S Berlin Shaheema
@stellamaryscoe.edu.in
Assistant Professor, Artificial Intelligence and Data Science
Stella Marys College of Engineering
RESEARCH, TEACHING, or OTHER INTERESTS
Computer Vision and Pattern Recognition, Computer Science, Artificial Intelligence
Scopus Publications
Scholar Citations
Scholar h-index
Scholar i10-index
Scopus Publications
S. Shiny, J. Jasper, R. Megiba Jasmine, and S. Berlin Shaheema
AIP Publishing
Berlin Shaheema S and Naresh Babu Muppalaneni
IEEE
Automated brain tumor classification is one among the most complicated and popularly used applications of medical imaging. Manual diagnosis of brain tumors is complex and inefficient. Therefore, identifying the appropriate type of tumor at an early stage plays a significant role in choosing the exact treatment plan. Therefore, a new method based on the YOLOv7 architecture was suggested to extract features, select features, identify tumors, and classify them from multimodal brain images. The brain tumor is categorised using the classifier as glioma, meningioma, non-tumor and pituitary tumors to increase confidence in tumor detection problems. YOLOv7 method is assessed utilizing the brain MRI dataset which includes Br35H dataset, Figshare dataset, and SARTAJ dataset, consisting of 7023 MRI images. Overall effectiveness of the suggested technique is assessed in terms of precision, specificity, sensitivity, F1-score, and accuracy as 98.73%, 99.69%, 99.81 %, 98.56% and 98.73 % respectively. The classification outcomes demonstrates that described method can be useful to help clinicians make quick and accurate decisions related to the diagnosis of brain tumors.
Liping Yu, S. Berlin Shaheema, J. Sunil, Vediyappan Govindan, P. Mahimiraj, Yijie Li, Wasim Jamshed, and Ahmed M. Hassan
Walter de Gruyter GmbH
Abstract Breast cancer diagnosis relies on breast ultrasound (BUS) and the early breast cancer screening saves lives. Computer-aided design (CAD) tools diagnose tumours via BUS tumour segmentation. Thus, breast cancer analysis automation may aid radiologists. Early detection of breast cancer might help the patients to survive and in context with this many approaches have been demonstrated by different researches, however, some of the works are weak in the segmentation of breast cancer images. to tackle these issues, this study propose a novel Hybrid Attendseg based gravitational clustering optimization (HA-GC) method which is utilized to segment breast cancer as normal malignant, and benign. For this we have taken the dataset known as breast ultrasound (BUS) images. This method constructively segments the breast cancer images. Prior to the segmentation, pre-processing is carried out which can be used to normalize the images incorporated with the removal of unwanted noises and format the images Optimization selects the best qualities. An experiment is conducted and compared the results with the parameters such as Dice coefficient, Jacquard, Precision, and Recall and attained over 90% and ensures the usage of present work in the segmentation of breast cancer images.
Berlin Shaheema S and Naresh Babu Muppalaneni
IEEE
Realistic Scene understanding is a challenging task by recognizing instances along with the semantic scene. This work, efficient panoptic image segmentation through a deep UNet combined with the melody search optimization algorithm for realistic scene images understanding. The visually appealing things have a variety of constituent pieces that capture the unintended organisational linkages between the crucial components of the carefully considered objects and, as a result, group them altogether according to a perceptual organisation model. The melody search optimization algorithm reduces the Unet parameters, thereby improving the accuracy, UNet combines semantic and instance segmentation predictions to form panoptic predictions and reduces the computational time and memory resources. Hence, the scene understanding is performed without requiring knowledge of the physically inspiring things in advance. Our approach is evaluated on four challenging benchmarks: Mapillary Vistas, Indian Driving Dataset, KITTI, and cityscapes. The results show that our method performs better than the current state-of-the-art methodologies for comprehending scene photos. The performance of the presented method was assessed utilizing panoptic quality metrics and computational time.
Berlin Shaheema S, Naresh Babu Muppalaneni, and Jasper J
IEEE
Brain tumor is a terrible disease that affects people worldwide. The main reason behind the growth of brain tumors is the uncontrolled and abnormal development of cells in the brain. Early detection of such growth improves survival. Therefore, developing automated systems that detect abnormal growth will help radiologists make accurate diagnoses. This paper presents a new metaheuristic-based methodology for early detection of brain tumors using the Melody Search Optimization Algorithm with Extreme Softplus learning. In the pretreatment step, image quality is improved by intensity normalization in combination with an adaptive bilateral filter. Histogram Oriented Gradient (HOG) extracts slope features. Automatic brain tumor segmentation splits tumor into sub-regions using Melody Search Optimization Algorithms combined with Extreme Softplus Learning. The performance of the proposed tumor segmentation technique is assessed based on the images acquired from the database. Our experiments have demonstrated that improved segmentation can help avoid the next level of danger. Segmentation metrics were calculated and compared to manual depictions, such as dice similarity, Jaccard index, and percentage of relative error (RE %).
J. Jasper, S. Berlin Shaheema, and S. Berlin Shiny
Hindawi Limited
This paper addresses a novel and efficient algorithm for solving optimization problem in image processing applications. Image enhancement (IE) is one of the complex optimization problems in image processing. The main goal of this paper is to enhance color images such that the eminence of the image is more suitable than the original image from the perceptual viewpoint of human. Traditional methods require prior knowledge of the image to be enhanced, whereas the aim of the proposed biogeography based optimization (BBO) enhanced with blended migration operator (BMO) algorithm is to maximize the objective function in order to enhance the image contrast by maximizing the parameters like edge intensity, edge information, and entropy. Experimental results are compared with the current state-of-the-art approaches and indicate the superiority of the proposed technique in terms of subjective and objective evaluation.