| 摘要: | 目的:由於亞洲人得到肝臟疾病的機率很高，造成越來越多人想藉由肝臟移植手術解決肝臟疾病，在肝臟移植手術前，醫師或技術員需要於電腦斷層造影的影像上先模擬計算切割肝臟前後的體積與範圍，以免造成捐肝者剩餘肝臟體積過小而導致生命危險，本研究目的為建立一個半自動肝臟體積計算系統在施打顯影劑的腹部CT影像中執行，制定一個統一的標準，有助於醫師在肝臟移植手術之前的評估與計畫，並且有效減少人力和時間成本。
材料: 取20位肝臟捐贈受檢者有施打顯影劑的腹部電腦斷層影像進行演算法測試，影像格式為消除病人個人資料後的DICOM檔，從DICOM中獲得以利於計算肝臟體積大小的參數。利用MATLAB 2013a進行程式設計。
方法: 本實驗以人工方式給出初始中心點，利用肝臟灰度值分布進行運算，接著使用梯度向量流結合主動輪廓模型和傅立葉描述子進行邊界偵測，經過上述的運算之後將會得到肝臟全部序列的邊界，並接著進行分葉補償，最後從DICOM檔中獲得的參數去計算出肝臟體積。
結果: 實驗結果跟醫生描繪去計算誤差以及程式執行時間，跟人工描繪的比對結果為敏感性90.6%、特異性92.39%，PPV為94.87%，整個程式運算時間平均為257.77秒，可以有效地節省人力成本與時間。
討論: 從實驗結果中可以看出本實驗方法可以有效地將肝臟周圍器官排除，並且在整個運作時間包括人工介入的部分所耗費的時間一組序列也不超過5分鐘。
結論:從結果顯示我們的方法在肝臟邊界分割上有優秀的準確度和運算速度。在未來我們會著重在減少人工介入的部分，進而完成全自動肝臟分割程式。
Objective : Since Asians get a high probability of liver disease caused by more and more people want to solve a liver disease using liver transplant surgery, before liver transplant surgery, the doctor or technician needed in computer tomography images first simulated cutting the volume and scope of the liver before and after liver donation avoid residual liver volume were too small to cause life-threatening, the purpose of this study was to establish a semi-automatic computing system to perform liver volume in abdominal CT images with contrast, develop a uniform standard, help physicians before liver transplant evaluation and planning, and effectively reduce the time and manpower costs.
MATERIALS: 20 subjects with liver donors have applied to abdominal CT images with contrast to test the algorithm, the image format is DICOM files which eliminate the personal data of patients, access parameters in order to facilitate the calculation of liver volume from DICOM. Programming with the use of MATLAB 2013a .
Methods: In this study, manually give the initial center point, and use liver gray value distribution operations, and then use the gradient vector flow combined with active contour models and Fourier descriptors for computing , after the above operation will be entire sequence of liver border , perform lobes compensation and then use the parameters obtained from the DICOM file to calculate the volume of the liver .
Results: The results depicted with the doctors to calculate the error and program execution time , the sensitivity is 90.6%, specificity 95.57%, PPV is 94.87%, the entire program computing the average time of 257.77 seconds, can effectively save labor costs and time.
Discussion: As can be seen from the experimental results of the experimental method can be effectively exclude ambient liver organ , and the operating portion includes a time -consuming manual intervention of a set of sequences of time not more than 5 minutes.
Conclusion: The results show that our method on the border dividing the liver has excellent accuracy and speed of operation . In the future we will focus on reducing the part of the manual intervention, complete the automatic liver segmentation program. |
