 |
 |
 |
|
||
|
|||
|
|
2008 SURI Intern Year: Sophomore Research Project 1 : Development of a Fully-automated Gabor Wavelet Technique for Ventricular Segmentation and Volumetry Neuronal plaque formation, cortical degeneration, and beta-amyloid accumulation are characteristic of Alzheimer's disease (AD). These changes concurrently induce ventricular enlargement because of cortical tissue atrophies in diseased subjects. Development of AD in humans can be reproduced in the rabbit model by feeding them with food containing increased amounts of cholesterol and water with trace amounts of copper (Sparks et al. 2003). Our project is aimed at developing a fully-automated method to measure changes in ventricular volumes in rabbits. Two different brain and ventricular segmentation techniques will be applied independently prior to volumetry. The methods will then be compared to check consistency. The specific application of the developed process will be to track beta-amyloid plaque related atrophy in the cholesterol-fed rabbit model by monitoring the abnormal enlargement of the ventricles. The methods will further verify the usefulness of the rabbit brain as a model of AD in humans. There is currently no rabbit MRI atlas that can be used as a template or model based segmentation tool. Hence, standard intensity based and probabilistic segmentation methods like Standard Parametric Mapping (SPM) have failed (Sampath et al. 2007). Further long-term applications of the developed methods will be in the development of a rabbit brain MRI atlas and automated segmentation of various neuro-anatomical structures including the corpus callosum, hippocampus and the cerebral cortex. Project Goals
Click here to review the summary report of this project. _______________ Research Project 2: Review Motion-detection System to Reduce MRI Artifacts and Scan Time: Ghosting vs. Motion Motion by a scan subject at any point in k-space could lead to artifacts, ghosting and blurring in the resulting MR images. Background noise in the image could also possibly arise due to sharp changes in intensity caused by the finite number of sampling points and external RF signal interference during image acquisition. Motion by a subject being scanned is however believed to have the highest physiologically-induced negative effect on the usability of MR data. Often times, image ghosting can render an image useless to the radiologist. The common way of dealing with this problem is image registration-realignment of the image back to the orientation and location in which it ought to have been acquired. However image realignment and post-processing takes much longer than the data acquisition. (Cox et al., 1999). This project is aimed at reducing motion during MRI scans by monitoring head and knee movements in the scanner. The display of estimated head motion during scans could prove to be very useful. If the movements are too large, it is possible to reacquire the image time series immediately. This would help prevent tedious and time-consuming post processing of the acquired images. The apparatus includes a complete GE 3T scanner, the head and knee coils and an independent motion detector device consisting of four infrared channels developed by us. It consists of a setup attached to the head coil to monitor movement from four different points on the human head. These channels will be positioned to enable maximum detection of one type of head movement from at least two channels. The head movements examined are yaw (shake) and pitch (nod). Project Goals
Click here to review the summary report of this project. |
|
||||