Mitosis Detection in Breast Cancer Using Superpixels and Ensemble Classifiers

  1. César A. Ortiz Toro
  2. Consuelo Gonzalo Martín
  3. Angel García Pedrero
  4. Alejandro Rodriguez Gonzalez
  5. Ernestina Menasalvas
Livre:
11th International Conference on Practical Applications of Computational Biology & Bioinformatics
  1. Fernández Riverola, Florentino (ed. lit.)

Éditorial: Springer Suiza

ISBN: 978-3-319-60815-0

Année de publication: 2017

Pages: 137-145

Congreso: Practical Applications of Computational Biology & Bioinformatics (PACBB). International Conference (11. 2017. null)

Type: Communication dans un congrès

Résumé

Determining the severity and potential aggressiveness of breast cancer is an important step in the determination of the treatment options for a patient. Mitosis activity is one of the main components in breast cancer severity grading. Currently, mitosis counting is a laborious, prone to processing errors, done manually by a pathologist. This paper presents a novel approach for automatic mitosis detection, where promising candidates are selected from a superpixel segmentation of the image and classified using an ensemble classifier created from a selection from a pool of different color spaces, different features vector