Article Preview
TopIntroduction
Image segmentation can visualize the region of interests (ROIs) (Seung-Hyun Lee, Jaekyoung Moon and Minho Lee, 2006) for the scanned MRI images. Its approaches are widely used in the field of medical sciences. The early stages of brain tumors detection with suspicious regions are the current research era in image segmentation. The brain is connected with billions of neurons and giant cells. For genetic issues, the human body procreates infected organs or cells in the brain and causes tumors. The tumors or inlaid masses in the brain need to be detected at early stages for curing the issues of the severe ailment. Different digitized types of equipment like MRI (J. Liu and L. Guo., 2015), computed tomography (CT) (V. Venkatesan, Sandhya and Jenefer, 2017), positron emission tomography (PET) (Juan Jose Vaquero, Paul Kinahan, 2017) are recommended for the scanning of the human brain. The classification of tumors is found in the human body with two types: benign and malignant, in which benign may cause the non-cancerous and malignant cause of cancerous cells. Such tumors are identified with suspicious regions (K Rajendra Prasad, Suneetha Rani, Suleman Basha, 2018) in the brain image (Sudipta Roy, Ayan Dey, Kingshuk Chatterjee, Prof. Samir K. Bandyopadhyay, 2012). Segmentation of images with suspicious and non-suspicious regions is a vital task in tumors detection. Traditional methods, such as morphological based segmentation and fuzzy-c-means, are the practical techniques which extracts the region of interests while performing the binarization segmentation of brain image. It helps to detect and recognize the tumors in the human brain with two modularity intensity levels with binarized segmentation in the techniques of morphological and fuzzy-c-means (Q. Li et al., 2018). These segmentation algorithms (Rao, C.S., Karunakara, K. 2021), (Srinivasa Reddy, A., Chenna Reddy 2021), are used to auto detect suspicious regions of brain images and uncover the irregular region boundaries in the ROI generations, which affects the quality of brain tumors image segmentation. Thus, this paper presents the proposed FCM-RSCP that initially extracts the unique features of brain image and then analyses regions' boundaries with the nearest neighboring pixels. Nearest neighboring pixels information is derived with a fuzzy concept to overcome irregular region boundaries of ROIs in brain image.
Essential work of proposed FCM-RSCP are summarized as follows:
- 1.
Extract the spatial data of brain image according to windowing of regions.
- 2.
The nearest neighboring pixels is derived based on 2D similarity metrics within the selected regions.
- 3.
Develop fuzzy update membership for the concept of the selected region and get the brain image with an update of the nearest neighboring pixels.
- 4.
Use the threshold for the creation of two modularity intensity levels of image.
- 5.
Initialize the segmented cluster (P Anjaiah, K Rajendra Prasad, C Raghavendra, 2018) points with maximum distance and generate the ROIs' spatial region boundaries, which results in effective brain tumor detection.