Dual Dictionary Learning for Cell Segmentation in Bright-field Microscopy Images

Introduction

Cell segmentation is an important but time-consuming and laborious task in biological image analysis, therefore, an automated, robust, and fast method is required to overcome burdensome processes.
The majority of existing work are based on input images and predefined feature models only – for example, using a deformable model to extract edge boundaries in the image.

이미지 8

In our method, we propose a novel data-driven cell segmentation algorithm for bright-field microscopy images.
이미지 10

Dual Dictionary Learning based on Convolutional Sparse Coding

이미지 12
The proposed method minimized an energy formula defined by two dictionaries and a common sparse code which aims to find the pixel-level classification by deploying the learned dictionaries on new images.
Our method does not need to know prior knowledge of objects like edge information.
이미지 15

Once we obtain well-trained filter banks D and B, they can be deployed to the segmentation process by finding the sparse codes from the first square norm and approximate the results as L above Eq. 2.
The training process derived in Alternating Direction Method of Multipliers (ADMM) is running on the GPU device by using a gpuarray of MATLAB.

 

Results

이미지 17

  • [PDF] G. Lee, T. M. Quan, and W. Jeong, “명시야 현미경 영상에서의 세포 분할을위한 이중 사전 학습 기법,” Journal of the Korea Computer Graphics Society, vol. 22, pp. 21-29, 2016.
    [Bibtex]
    @article{ghlee_2016_dualdictionary,
    author={Gyuhyun Lee and Tran Minh Quan and Won-Ki Jeong},
    title={{명시야 현미경 영상에서의 세포 분할을위한 이중 사전 학습 기법}},
    booltitle={{Vol.22 No.3}},
    journal={{Journal of the Korea Computer Graphics Society}},
    volume={22},
    issur={3},
    publisher={Korea Computer Graphics Society},
    year={2016},
    pages={21-29},
    url={http://www.dbpia.co.kr/Article/NODE06716016
    }
    }