A multiresolution approach to automated classification of protein subcellular location images
2007

Automated Classification of Protein Locations in Cells

Sample size: 90 publication Evidence: high

Author Information

Author(s): Chebira Amina, Barbotin Yann, Jackson Charles, Merryman Thomas, Srinivasa Gowri, Murphy Robert F, Kovačević Jelena

Primary Institution: Carnegie Mellon University

Hypothesis

Adaptive classification in multiresolution subspaces will improve the classification accuracy.

Conclusion

The study shows that using a multiresolution approach significantly improves the classification accuracy of protein subcellular location images.

Supporting Evidence

  • The best classification accuracy achieved was 95.3% using a neural network classifier.
  • The study demonstrated that multiresolution features significantly enhance classification performance.
  • A reduced set of features was sufficient for high classification accuracy.

Takeaway

This study helps computers better understand where proteins are located in cells by using special techniques that look at images in different ways.

Methodology

The study used a multiresolution decomposition followed by feature computation and classification in each subspace, yielding local decisions combined into a global decision.

Limitations

The system is not shift invariant, which can affect classification accuracy.

Participant Demographics

The dataset consists of approximately 90 single-cell images of HeLa cells across 10 classes.

Statistical Information

P-Value

p<0.05

Statistical Significance

p<0.05

Digital Object Identifier (DOI)

10.1186/1471-2105-8-210

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication