The Diagnosis of Lung Cancer in Computerized Tomography Image using Automatic Segmentation and Classification Approach

Abstract

Tumour detection is a process of crucial importance in oncology. Image enhancement, segmentation and classification are general problems in computer vision and are of dominant importance in medical imaging. The aim of this study is three-fold: (i) to investigate the strength and drawbacks of current medical image enhancement and tumour detection schemes, (ii) to develop and design a new approach to overcome the limitations, and (iii) to evaluate the new schemes using application scenarios vi for enhancement and tumour detection of lung CT images. The initial step is Anisotropic Diffusion Filter and Unsharp masking techniques which can eliminate the noise in the input images. The next step is the Adaptive cluster with super pixel segmentation process. It is implemented on an enhanced nodule image sequence for abnormal lung tissue forecast. Finally, the lung nodule images are obtained by utilizing a GWO based CNN classifier. Segmentation time for nodule portion order is 1.06s. The highest classification accuracy is 98.57% by the GWO based CNN method.