dc.contributor.author | Hope, Tyna A. | en_US |
dc.date.accessioned | 2014-10-21T12:33:24Z | |
dc.date.available | 2006 | |
dc.date.issued | 2006 | en_US |
dc.identifier.other | AAINR27641 | en_US |
dc.identifier.uri | http://hdl.handle.net/10222/54919 | |
dc.description | Preterm infants are susceptible to white matter damage (WMD), which is associated with cerebral palsy (CP) and cognitive impairment. Ultrasound (US) is the preferred imaging modality to detect WMD but suffers from poor sensitivity and specificity in the early postnatal period. To improve on existing diagnostic rates, quantitative measures incorporating new information are needed. Ultrasound texture measures have been shown to reveal diagnostic information about human tissue. In this research, unique texture measures are extracted using adaptive preprocessing and high-resolution feature enhancement. The clinical diagnosis of CP presently is made at 12 to 18 months. As it is desirable to detect the disease in its early stage, clinical B-mode images taken within days of birth are used in this research. | en_US |
dc.description | In this study, the images are not standardized but use the patient as his or her own control. Speckle is not removed as speckle may contain information. To test the hypothesis that ultrasonic texture in these early images are associated with patient outcome, a model using only texture measures is created and evaluated. The "Random Forest" algorithm is used to form the model. The design of the texture measures and the selection of the variables are performed with a data set distinct from the set used for design and evaluation of the model. The resulting model has an accuracy of 72.5%. Random noise would provide a model with 50% accuracy, and designating all patients as having CP would result in 54% accuracy. This result suggests that early quantitative texture measures contain diagnostic information relevant to patient white matter health. | en_US |
dc.description | Thesis (Ph.D.)--Dalhousie University (Canada), 2006. | en_US |
dc.language | eng | en_US |
dc.publisher | Dalhousie University | en_US |
dc.publisher | | en_US |
dc.subject | Engineering, Biomedical. | en_US |
dc.subject | Engineering, Electronics and Electrical. | en_US |
dc.subject | Health Sciences, Radiology. | en_US |
dc.title | Selecting and assessing quantitative early ultrasonic texture measures for their association with cerebral palsy. | en_US |
dc.type | Thesis | en_US |
dc.contributor.degree | Ph.D. | en_US |