電腦斷層掃描決定冠狀動脈分叉之最佳投影角度演算法：與傳統血管攝影之比較

Determination of the Optimal Projection Angles for Coronary Bifurcation by Analyzing Coronary Computed Tomographic Angiography: Comparison with Conventional Invasive Angiography

林邑城(Yi-Chen Lin)

指導教授 ： 陳中明

國立臺灣大學/工學院/醫學工程學研究所/碩士(2018年)

https://doi.org/10.6342/NTU201803458

摘要

心臟疾病的死亡率在台灣僅次於惡性腫瘤，其中最常見者為冠狀動脈疾病。不良的飲食與生活習慣，如抽菸、高血壓、高膽固醇指數、糖尿病或缺乏運動等皆為冠狀動脈疾病的危險因子。動脈粥狀硬化是斑塊聚積硬化的慢性過程，常好發於冠狀動脈分叉處。依據病情的嚴重程度將採取不同的治療。其中，經皮冠狀動脈手術的支架置入侵入性低且恢復較快，逐漸成為血運重建方式中治療冠狀動脈疾病的首選。然而，在放置支架的過程中，醫生根據個人經驗或手術準則所採取傳統投影角度有時無法提供足夠好的視野，若支架未對病變位置完整涵蓋或突出，可能會造成再狹窄與支架栓塞。此外，由於分叉夾角也影響支架技術的選用，本研究將建立一個電腦斷層掃描之電腦輔助支架放置系統，最佳投影角度和分叉角度的資訊可以幫助醫生在手術前選擇合適的支架，並透過術中冠狀動脈病變的最佳視野以更準確地放置支架。
本研究分析左、右冠狀動脈四個主要的分叉區域，包含LAD/LCX、LAD/diagonal、LCX/OM、PDA/PLA。首先，多假設追蹤的技術被用來提取影像上的冠狀動脈血管樹與血管中心線，並提出深度學習演算法改善血管分割的結果。在定義分叉平面中的分叉點與兩個分支點後，除了能計算分叉夾角，並可透過最小化分叉夾角差與分支縮短比例參數找出最佳投影角度。結果顯示，深度學習演算法在冠狀動脈血管分割中有著良好的表現。而最(次)佳投影角度在分叉夾角差與縮短比例較傳統投影角度來得低，代表其有更佳的手術視野，並透過傳統血管攝影驗證此分析結果。此外，在43個最(次)佳投影角度的統計分析中，發現各分叉位置並無集中的分布，且分叉夾角差的容許度中十度僅較兩度微幅地改善了縮短比例。
本研究提出了一個電腦斷層掃描之電腦輔助支架放置系統，除了讓醫生以分叉夾角資訊進行支架選擇的術前規劃，在臨床手術時更能以最佳投影角度的視野進行精確的支架放置，縮短心導管手術的時間、改善手術治療效果。

關鍵字

冠狀動脈疾病 ； 傳統血管攝影 ； 電腦斷層血管攝影術 ； 電腦輔助支架放置 ； 最佳投影角度 ； 深度學習 ； 冠狀動脈分割

並列摘要

Heart disease is the second leading cause of death in Taiwan, and coronary artery disease is the most common type of heart disease. Bad eating habits and lifestyle such as smoking, high blood pressure, high cholesterol level, diabetes or lack of exercise are risk factors for coronary artery disease. Atherosclerosis is a chronic condition in which arteries harden through build-up of plaques, and often occurs at bifurcations of coronary arteries. Different treatments are taken depending on the severity of disease. Patient with mild symptoms are suggested to take medicines while revascularization is recommended in severe cases. Percutaneous coronary intervention with stent implantation is less invasive and recovers more rapidly, and gradually becomes the preferred revascularization modality in treating coronary artery disease. However, physicians often perform stent placement through some conventional projection angles based on personal experience or clinical guideline, of which the view may be inappropriate during the surgery. Restenosis or stent embolism may occur if a stent doesn’t completely cover the lesion. Besides, bifurcation angles impact the selection of stent technology, so the research is proposed to develop a computer-aided stent placement system based on CTA. The information on optimal projection angles and bifurcation angles helps physicians to choose the suitable stent before surgery, and to place stent more precisely by having an optimal view for the visualization of coronary artery lesions during surgery.
Four main bifurcation regions of the left and right coronary arteries were analyzed in our research, including LAD/LCX, LAD/diagonal, LCX/OM, and PDA/PLA. First, multiple hypothesis tracking was used to extract coronary artery trees and vessel centerline from an image, and a deep learning algorithm was proposed to improve the result of vessel segmentation. A bifurcation angle could be calculated after a bifurcation point and two branch points were defined in a bifurcation plane. Parameters of angle discrepancy and foreshortening ratio were proposed to be minimized to acquire an optimal projection angle. Result showed that the deep learning algorithm had a good performance in coronary artery segmentation. Besides, Angle discrepancy and foreshortening ratio of the (obtainable) optimal projection angle were smaller than those of conventional projection angles, implying a better visualization in surgical view which was further validated by conventional angiography. In addition, the statistical analysis showed that (obtainable) optimal projection angles were not densely distributed among 43 cases. The foreshortening ratio was slightly improved if the tolerance of angle discrepancy changed from two to ten degrees.
A computer-aided stent placement system based on CTA is proposed in the research. Physicians are able to make pre-clinical planning by getting information on bifurcation angle to choose the suitable stent, and can place stents precisely under the view of optimal projection angle, which shortens the time spent during the surgery and improves the clinical outcome.

並列關鍵字

coronary artery disease ； conventional invasive angiography ； computed tomographic angiography ； computer-aided stent placement ； optimal projection angle ； deep learning ； coronary artery segmentation

參考文獻

[1] Benjamin, Emelia J., et al. "Heart disease and stroke statistics—2017 update: a report from the American Heart Association." Circulation 135.10 (2017): 146.

[2] Srinivas, Sunil Kumar, et al. "Incidence, predictors, clinical profile, management and outcome of patients with isolated left main coronary artery ostial disease." Indian heart journal 70.2 (2018): 214-219.

[3] Ashley, Euan A., and Josef Niebauer. "Cardiology explained." Remedica (2004).

[4] Bhatia, Sujata K. "Coronary Artery Disease." Biomaterials for Clinical Applications (2010): 23-49.

[5] Bergheanu, S. C., M. C. Bodde, and J. W. Jukema. "Pathophysiology and treatment of atherosclerosis." Netherlands Heart Journal 25.4 (2017): 231-242.