篇名 | To evaluate hemodynamics of aqueductal CSF flowing by accessing different image segmentation methods |
---|---|
卷期 | 12:1 |
作者 | Ming-Yen Chen |
頁次 | 021-027 |
關鍵字 | aqueduct 、 image segmentation 、 CSF |
出刊日期 | 202012 |
Along with cardiac pulsation causes systolic expansion and diastolic shrinkage of the cerebral hemispheres, the change in brain volume hence results in a biphasic to-and-fro cerebrospinal fluid motion. Cine phase-contrast MRI has been proven an effective tool in quantifying the hydrodynamics of CSF through the aqueduct of Sylvius. Thus, this project aimed to acquire high spatial and temporal resolution images and to adopt proper image segmentation method. Here, this project will compare 3 methods, i.e. adaptive threshold segmentation, pulsatile-based segmentation and independent component analysis segmentation. In phantom study, lumen area and stroke volume were considered. As for lumen area estimation, there were 95±3%, 92±5% and 93±6% for adaptive threshold, modified PUBS and ICA, respectively. As for stroke volume estimation, there were 88 ± 6%, 91 ± 5% and 91 ± 4% for adaptive threshold, modified PUBS and ICA, respectively. In normal volunteers’ study, three different image segmentation methods were evaluated precision as index (3x3 average mask and salt-and-pepper noise testing) results. As for noise estimation, there were 82±8%, 85± 8% and 90±9% for adaptive threshold, modified PUBS and ICA, respectively.