先天性心臟病影像：電腦斷層掃描攝影、數值發育分析以及生物模型

Abstract

在台灣先天性心臟病發生的機率比全世界的平均還要高，且是佔產嬰有先天畸型的第一位，其重要性不可言寓。然而在這些患童身上，因彼此年齡的不同、病症別的不同、合併異常的不同以及嚴重程度的不同，造就出較少所有狀況完全相同的兩個病人，也因此個別化（individualized）的診治變得非常重要。 在診斷上，必需能提供處置介入前完完整整的資訊以供決策的重大參考。電腦斷層掃描攝影因其擁有目前最高解析度的斷面影像，以及在其所掃描的範圍內的所有構造，不會被遮蔽而一覧無遺，而且能在非常短的時間內，受檢者僅需平躺於檢查台上不動便能在相當安全的狀況下完成檢查，獲致幾乎所有關切問題的解答之多重優點，目前已經成為診斷先天性心臟病不可缺少的診斷工具。除了能將心臟血管內部的構造清楚的展現外，電腦斷層掃描攝影在肺靜脈、肺動脈、冠狀動脈及複雜性心臟病異常的詮釋上，有優於其它的診斷工具，而且電腦斷層掃描攝影擁有不限組織的特性，還可以將心臟血管周遭的所有構造，如氣道、肺部、胸縱隔腔的資訊同時的呈現，這是其它單一檢查工具所無法作到的好處。 隨著電腦軟、硬體科技的進步，將電腦斷層掃描攝影的資料，重組成三度空間立體影像的能力比起五~十年前可謂突飛猛進，將之應用到先天性心臟病的患童上，讓醫療人員及病人、家屬透過這些三度空間立體影像，得以看到像真正的心臟血管構造一般，或甚至現實醫療技術上不可能出現的虛擬實境，這些各式各樣組像的演算法有各自的用處，並在本研究中顯示其臨床上的效果，唯目前仍需有許多人工徒手的介入，其所花費的時間及要有專業技術訓練的基礎是執行上的困難。未來如果能利用建立完整之各先天性心臟病的電子數值化的資料庫，並求得均值代表模型後，再依此一模型從組織區分 (segmentation) 階段開始，就有可能作到半人工智慧般的將所需的三度空間立體影像重建出來，如此才能有辦法更廣泛的應用及推廣。這些數值化的虛擬電子心臟血管器官，除了可以建構上述資料庫外，更重要的是成為可以運算及統計的基礎，在現今基因體及蛋白質體快速發展下，由基礎醫學累積的大量知識，正嘗試要連結到組織或器官的等級，這些更接近臨床上的徵象，稱作表現型，卻僅有相對少的類別變項可供對應，就以心臟血管的型狀來說，先天性心臟病的可對應變項就是這些個別的疾病診斷別，如；大動脈轉位、法洛氏四重症…等，然而假設某一基因是調控主動脈弓的形成，那此一基因應會同時與大動脈轉位、法洛氏四重症…等眾多的先天性心臟病都會有相關，那麼研究的主題就會失焦；但是如果不管疾病別而能針對主動脈弓取得眾多個案的數值化資料，來呈現出〝表現型多樣性（phenotype polymorphism）〞，有機會可以透過資料探勘 (data mining) 與分析的手段，擔起由基因、蛋白質、胚胎發育到臨床形狀表現中的橋樑，或許能將一些未知的基因更快的付予義意。而這一切都可以從本研究，所成功並初步揭示之方法及驗證的數值化心臟血管發育資訊中，開始看出端倪。 對於先天性心臟病的研究與探索，最後還是得回到個別患者的介入治療，由於這些先天性心臟病患者間變化的因素大，尤其是複雜性的先天性心臟病，個別化的治療絕對是最好的方針，再加上心臟血管組織器官的矯治手術，最最好能在一次開刀中完全解決以避免不必要的傷害，所以術前的演練就顯得異常重要，透過快速成型術 (rapid prototyping)，本研究已經建立起一套作業流程可以完全複製一個活體的心臟模型，這模型除了可以提供一個絕對具體，並與患者一模一樣的複製心臟供診斷、討論、講解說明外，更重要的功能是可讓心臟血管外科醫師能在手術前一再的演練，並依此一等比例 (1：1)的模型，事先剪裁修飾好術中會用到的補片模板，好讓心臟血管外科醫師在手術當下能非常從容且肯定的幫助病人完成矯治，本研究所展示的初步成果，恰可成為未來更進一步應用的方向。 因此，本研究所揭示有關先天性心臟病的研究上，從診斷工具的開發、應用、延伸到實證的每一個步驟都是環環相扣的，這些成果更重要的是要回饋並滿足每一位先天性心臟病患者的個人化的需求，以增進個別病人最大的福祉。

The incidence of the congenital heart disease in Taiwan is higher than that was reported in the world. And, the congenital heart disease is the highest occurrence of congenital deformity of live birth in our country. Thereafter, the importance of the congenital heart disease in Taiwan is straightforward. However in these patients, owing to differences of age, disease entity, associated anomalies, severity.. etc, there are no two patients have the exactly same condition. So, individualization both on diagnosis and treatment become very important. In diagnosis, comprehensive information should be got before any intervention or disposition. Advanced computed tomography becomes an essential imaging tool in the diagnosis of the congenital heart disease by providing key evidence for decision making because it has the following advantages: highest spatial resolution in sectional imaging modalities, no limitation in the field of be examined, complete the study in very short time, non-invasiveness, simple procedure by lying of the patient on the examination table. Basically computed tomography can well delineate the internal structures of the disease heart. Computed tomography is superior to other imaging modalities in demonstration of the pulmonary vein, pulmonary artery, coronary artery and the complex geometry of the heart. Structures outside of the heart also can be well shown by computed tomography; they are trachea, lung, and mediastinum and chest wall. This character let the computed tomography to be a single modality that can overcome many fields of exploring that is hardly be achieved by other imaging modalities. Accompany with the rapid developing of the software and hardware in computer science, reconstruction of the raw data from the volume acquition by computed tomography scanner has great achievement. Using this three-dimensional reconstruction on the cases with congenital heart disease, patients, families and medical practices all could see the diseased heart in global and just like the real object on images. By providing the virtual reality, we also can perceive the image that cannot be identify in life. In this thesis, many algorithms that can get different effect on image reconstruction is demonstrated and discussed. However, it takes a lot of time for a specialist to manually manipulate these raw data in order to make this reconstruction. In the further, if we can set up a comprehensive electronic database of each types of the congenital heart disease and get a lot of averaged models, we will have the opportunity to make a semi-automatic algorithm in images reconstruction. At that time, the widespread apply of the three-dimensional reconstruction in the congenital heart disease will be feasible. These numerical electronic hearts will be the basic component of the electronic database. More important, these numerical cardiac models will be the platform for quantitative simulation in both calculation and statistics. Owing to the rapid development of the genomic and proteomics, the knowledge on basic medicine is quickly accumulated. However, at present there is no comprehensive interface for bridge this basic biomedical knowledge (genotype) to diverse clinical presentations (phenotype). In case of the shape of the congenital heart disease, there are only limited phenotypes (disease entities: such as transposition of the great arteries, tetralogy of Fallot … etc) that can be used to correlate to the genotype. However, there maybe a possibility that some particular one gene may control the development of the aortic arch, thereafter this gene should have impact on the final shape of the aortic arch in all kinds of the congenital heart disease. If researcher only takes one specific diseased heart into consideration and focus on this particular gene, then the correlation maybe not strong enough to get attention. On the other hand, if we could collection the electronic numerical shape of the aortic arch of any types congenital heart disease, maybe we can use data mining and analysis strategy to integrate from the genomic, developmental embryology to the clinical manifestation. Such strategy may fasten disclose the function of many genes of unknown function. This study demonstrates the preliminary application of this strategy to analyze the possible development embryology of the systemic venous return of the left upper body by using the numerical cardiovascular information. Individualized intervention in each patient with the congenital heart disease is the final practical purpose of the research and explore on this topic. Because of the wide variation of the involvement of each component in the congenital heart disease, individualized intervention on each patient is the best policy of the disposition particularly on those patients with the complex congenital heart disease. Additionally, cardiovascular surgery usually has only limited time of revision to prevent unnecessary injury let the pre-operative planning and simulation is extremely important. By way of stereolithography, this report has build up a systemic protocol and a reasonable procedure to replicate a living heart model. This model provides a solid presentation of exactly the same of the patients’ heart for diagnosis, discussion and interpretation. Moreover, surgeon can repeat and repeat practice on this model before really put the patient on the operative table and perform such surgical procedures. This model also can provide a real size template for cutting any augmented patch. The preliminary result on the application of this cardiac model revealed the improvement of the confidence both on surgical manipulation and patient education. In conclusion, the present thesis reports the integrate research on the imaging application of the congenital heart disease step by step, it first made by the development of the diagnostic tool, then continue by the wide range of the application of this tool, and finally extends to the verification of this tool. These results try to satisfy the need of each individual with different congenital heart disease to improve the quality of their life.