The International journal of analytical and experimental modal analysis ISSN NO:0886-9367 Volume XII, Issue IX, September/2020 Page No:1103 Lung Cancer Prognosis by Implementing Various Evolutionary Image Processing Steps Prajakta S. Kale #1 , Prof. Kailash D. Kharat #2 , Dr. Santosh K.Yadav #3 #1&2 AssistantProfessor, Department of Computer Science & Engineering Government Engineering College, Aurangabad, Maharashtra, India #3 President cum Director, Department of Computer Science & Engineering JJT University, Jhunjhunu, Rajasthan, India 1 prajaktakale@geca.ac.in 2 kailashdkharat@geca.ac.in 3 drskyadav@hotmail.com Abstract— with the latest technical developments in healthcare, the emphasis has always been upon improving exi sting systems’ speed and accuracy. Increasing survival rate of the patient from lung malignance is one of such challenge which interests researchers. lung cancer survival rate of patients depends upon the stage in which the cancer is detected. If detected in early (first) stage, there are more chances of the diseases to be cured and the survival chances of the patient are more than 50%. This requires the advanced analysis because the symptoms normally overlap with other lung diseases. CT scan images are widely used all over the world for this analysis. Image processing contributes significantly in analysis of the CT scan images. This paper sheds light on current literature strategies, with thorough studies of each phase. A comparative analysis has been presented. Final steps for the detection of lung cancer are then proposed based on the conclusions of the analysis. Keywords— CT Images, Image Processing, Lung Cancer Detection, Image Enhancement, Image Segmentation. I. INTRODUCTION Cancer stands out in the list of fatal diseases being the second greatest reason of mortality in the world. [1] Lung cancer is the worst of all types of cancers causing the greatest contributor of deaths, which increases with each year. According to the Global Burden Cancer, there were total of 1, 80, 78,957 cases with 11.6% cases of lung cancer occupying the first position in the list. The number of deaths due to cancer in the same year was 95, 55,027 with 18.4% of deaths caused by lung cancer, again the largest of all cancers. [2] The survival rate after detection of the cancer is very low, only up to 15%. There is a direct relation between early detection of lung cancer and the survival rate. Research proves that early detection can improve this rate up to 50%.[3] Therefore, the need to improve the techniques of primordial findings for this disease to increase the endurance scale in terms of time as well as accuracy is the need of the hour. For the precision of diagnosis, radiologists need an automatic system for second opinion. The evolution in the automation techniques have proved that the Computer-Aided Diagnostic Systems are more capable of identifying lung nodules and detecting lung cancer as they can detect even a small nodule of size 6mm. This brings the light of concern on increasing precision of automated systems. Various image processing and segmentation techniques have been suggested in the early research in the same direction. [4] CT (Computed- Tomography) scan images give well results as associated to another types of figures such as X-rays, MRI, and others. [5] The CT scan images are 3D images which provide wider perspective of analysis. However, these images need a lot of pre-processing before the extraction process. There has been a wide research in this field presenting various combinations of these processes. The dataset used for experiments is real time dataset from Lung Image Database Consortium (LIDC). The images directory has been originated by the National Centre for Cancer for research and development of lung cancer diagnostic systems. There are about 1018 cases in the database by eight companies and seven centres of academic. This paper throws light on some of the research in the literature and performs a comparative analysis of the same on the mentioned dataset.