Hippocampus segmentation in MR Images with deep learning

نوع: Type: thesis

مقطع: Segment: masters

عنوان: Title: Hippocampus segmentation in MR Images with deep learning

ارائه دهنده: Provider: Alireza Sadeghi

اساتید راهنما: Supervisors: Dr. Hassan Khotanlou - Dr. Muharram Mansoorizadeh

اساتید مشاور: Advisory Professors: Dr. Mohammadreza Rezaeiyan

اساتید ممتحن یا داور: Examining professors or referees: Mir Hossein Dezfoulian - Dr mahlagha afrasiabi

زمان و تاریخ ارائه: Time and date of presentation: 2023-09-19 10:30

مکان ارائه: Place of presentation: faculty of engineering class No. 27

چکیده: Abstract: The Hippocampus is a small, intermediate, and subcortical brain structure that is associated with both long-term and short-term memory. The shape and appearance of the hippocampus can change under the influence of factors such as neurodegeneration or Alzheimer's disease. In this research, the goal is to identify and segment the hippocampal region from Magnetic Resonance Imaging (MRI) scans. Hippocampal segmentation from MRI images is of great importance in neurological and psychiatric research and can also be used in the investigation of diseases such as Alzheimer's, epilepsy, and schizophrenia. One of the reasons for the necessity of analyzing hippocampal images, as mentioned, is the prediction of the likelihood of developing Alzheimer's. If this disease is detected early and treatment begins in the initial stages, the chances of improvement are higher, and the effectiveness of treatment methods is also greater. The high cost associated with manual hippocampal segmentation has led to research in the field of automatic hippocampal segmentation from medical images. One of the main challenges in hippocampal segmentation from medical images is the small size of the region, making it difficult to identify the region with the naked eye. The main objective of this research is to provide a model for segmenting and accurately determining the boundary of the hippocampal region in magnetic resonance imaging. Deep learning in computer vision plays a key role today and is used for various purposes such as image recognition, face detection, image segmentation, and medical image analysis. In this research, an ensemble model is presented using a pre-trained fuzzy mask for hippocampal segmentation with the use of deep convolutional neural networks. To improve the accuracy of the final model in this research, a fuzzy mask was created using a convolutional neural network model. The use of a pre-trained fuzzy mask with the removal of unnecessary parts of the MRI image reduced the complexity of the final ensemble model and increased the accuracy of hippocampal segmentation. After masking the images, an ensemble model consisting of convolutional neural networks was used for hippocampal segmentation. The final model is capable of accurately identifying the hippocampus if it is present in a given MRI scan. Subsequently, after creating the model with a portion of the available dataset, the model was evaluated. The evaluation results, compared with similar research, indicate favorable and acceptable results in the field of hippocampal segmentation.

فایل: ّFile: Download فایل