Segmentation And Assessment Of Olfactory Bulb Based On MRI Images

Olfactory Sensation plays an important role in human’s daily life. Research on human anatomy indicates that olfactory bulb (OB) and tract (OT) functions as a media or a transfer station in the path of olfactory transmission and formation. Large amounts of clinical experiments indicate that the volumes of OB could reflect the olfactory sensibility significantly. Data from latest research shows that, dysfunction or loss of olfactory could have some relation with pathological changes of neural networks like temporal lobe (TL) or hippocampus in the brain, and act as an early symptom of central nervous system diseases, such as Alzheimer’s Disease (AD), Parkinson’Disease (PD) or schizophrenia. Olfactory loss caused by Craniocerebral trauma and Kallmann syndrome may decrease in OB volumes and lose a part of olfactory sulcus. Thus, measurement of OB volumes values significantly on the prediction and early detection of nervous system diseases.Since OB and OT locate in depth in the brain on anatomy, clinical techniques could not detect the pathological changes of them. High resolution magnetic resonance imaging (MRI) becomes an ideal detection method for measurement of OB volumes. MR image can distinguish pathological tissues with the normal through application of different imaging sequences combined with physiological features like flowing or diffusion, especially for brain and soft tissue imaging.Medical image segmentation as a critical computer aided diagnose method is the basic step of3D marking of the volumes and location of brain structures, making anatomy atlas, surgery navigation and visualization, significant and valuable for orientation of pathological changes and analysis of brain. Since the fuzzy, multi-mode, uncertainty features of medical images, the segmentation algorithm varies, and the quality of segmentation should be highly exact. Segmentation based on model, such as active contours (snake) and Chan-Vese model is an efficient research direction. The paper studied deeply into principles of active contour model (ACM) and ACM without edges (Chan-Vese model), which used level set method and region information, and the mathematical and physical meanings. Also, combined with seeded region growing generated a semi-automatic algorithm to extract OB from brain MR images successfully.The T1W1brain MR images of21normal adult volunteers aging from20to30were used. Statistics analysis were applied to OB volumes of the21people, and the result proved the validity of the segmentation algorithm generated in this paper.