Adaptive transfer learning based feature selection and classification for lung cancer detection in ct image

Abstract

Lung cancer is the deadliest type of cancer worldwide, but the morbidity newlineand mortality rates can be significantly reduced if the diagnosis is performed newlineearly enough. Lung cancer screening is non-trivial because lung nodules can newlinepresent a wide range of opacities, commonly referred as textures, shapes, newlinedimensions and locations, and thus the experience of the specialist tends to play newlinean important role on the success of the nodule hunting and corresponding newlinecharacterization. But a major key obstacle to early detection is the absence of newlineobvious symptoms since the cancer has started becoming prevalent. Diagnosis newlineand the use of non-invasive imaging such as Computed Tomography (CT) newlinescreening is a potential solution. newlineRecently, Image processing technology has been widely used in several newlinemedical fields of detection and treatment levels. Lung cancer computer-aided newlinedetection and diagnosis systems can help to further increase the success of newlinescreening programs by identifying potential abnormalities to the radiologists. newlineHowever, to achieve accurate automatic analysis of these high-resolution newlineimages, a novel technique is required. For the recognition of lung cancer in the newlineimage processing, four different stages are analyzed namely (i) Pre-processing newline(ii)Segmentation (iii)Feature extraction and newline(iv) Classification. newlineIn the first approach, during the first stage of the process, the Adaptive newlineMedian Filtering techniques are implemented to reduce the noise in the CT newlineimages. newline