Waykule Jyoti M
Pulmonary cancer, also known as lung carcinoma, is the primary cause of cancer death worldwide. Every year, earlystage cancer detection using Computed Tomography (CT) could save hundreds of thousands of lives. However, analysing hundreds of thousands of these scans is a huge burden for radiologists, and they frequently experience observer tiredness, which can harm their performance. As a result, there is a requirement to efficiently read, detect, and evaluate CT scans. As a result, there is a requirement to efficiently read, detect, and evaluate CT scans. Using the midpoint of the lung cancer provided in the dataset, the author cropped 2D cancer masks on its reference image and trained a model with various techniques. The proposed system consists of many steps such as image acquisition, pre-processing, binarization, thresholding, segmentation and feature extraction. In first stage, Binarization technique is used to convert binary image and then compare it with threshold value to detect lung cancer. In second stage, segmentation is performed to segment the lung CT image and a strong feature extraction method has been introduced to extract the some important feature of segmented images. Extracted features areused to train the neural network and finally the system is tested any cancerous and noncancerous images.
Published Date: 2021-06-30; Received Date: 2021-06-14