AN INNOVATIVE MECHANISM FOR SEGMENTATION TECHNIQUES ON FETAL BRAIN MRI IMAGES

Abstract

In the medical field, the intensifying use of Artificial Intelligence and Machine Learning tends to lower the time and accuracy in detecting illnesses in their early stages. AI will have a 10.4 percent influence on the Indian economy in 2030, amounting to 0.9 trillion dollars. More than 80 anomalies in a child's fetal development have been discovered yet. Machine Learning in clinical imaging brings in an exciting new era of reengineered and rethought clinical capabilities. Deep Learning makes it simpler for doctors to feel like they're wandering in the park, but with a few more desirable resources. Brain segmentation of fetal MRI is a new work on the completely automated treatment has been published. Automatic brain segmentation methods established for adult MRI cannot be used for studying fetal brain development, since the geometry and tissue morphology of the fetal brain are substantially different. In this thesis, two approaches for segmenting the brain and two methods to localize the fetal brain and its abnormalities are performed. It takes more time to manually identify the lesion tissues in the fetal brain. A convolutional neural network is a form of neural network that is most frequently applied to image processing problems. A computer recognizes artifacts in a picture and utilizes convolutional neural networks that are so essential in deep learning and artificial intelligence today. Object Detection is one of the thought-provoking tasks to be accomplished in computer vision. The basic aim of this initial phase of the study is to detect objects and to interpret the type of object. The research work includes the identification and description of clinical items. The model is trained in combination with photographs and images using YOLO v3 object detection techniques to identify the medical objects. The model is programmed in the cloud system to do well in a short time of preparation. The work summarizes the model that has been developed with the potential to recognize and detail unknown objects, viii whether static or moving in a real-life context. The program not only shows knowledge in the form of text but also spells out text in an artificial voice to help you easily understand the object. The model has an accuracy of 98.62% to detect the objects in the clinical practices.