DEEP LEARNING TECHNIQUES FOR LUNG CANCER SEGMENTATION USING MULTIPLE NEURAL NETWORKS
European Journal of Molecular & Clinical Medicine,
2020, Volume 7, Issue 9, Pages 1507-1514
AbstractDeep learning in pattern recognition and classification is considered a common and potent tool. There are, be that as it may, very few profoundly composed frameworks utilized in the field of clinical imaging conclusion, since there isn't constantly a wide store accessible for clinical pictures. In this investigation we tried the possibility of utilizing the Lung Image Database Consortium (LIDC) database cases to utilize profound learning calculations for lung disease conclusion. The knobs were portioned by the imprints given by the radiologists on each figured tomography cut. In the wake of inspecting and turning down we gained 174412 examples with 52 by 52 pixels each and the relating records of truth. Three profound learning calculations, including the Convolutionary Neural Network (CNN), Deep Belief Networks (DBNs), Stacked Denoising Auto encoder (SDAE), have been created and executed. We built up a plan with 28 picture highlights and bolster vector machine to analyze the presentation of profound learning calculations with traditional PC helped determination (CADx) system. CNN, DBNs, and SDAE correctness’s are individually 0.7976, 0.8119, and 0.7929; our fabricated regular CADx precision is 0.7940 which is possibly lower than CNN and DBNs.
- Article View: 59
- PDF Download: 137