Abstract:
In recent time the Lung Cancer disease is evolving as highly life-threatening disease for human beings, as per world health
organization lung cancer disease is becoming second largest cause of deaths as compared to all other types of cancer. Although the
prevailing available technology is striving to get more exposer in the field of medical science using Computer Assisted Diagnosis (CAD)
system, where image processing is playing crucial role for detecting the cancerous nodules in computer tomographic images. Augmenting
the Machine Learning techniques with image processing algorithms is becoming more comprehensive examination of cancer disease
in proposed CAD system. This paper is exhibiting the heuristic approach for lung cancer nodule detection, the purported model is
predominantly categorized the foremost tasks, which are Image Enhancement, Segmenting ROI (Region of Interest), Features Extraction,
Nodule Classification. In preprocessing, primarily the AMF filtering method is applied to eliminate the speckle noise in input CT image
of LIDC-IDRI dataset through, and quality of input image is improved by applying Histogram Equalization technique with Contrast Limited Adaptive approach. Secondly, in successive stage the Improved LevelSet (ILS) algorithm is used to segment the interest region
(ROI). Furthermore, the third step of projected work is applied to extract the definite learnable Texture Features and Statistical Features
from segmented ROI. Based on the extracted features in aforementioned stage are applied to pioneering improved Convolutional Neural
Network CNN architecture with Hybrid-Layer to classify the lung cancer nodule is either benign or malignant. Principally this research
is carried out by contributing to each stage of it, where novel concept of improved Hybrid-Layer Convolutional Neural Network (CNN)
is employed by optimizing and selecting the optimal weight using the Enhanced Cat Swarm Optimization (ECSO) algorithm. The
experimental result of proposed Hybrid-Layer CNN using weight optimization algorithm ECSO is achieved the accuracy of 93%, which
is comparatively efficient with respect to existing model such as DBN, SVM, CNN, Hybrid-Layer: WOA, MFO, CSO. Moreover, this
work provides conclusive decision on detected nodule is either benign or malignant.