本研究提出了三維動態規劃演算法，可以在一個三維矩陣中，尋找出一個最佳平面，可應用於磁振影像序列中股動脈的邊界的偵測。本研究的目的為在長時間極度耐力負荷的狀態下，觀察極限運動員的股動脈血管截面積隨著心臟周期的變化，方法是使用磁振造影儀（MRI）取得超級馬拉松比賽的運動員股動脈影像，量化隨著時間的推移而改變的股動脈截面積變化。在本次實驗中，使用了模擬血管的影像，藉由添加不同的雜訊及改變影像的對比度來增加影像的真實性，另外在模擬影像的實驗中加入斑塊(plaque)來測試此演算法有應付血管斑塊的準確性。最後我們利用三維動態規劃演算法來偵測下肢股動脈未施打顯影劑的真實磁振影像，再將自動偵測的結果與專家描繪的結果比較，其平均相對誤差小於3%。此結果表示相對於傳統的2D動態規劃演算法，我們所提出的三維動態規劃演算法能夠準確地偵測出磁振影像序列的血管邊界。
This study proposes a 3D dynamic programming algorithm to find an optimal surface in a 3D matrix. This algorithm can detect boundaries in an image sequence. The aim of this study is in the long term with an extreme endurance physical load to assess the cross-sectional vessel area within a heart cycle in ultra endurance runners. By means of using MRI to obtain the femoral artery of ultramarathon runners and to quantify the changes of transverse view of femoral artery with respect to time. We use phantom image studies with different added noises to test its accuracy. Moreover, we change the contrast between the vessel and its background to simulate the real MRA image. We also add a plaque in phantom images to test the accuracy of the proposed algorithm in dealing pathologic images. Finally we use 3D dynamic programming algorithm to detect the boundary of superficial femoral artery in real MRA image sequences without contrast injection. The accuracy is performed via comparisons between the manual tracings and automated results. The mean relative error rate is under 3%. The results demonstrate that the proposed method can perform qualitatively better than the conventional 2D dynamic programming for vessel boundary detection on MRA image sequences.
