利用醫學影像重建精確的完整人體模型

Abstract

正確的內部組織結構的人體虛擬模型對於醫學應用而言相當重要，例如手術模擬、構造分析等，然而，建構一個三維空間的精準虛擬模型通常需要有經驗的模型師花費大量的時間與精力來完成。我們在本篇論文提出了一套半自動化的方法來更有效率的建構之於特定目標的人體模型，我們不只將人體樣版模型轉換以符合目標外觀，還結合醫學影像使內部器官組織更加準確定位。我們透過求得 Laplacian 變形場來變形我們的樣版模型，而整套流程將能處理不同的人體醫學內部結構資料如電腦斷層掃描與核磁共振的切片資料。我們呈現了幾個變形結果，這些結果展示我們的方法在目標區域的準確度不只是在視覺上有效，而在醫學上也兼顧其內容的正確性，而我們也透過兩套不同的 Error Metrics 來評估我們的變形結果。

Virtual character with accurate internal anatomy is critical for medical applications, such as surgical simulation. However, constructing an accurate anatomic model is both labor and time intensive process. In this paper, we propose a semi-automatic framework to reconstruct accurate patient-specific anatomic model efficiently. We do not only transfer a template anatomic model to match the patient's external body shape but also accurately match the internal anatomic structure represented by medical imaging by solving a Laplacian deformation field. Our framework is able to handle different medical imaging techniques, including CT and MRI. We demonstrate several results to show that our transferred results are not only visually plausible but also medical accurate around the affected regions, and also evaluate our accurate anatomic reconstruction using two different metrics to quantitatively measure the accuracy.