A Computer-Aided Diagnosis-Based Analysis for a Model-Based Approach for Lung Segmentation

• Published in: SN computer science 2023-09, Vol.4 (5), p.601, Article 601
• Main Authors: Joseph, J. Sharmila, Ganesan, Srividhya, Chaudhary, Prachi, Mehra, Rajni, Saini, Himanshi, Pund, Sachin S.
• Format: Article
• Language: English
• Subjects: Artificial intelligence; Chronic obstructive pulmonary disease; Computed tomography; Computer Imaging; Computer peripherals; Computer Science; Computer Systems Organization and Communication Networks; Data Structures and Information Theory; Disorders; Image segmentation; Information Systems and Communication Service; Light beams; Lung cancer; Lung diseases; Magnetic resonance imaging; Medical imaging; Morphology; Original Research; Pattern Recognition and Graphics; Research Trends in Computational Intelligence; Respiratory system; Return on investment; Software Engineering/Programming and Operating Systems; Tomography; Vision
• ISSN: 2661-8907; 2662-995X; 2661-8907
• DOI: 10.1007/s42979-023-02034-2

One of the most important and vital organs in the human body’s anatomy is the lung. The most prevalent medical disorders in the world today are lung ailments. The World Health Organization (WHO) has reported that lung cancer and chronic obstructive pulmonary disease (COPD) have high mortality rates. Lung disorders can be challenging to diagnose in oncology and radiology. Computer-aided diagnostic (CAD) systems often highlight worrisome or aberrant patterns, commonly known as “regions of interest”, after receiving inputs from medical imaging modalities including computer tomography (CT) and magnetic resonance imaging (MRI). The return on investment can be evaluated based on its form, size, pattern of occurrence now, and expected future development. Segmentation in a medical image is difficult because of elements, such as the scanning environment, the quick changes in the strength of the light beams from the scanners, and the variations in noise. Segmenting the lung’s many components gets more difficult as microscopic differences between healthy and diseased lung tissues increase. This necessitates the development of a method for more accurate extraction of the lung model. A model-based segmentation approach was introduced to extract the lung’s shape and link it to a reference model. The suggested model builds sound reference models and compares the shape features to the input slices as a correlation metric. As compared to other frequently used methodologies, the proposed segmentation strategy produced better results, according to a numerical study. It was mathematically compared to other popular segmentation methods including region expanding and flood fill. The proposed method has higher accuracy (98%), specificity (95%), and sensitivity (98%) Precision by 94% than these other methods.