DeepCAN: A Modular Deep Learning System for Automated Cell Counting and Viability Analysis

Eren F., Aslan M., Kanarya D., Uysalli Y., Aydin M., Kiraz B., ...More

IEEE Journal of Biomedical and Health Informatics, vol.26, pp.5575-5583, 2022 (SCI-Expanded) identifier identifier identifier

  • Publication Type: Article / Article
  • Volume: 26
  • Publication Date: 2022
  • Doi Number: 10.1109/jbhi.2022.3203893
  • Journal Name: IEEE Journal of Biomedical and Health Informatics
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Aerospace Database, Applied Science & Technology Source, BIOSIS, Biotechnology Research Abstracts, Communication Abstracts, Compendex, Computer & Applied Sciences, EMBASE, INSPEC, MEDLINE, Metadex, Civil Engineering Abstracts
  • Page Numbers: pp.5575-5583
  • Keywords: Computer architecture, Microprocessors, Image segmentation, Convolutional neural networks, Deep learning, Biomedical imaging, Visualization, Bioimage segmentation, bright field imaging, cell counting, convolutional neural network, viability analysis, IMAGE, SEGMENTATION, FLOW
  • Kayseri University Affiliated: No


IEEEPrecise and quick monitoring of key cytometric features such as cell count, cell size, cell morphology, and DNA content is crucial for applications in biotechnology, medical sciences, and cell culture research. Traditionally, image cytometry relies on the use of a hemocytometer accompanied with visual inspection of an operator under a microscope. This approach is prone to error due to subjective decisions of the operator. Recently, deep learning approaches have emerged as powerful tools enabling quick and highly accurate image cytometric analysis that are easily generalizable to different cell types. Leading to simpler, more compact, and less expensive solutions, these approaches revealed image cytometry as a viable alternative to flow cytometry or Coulter counting. In this study, we demonstrate a modular deep learning system, DeepCAN, that provides a complete solution for automated cell counting and viability analysis. DeepCAN employs three different neural network blocks called Parallel Segmenter, Cluster CNN, and Viability CNN that are trained for initial segmentation, cluster separation, and cell viability analysis, respectively. Parallel Segmenter and Cluster CNN blocks achieve highly accurate segmentation of individual cells while Viability CNN block performs viability classification. A modified U-Net network, a well-known deep neural network model for bioimage analysis, is used in Parallel Segmenter while LeNet-5 architecture and its modified version called Opto-Net are used for Cluster CNN and Viability CNN, respectively. We train the Parallel Segmenter using 15 images of A2780 cells and 5 images of yeasts cells, containing, in total, 14742 individual cell images. Similarly, 6101 and 5900 A2780 cell images are employed for training Cluster CNN and Viability CNN models, respectively. 2514 individual A2780 cell images are used to test the overall segmentation performance of Parallel Segmenter combined with Cluster CNN, revealing high Precision/Recall/F1-Score values of 96.52%/96.45%/98.06%, respectively. Overall cell counting/viability analysis performance of DeepCAN is tested with A2780 (2514 cells), A549 (601 cells), Colo (356 cells), and MDA-MB-231 (887 cells) cell images revealing high counting/viability analysis accuracies of 96.76%/99.02%, 93.82%/95.93%, and 92.18%/97.90%, 85.32%/97.40%, respectively.