2020/08/13 by Parnian Afshar, Farnoosh Naderkhani, Afshar, Parnian +9 · 1 citation
Computer Science · Medicine · #AI in cancer detection #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences #FOS: Electrical engineering #Image and Video Processing (eess.IV) #Lung Cancer Diagnosis and Treatment #Machine Learning (cs.LG) #Radiomics and Machine Learning in Medical Imaging #electronic engineering #information engineering
paper · pdf · doi:10.48550/arxiv.2008.06072
openalex publication_date 2020/08/13 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
Lung diseases including infections such as Pneumonia, Tuberculosis, and novel\nCoronavirus (COVID-19), together with Lung Cancer are significantly widespread\nand are, typically, considered life threatening. In particular, lung cancer is\namong the most common and deadliest cancers with a low 5-year survival rate.\nTimely diagnosis of lung cancer is, therefore, of paramount importance as it\ncan save countless lives. In this regard, deep learning radiomics solutions\nhave the promise of extracting the most useful features on their own in an\nend-to-end fashion without having access to the annotated boundaries. Among\ndifferent deep learning models, Capsule Networks are proposed to overcome\nshortcomings of the Convolutional Neural Networks (CNN) such as their inability\nto recognize detailed spatial relations. Capsule networks have so far shown\nsatisfying performance in medical imaging problems. Capitalizing on their\nsuccess, in this study, we propose a novel capsule network-based mixture of\nexperts, referred to as the MIXCAPS. The proposed MIXCAPS architecture takes\nadvantage of not only the capsule network's capabilities to handle small\ndatasets, but also automatically splitting dataset through a convolutional\ngating network. MIXCAPS enables capsule network experts to specialize on\ndifferent subsets of the data. Our results show that MIXCAPS outperforms a\nsingle capsule network and a mixture of CNNs, with an accuracy of 92.88%,\nsensitivity of 93.2%, specificity of 92.3% and area under the curve of 0.963.\nOur experiments also show that there is a relation between the gate outputs and\na couple of hand-crafted features, illustrating explainable nature of the\nproposed MIXCAPS. To further evaluate generalization capabilities of the\nproposed MIXCAPS architecture, additional experiments on a brain tumor dataset\nare performed showing potentials of MIXCAPS for detection of tumors related to\nother organs.\n