透過您的圖書館登入
IP:3.20.235.88
  • 期刊

核磁共振影像在良性和惡性腦腫瘤之自動偵測與分類上的應用

Application of Magnetic Resonance Imaging in Automatic Detection and Classification of Benign and Malignant Brain Tumors

摘要


核磁共振影像是檢測大腦腫瘤的重要醫學診斷工具,除了提供了詳細的大腦解剖構造外,更可提供較佳的組織對比度。在臨床應用中,核磁共振影像可以提供神經放射科醫師偵測異常細胞的增生或腫瘤型態的相關資訊。本研究主要在開發自動檢測工具並應用在核磁共振影像中不同類別腦腫瘤的分類。雖然目前自動檢測和影像分類在醫學上的應用上仍然是一個具有挑戰性的任務,但在我們的研究中,藉由「支持向量機技術」(Support Vector Machine, SVM) 的運用,我們仍嘗試對磁振影像中的良性與惡性腦腫瘤進行自動分類。方法包括兩個階段:特徵值擷取和分類。在第一階段中,首先使用分割(segmentation) 影像並擷取特徵值 (features)。在第二階段中,將擷提出的影像特徵值輸入支持向量機並進行分類,在此一過程中,我們主要取決於大腦良性和惡性之間腫瘤的特徵類型作為分類的標準。而在最後的結果中我們也發現,對於大腦的核磁共振影像的特徵值分類上,使用線性核心函數的支持向量機具有高達有80%的正確率,可以做為臨床醫師一種極有價值的診斷輔助工具。

並列摘要


Magnetic resonance imaging (MRI) is an important medical diagnosis tool for the detecting brain tumors. In addition to providing a detailed brain anatomy, but also to providing better tissue contrast. MRI can offer information about neuroradiologists' detection of abnormal cell proliferation or tumor patterns. In this study, we have developed an automatic detection tool and applied the classification of different types of brain tumors in MRI. The application of automatic detection and image classification in medicine is still a challenging task at present. In our study, we still try to classify benign and malignant brain tumors in MRI by the use of SVM (Support Vector Machine) technique. The method consists of two stages: feature extraction and image classification. In the first stage, the segmentation image first used and the features are retrieved. In the second stage, the extracted feature values are input into the SVM classifier. In this process, we mainly depend on characteristics types of benign and malignant tumors of brain as the classification criteria. In the final results, we also find that for the feature classification of brain MRI, the support vector machine with linear kernel function has been as high as 80% correct rate, which can be used as a valuable diagnostic tool for clinicians.

延伸閱讀