基於醫學影像建立個人化三維人工耳蝸模型的方法

Abstract

人工耳蝸植入手術對於聽覺感知的嚴重損失的患者，已經成為一個標準的臨床治療。電極陣列植入耳蝸時的位置，是影響語音識別的重要因素之一。人類耳蝸的三維模型對於手術前的規劃是有用的。三維人工耳蝸（Cochlear Implant, CI）模型經過高擬合表現後，對於CI的術後編程(programming)是有用的。 由於植入的電極陣列在二維X光影像提供很高的辨識率，因此二維X光影像最廣泛使用於手術後的診斷。然而，二維X光影像並無法獲得三維空間的資訊。電腦斷層影像（Computed Tomography, CT），對於耳蝸和植入的電極陣列提供高辨識度的影像，並且CT的每一張二維截面圖可以被用來重建三維影像。但CT卻很難區分每一個植入電極的位置，因為植入的電極在CT中會產生金屬假影。耳蝸模型的空間訊息可以從CT的截面影像中獲得。而電極陣列的螺旋模板以及每一個電極的位置可從二維X光影像中得知。同時使用兩種影像所得到的資訊，使人工耳蝸的建模能夠更完整。 由於目前實驗室所建立的電子耳模型是一個理想化的模型，但每個人都有生理結構上的差異，因此一個理想化的模型很難去符合每個人的狀況。藉由建立一個個人化的模型，使的實驗的結果會更符合患者的狀況。 在這篇論文中，基於先前研究的統計參數，以及由CI公司提供的電極陣列尺寸，分別建立了平均尺寸的鼓階（Scala Tympani, ST）和CI電極陣列模型。接著利用受試者的醫學影像建立患者的個人化的ST模型和電極陣列的模型。

Surgeries of CI for patients with acute loss of auditory perception have become a standard clinical treatment. The location of the electrode array has been one of the important variables used in speech recognition. The 3D model of the human cochlea can be useful for preoperative planning. Likewise, 3D CI model can be useful for postoperative programming by improving the fitting performance of CI. 2D X-ray images are the most widely used post-operatively because the 2D X-ray images of the implanted electrode array provide the highest resolution. However, the 2D X-ray images cannot obtain the 3D information. CT images of the cochlea and the implanted electrode array provide high-resolution images. However, CT images are difficult to distinguish the implanted electrodes, because the electrodes may cause metallic artifacts. The spatial information of the geometric model can be obtained from CT images. A spiral template of the implanted electrode array can be utilized to fit to the position of each implanted electrode in the X-ray images. The created CI model in our laboratory is an idealized model, but everyone has physiological structural differences. Therefore, an idealized model is very difficult to correspond with the condition of each patient. By creating a patient-specified model, the results of the experiment will be more close to the condition of each patient. In this thesis, based on the statistical parameters of previous studies and the size of the electrode array provided by CI companies, we create a mean size of the Scala Tympani (ST) model and the CI electrode array model. Then base on medical images of subject to create a patient-specified ST model and the electrode array of the patients.