shift algorithm, in which several kernels with adaptive scale, shape, and direction are designed.
U-Net: Convolutional Networks for Biomedical Image Segmentation
It is shown that such a network can be trained end-to-end from very few images and outperforms the prior best method (a sliding-window convolutional network) on the ISBI challenge for segmentation of neuronal structures in electron microscopic stacks.
Segmentation of Nuclei in Histopathology Images by Deep Regression of the Distance Map
A new method to automatically segment nuclei from Haematoxylin and Eosin stained histopathology data with fully convolutional networks is described and superior performance is demonstrated as compared to other approaches using Convolutional Neural Networks.