Article Preview
Top1. Introduction
The requirement for intelligent health care systems is increasing day by day as human health diseases, such as brain tumors, liver cysts, breast tumors, kidneys, covid-19, etc. These ailments are diagnosed by various imaging techniques such as X-ray, ultrasound CT scan, and MRI scan. Doctors use these scanned images for diagnosis purposes (Gupta et al., 2020). One such important organ of the human body is the liver and it's important to diagnose every disorder related to the liver. The liver cyst is a rare disease. A cyst is a fluid or a solid mass. It is present either inside or outside the human body(Nathan & Mulholland, 2019)(Matsuura et al., 2017). Only 5% of the population has this condition. Patients usually do not have any symptoms. The presence of a cyst can be detected by using imaging techniques only. If these are not taken care of, cysts can sometimes be the reason for cancer. Cysts grow in size day by day(Oda et al., 2020). The captured computed tomography (CT)/magnetic resonance imaging (MRI) images typically have limited spatial resolution, low contrast, noise and non-uniform variability in intensity due to environmental effects. Therefore, the objects' distinctions are blurred and distorted and the meanings of the objects are not quite precise(Kaur et al., 2020). Image processing field is very helpful to enhance the quality of images so that each disorder can be diagnosed clearly. Image segmentation plays an important role in diagnosing such conditions. With the help of image segmentation, identification of any disease such as brain tumor, cyst, breast tumor, kidney, COVID-19, etc., from a medical image is very simple. Image segmentation process is useful for feature extraction and pattern recognition (Qureshi & Ahamad, 2018). The segmented chromosome images are used for feature selection in classifying chromosomes (Arora & Dhir, 2020). Many applications have been developed for image segmentation in recent years, but this is a challenging task(Anter et al., 2020). So, in this paper the segmentation process is used to detect internal cysts in a liver. To improve the visual quality, median filter is applied to the dataset (Joao et al., 2020). A median filter is a type of non-linear filter (Zhu & Huang, 2012)(Arasi & Suganthi, 2019). It is widely used in image processing field as compared to other filters. The main advantage of the median filter is that it keeps the edges while removing noise from images. These filters use a m×n neighborhood mask and salt-and-pepper noise to filter the image (Liu et al., 2020). To improve the efficiency of results, soft computing based approaches are used to support health care diagnosis. There are numerous types of soft computing approaches like Evolutionary algorithms (Abdel-Khalek et al., 2017), an artificial neural network (ANN)(Hamad et al., 2019), Fuzzy logic based algorithms(Kyi et al., 2019), particle swarm optimization algorithms (PSO)(Xiaoqiong & Zhang, 2020), nature-inspired algorithms and metaheuristic optimization algorithms. Some of the metaheuristic optimization algorithms are Harmony search (HS) (Srikanth & Bikshalu, 2020), Ant lion optimization (ALO)(Anter et al., 2020), Ant colony Optimization (Tao & Jin, 2007), bat algorithm(G. Zhou et al., 2015), Grey wolf optimization (GWO)(Kapoor et al., 2017), Whale optimization algorithm (WOA)(Abdel-basset et al., 2020), Firefly algorithm(Rodrigues et al., 2017), moth flame optimization (MFO), Teaching learning optimization algorithm (TLBO)(Rao et al., 2011) and many more. The segmentation process is done using different methods like clustering, region growing, region splitting & merging and multilevel thresholding based upon the image intensity values. In thresholding, the grouping of pixels is according to the threshold value (T) (S. Singh et al., 2020)(Upadhyay & Chhabra, 2019). Threshold level or value is selected for the segmentation purpose in many cases. Levels are of many types like binary level, tri-level, four levels and n levels. In the case of binary level, one selected value is used as a threshold that converts the whole image into a binary image. Here binary image means the intensity of each pixel is either 0 or 1. More than one level is called multilevel thresholding. Increasing the number of threshold levels means grouping pixels in more than two groups. As the level of thresholds increases, computation time also increases. Many algorithms have been developed till date to overcome this drawback. The major problem is a selection of optimal number of thresholds(Aziz et al., 2017).