Feature selection for outcome prediction in oesophageal cancer using genetic algorithm and random forest classifier

Highlights • Proposition of a new feature selection strategy in two steps called GARF. • Selection of relevant subset of features extracted from PET images and clinical data. • Proposition of a multiparametric fitness function depending on RF and AUC. • Excellent performances obtained in comparison...

Ausführliche Beschreibung

Gespeichert in:

Bibliographische Detailangaben
Veröffentlicht in:	Computerized medical imaging and graphics 2017-09, Vol.60, p.42-49
Hauptverfasser:	Paul, Desbordes, Su, Ruan, Romain, Modzelewski, Sébastien, Vauclin, Pierre, Vera, Isabelle, Gardin
Format:	Artikel
Sprache:	eng
Schlagworte:	Algorithms Area Under Curve Cancer Esophageal cancer Esophageal Neoplasms - diagnosis Esophageal Neoplasms - diagnostic imaging Esophageal Neoplasms - genetics Esophagus Feature extraction Feature selection Genetic algorithm Genetic algorithms Humans Internal Medicine NMR Nuclear magnetic resonance Oesophageal cancer Other Patients Positron emission Positron emission tomography Predictions Prognosis Radiation Radiation therapy Radiomics Random forest ROC Curve Tomography
Online-Zugang:	Volltext
Tags:	Tag hinzufügen Keine Tags, Fügen Sie den ersten Tag hinzu!

Beschreibung
Zusammenfassung:	Highlights • Proposition of a new feature selection strategy in two steps called GARF. • Selection of relevant subset of features extracted from PET images and clinical data. • Proposition of a multiparametric fitness function depending on RF and AUC. • Excellent performances obtained in comparison to 3 other methods.
ISSN:	0895-6111 1879-0771
DOI:	10.1016/j.compmedimag.2016.12.002