Diagnose the premalignant pancreatic cancer using high dimensional linear machine

Abstract

High throughput mass spectrometry technique has been extensively studied for the diagnosis of cancers. The detection of the pancreatic cancer at a very early stage is important to heal patients, but is very difficult due to biological and computational challenges. This paper proposes a simple classification approach which can be applied to the premalignant pancreatic cancer detection using mass spectrometry technique. Computational experiments show that our method outperforms the benchmark methods in accuracy and sensitivity without resorting to any biomarker selection, and the comparison with previous works shows that our method can obtain competitive performance. © 2012 Springer-Verlag.