Robust Automated Tumour Segmentation on Histological and Immunohistochemical Tissue Images
2011

Automated Tumor Segmentation in Tissue Images

Sample size: 18 publication Evidence: moderate

Author Information

Author(s): Ching-Wei Wang

Primary Institution: National Taiwan University of Science and Technology

Hypothesis

Can an automated method improve tumor cell segmentation in histological and immunohistochemical images?

Conclusion

The automated tumor segmentation method achieved 80% accuracy for H&E images and 78% accuracy for IHC images.

Supporting Evidence

  • The automated method achieved 80% accuracy in H&E images.
  • The method also achieved 78% accuracy in IHC images.
  • The technique significantly reduces the workload for pathologists.

Takeaway

This study created a computer program that helps doctors find cancer cells in tissue samples faster and more accurately.

Methodology

The study developed a tumor segmentation system using a tissue architecture extraction model and a Markov Random Field image segmentation method.

Limitations

The method may struggle with poorly differentiated cases where even experienced pathologists find it difficult to define cancerous cells.

Digital Object Identifier (DOI)

10.1371/journal.pone.0015818

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication