NATIONAL YUNLIN UNIVERSITY OF SCIENCE AND TECHNOLOGY
发明人:
SHEN, DAY FANN,沈岱范,沈岱範,CHENG, CHE FU,张哲辅,張哲輔,LIN, GUAN YI,林冠毅,林冠毅,SYU, WUN CING,徐文庆,徐文慶,LI, YA WUN,李雅雯,李雅雯
申请号:
TW100102692
公开号:
TWI399194B
申请日:
2011.01.25
申请国别(地区):
TW
年份:
2013
代理人:
摘要:
This invention relates to an improved MRI knee cartilage semi-automatic segmentation method, featuring Cellular Automata (CA) technique. The process includes: (a) pre-processing of MRI image sequence (b) human selection of rectangular area that cover knee cartilage (c) initial seeds detection and segmentation (d) Human selection of foreground and background seeds (e) Cellular automata segmentation (f) Human inspection of segmentation results and modifications through provided interface (g) refinement of segmentation. The above procedure provides a solution to the knee cartilage segmentation: which effectively resolves the problem of difficult-to-distinguish between the knee-cartilage and surrounding tissues.本發明係為細胞自動機技術為特色之半自動MRI膝蓋軟骨影像分割改良方法,其步驟為(a)自動MRI影像序列前處理;(b)人工選取影像中涵蓋膝蓋軟骨之區域;(c)自動進行種子點偵測及初始分割;(d)人工選擇前景及背景種子點;(e)將(d)之資料輸入給細胞自動機以進行自動分割;(f)人工檢視分割結果並透過系統之介面作修正;(g)根據(f)之介面輸入資料自動作精細切割,重複(f)(g)直至滿意為止,本方法有效穩定解決因軟骨像素與周遭組織像素過於相近所導致的分割失敗及時間過長等問題。(a)、(b)、(c)、(d)、(e)、(f)、(g)‧‧‧步驟流程