China 
Africa 
Explore chapters and articles related to this topic
Role of Image Processing in Artificial Intelligence and Internet of Things
Published in Lavanya Sharma, Mukesh Carpenter, Computer Vision and Internet of Things, 2022
Researchers have achieved extraordinary advances in unleashing the power of AI on applications related to image processing. Techniques of image processing collaborated with AI are extensively utilized in different domains, such as medicinal drugs, law enforcement, cybersecurity, and retail. AI integrated with image processing has paved the way for computing devices to acquire new functionalities by using the method of ML [6–28,33–42]. Some of these areas are as follows: Image categoryObject recognitionObject trackingImage generationImage retrieval
Deep Learning in Brain Segmentation
Published in Saravanan Krishnan, Ramesh Kesavan, B. Surendiran, G. S. Mahalakshmi, Handbook of Artificial Intelligence in Biomedical Engineering, 2021
Intensity thresholding, pixel clustering, and histogram-based methods are known as conventional image processing segmentation methods. These methods rely solely on the intensity information and distribution of pixels in the image to segment objects of interests. The convoluted structures of the brain and subtle discrepancy between normal tissues and lesions prove to be too complex for these simple methods to capture. Machine learning, a branch of artificial intelligence (AI) that uses pattern recognition techniques and learnable parameters, has shown the capabilities to replace simple image processing as the method of choice for image segmentation. Methods without stacked layers or deep representation are known as “traditional” machine learning, which includes algorithms like support vector machine (SVM), random forest (RF), and Markov random field (MRF). While demonstrating better performance than image processing methods, traditional machine learning still does not scale well to high-dimensional data such in three-dimensional (3D) or four-dimensional (4D) MRI sequences. Furthermore, it requires meticulous manual feature extraction and may show poor generalization facing different scanners.
Automatic Detection of Brain Tumor using NSR Filter and K-means Clustering
Published in P. C. Thomas, Vishal John Mathai, Geevarghese Titus, Emerging Technologies for Sustainability, 2020
P. Athira, Therese Yamuna Mahesh
The segmentation is a digital image processing method, which divides an image into various regions, such that the pixels within the region have similar characteristics. In the case of MRI brain image, separation of different tumor tissues from normal tissues is considered as the segmentation process. In the medical technique, segmentation of brain tumor is done manually. The manual segmentation of tumor from the images requires huge processing time and may produce inaccurate results. To help doctors for diagnosis, treatment of tumor and to help researchers for studying the brain activities, the research in automatic segmentation techniques of brain tumor are gaining more importance. Furthermore, segmentation of brain tumor is challenging task because of its unpredictable shape and appearance.
An efficient stacked ensemble model for the detection of COVID-19 and skin cancer using fused feature of transfer learning and handcrafted methods
Published in Computer Methods in Biomechanics and Biomedical Engineering: Imaging & Visualization, 2023
Recently, the predominant approach for image processing operations like feature extraction, segmentation and classification is convolutional neural networks (CNN) (ElOuassif et al. 2021; Shekar and Hailu 2022). CNN has several forms starting from the original LeNet (LeCun et al. 1998) which has only five layers. In this paper, we use one of the best and well-performed variants of CNN, DenseNet-169 (Huang et al. 2017), which is trained on the ImageNet dataset (Russakovsky et al. 2015) and contains 169 layers (165 convolution +3 transition +1 classification) grouped in 4 dense blocks. DenseNet-169 is chosen because it is easier and faster to train with no loss of accuracy due to the improved gradient flow. This algorithm is formed by connecting all layers directly to each other to secure the flow of the highest information between layers. Each layer receives input from all prior layers and gives its output feature-map to all successive layers. Figure 3 illustrates the layout of the DenseNet algorithm. We extract features from the skin lesion and COVID-19 image datasets by using transfer learning a pre-trained DenseNet-169. We remove the top layer and freeze the other layers. The prediction output from the last dropout layer is taken as the feature vector.
A Smart and Secured Approach for Children’s Health Monitoring Using Machine Learning Techniques Enhancing Data Privacy
Published in IETE Journal of Research, 2023
Labelling data for object recognition are challenging since there are several ways used to train algorithms that learn from data sets and anticipate the results. Image segmentation is a subset of digital image processing that focuses on dividing an image into distinct segments based on its characteristics and qualities. Image annotation is a way of labelling pictures including points of interest in order to make them identifiable to systems. The objective of image segmentation is to simplify or transform an image’s representation into something more relevant and easier to evaluate. It might be difficult to train an image segmentation model on new images, especially when you have to label your own data. The fundamental objective of image segmentation is to simplify the image so that it can be analysed more easily. Image segmentation is a method of breaking down a digital image into several subsets called Image segments, which serves to reduce the complexity of the image and make further processing or analysis of the image easier. In simpler terms, segmentation is the process of assigning labels to pixels. Image segmentation is commonly used to find objects and boundaries (lines, curves and so on) in images (Figure 7).
Research perspective and review towards brain tumour segmentation and classification using different image modalities
Published in Computer Methods in Biomechanics and Biomedical Engineering: Imaging & Visualization, 2022
Image segmentation is the major step in image processing that distinguishes the given images into different regions in the voxel format or divides the image into various objects. The images segmentation is required for making the image representation without any noises. These segmented images are used for detecting the objects, similar data in digital images, and also its boundaries. The image segmentation uses the regions, edges, and boundaries for the identification process. Tumour segmentation has numerous techniques based on region, soft computing, and some other types. Brain tumour segmentation is processed for determining the cancerous brain tissues and also performs automatic labelling them according to their tumour types. The brain tumour segmentation techniques are categorised and described in Figure 3.