Article Preview
Top1. Introduction
The disease has different stages whereby it starts from the small tissue and spreads throughout the different areas of the lungs by a process called metastasis. It is the uncontrolled growth of unwanted cells in the lungs. It is estimated that around 12,203 individuals had lung cancer 2016, 7130 males and 5073 females deaths from lung cancer in 2016 were 8839.
In recent years, advanced analysis of medical imaging using radiomics, machine, and deep-learning, including convolutional neural networks (CNNs), has been explored. These approaches offer great promise for future applications for both diagnostic and predictive purposes. CNNs are non explicitly programmed algorithms that identify relevant features on the images that allow them to classify an input object. They have been applied in various tasks such as detection (e.g., breast lesions on mammographic scans), segmentation (e.g., liver and liver lesions on computed tomography (CT)), and diagnosis (e.g., lung lesions on screening low-dose CT).
CNNs are a machine-learning technique based on an artificial neural network with deep architecture relying on convolution operations (the linear application of a filter or kernel to local neighborhoods of pixel/voxels in an input image) and down sampling or pooling operations (grouping of feature map signals into a lower-resolution feature map). The final classification or regression task relies on higher-level features representative of a large receptive field that is flattened into a single vector. The development of an algorithm entails (a) selection of the hyper parameters, (b) training and validation, and (c) testing. The hyper parameters include the network topology, the number of filters per layer, and the optimization parameters. During the training process, the dataset of input images (divided into training and validation sets) is repeatedly submitted to the network to capture the structure of the images that is salient for the task. Initially, the weights for each artificial neuron are randomly chosen. Then, they are adjusted at each iteration, targeting minimization of the loss function, which quantifies how close the prediction is to the target class. The performance of the trained model is then evaluated using an independent test dataset. This is also aimed at assessing whether an “over fitting” has occurred. The over fitting problem can arise in the case of limited datasets with too many parameters compared with the dataset size, in which case a model “memorizes” the training data rather than generalizing from them.
Lung cancer is one of the most dreadful diseases in the developing countries and its mortality rate is 19.4% (Globocan, n.d.). Early detection of lung tumor is done by using many imaging techniques such as Computed Tomography (CT), Sputum Cytology, Chest X-ray and Magnetic Resonance Imaging (MRI). Detection means classifying tumor two classes (i) non-cancerous tumor (benign) and (ii) cancerous tumor (malignant) (Livemint, n.d.). A number of researches on the image processing techniques to detect the early stage cancer detection are available in the literature. But the hit ratio of early stage detection of cancer is not greatly improved. With the advancement in the machine learning techniques, the early diagnosis of the cancer is attempted by lot of researchers. Neural network plays a key role in the recognition of the cancer cells among the normal tissues, which in turn provides an effective tool for building an assistive AI based cancer detection. The cancer treatment will be effective only when the tumor cells are accurately separated from the normal cells Classification of the tumor cells and training of the neural network forms the basis for the machine learning based cancer diagnosis.
The disease has different stages whereby it starts from the small tissue and spreads throughout the different areas of the lungs by a process called metastasis. It is the uncontrolled growth of unwanted cells in the lungs. It is estimated that around 12,203 individuals had lung cancer 2016, 7130 males and 5073 females deaths from lung cancer in 2016 were 8839.
This is used as a classifier to detect early lung cancer to improve the accuracy and sensitivity of the system in preference to other machine learning languages. In this software we are going to detect early signs of lung cancer. Whereas, we are not going to ensure to measure the size of nodule and the stages of lung cancer.