Enhanced convolutional neural network for non-small cell lung cancer classification

<p>Lung cancer is a common type of that causes death if not detected early enough. Doctors use computed tomography (CT) images to diagnose lung cancer. The accuracy the diagnosis relies highly on doctor's expertise. Recently, clinical decision support systems based deep learning valuable recommendations doctors in their diagnoses. In this paper, we present several models detect non-small cell CT and differentiate its main subtypes namely adenocarcinoma, large carcinoma, squamous carcinoma. We adopted standard convolutional neural networks (CNN), visual geometry group-16 (VGG16), VGG19. Besides, introduce variant CNN augmented with block attention modules (CBAM). CBAM aims extract informative features by combining cross-channel spatial information. also propose variants VGG16 VGG19 utilize vector machine (SVM) at classification layer instead SoftMax. validated all study through extensive experiments dataset. Experimental results show supplementing leads consistent improvements over vanilla CNN. Results VGG SVM classifier outperform original VGGs significant margin.</p>