Enhanced massive training artificial immune recognition system for false positives reduction in lung nodules classification

Abstract

Massive Training Artificial Immune Recognition System (MTAIRS) had been implemented in the computerized system to classify lung nodules on Computed Tomography (CT) scans. In this algorithm, large training sub-regions are trained, and the classification algorithm shows promising results in the lung nodules classification. However, in the output images of non-nodule cases, some false positives are still identified in the MTAIRS. False positives are always considered as a common issue in most of the development of classification algorithms of lung nodules detection. The effort of reducing false positives in the output images from MTAIRS is presented where the enhancement is based on the affinity function in MTAIRS algorithms. The quantitative assessment on the classification results for detection of lung nodules will be presented in this research.