利用三維形態分析方法診斷肺臟腫瘤

The 3D Morphological Analysis Method for Pulmonary Nodule Diagnosis

葉嘉芬(Chia-Fen Yeh) ； 鄭焕勳(Huan-Hsun Cheng) ； 張允中(Yeun-Chung Chang) ； 蘇振隆(Jenn-Lung Su)

《中原學報》 32卷3期 (2004/09) Pp. 363-372

https://doi.org/10.6358/JCYU.200409.0363

摘要

本研究透過影像處理技術來進行肺部CT影像分析及利用三維形態診斷肺臟腫瘤，所使用之技術包含影像增強、半自動邊緣選取、高斯平滑、曲率計算、似圓性和紋理計算等函數來分析及診斷之形態。此外，以重建後之立體影像中之高斯曲率及平均曲率、似球性與平均密度等參數，提供形態之資訊。最後建立類神經網路之模式，對肺臟腫瘤進行鑑別。初步結果顯示，切片厚度為3mm之肺腫瘤測試病例中，鑑別正確率為0.67，敏感度為1和系統之信賴度Kappa值為0.378；切片厚度為5mm之肺腫瘤測試病例中，鑑別正確率為0.833，敏感度為1和系統之信賴度Kappa值為0.515。研究發現，良性腫瘤之鈣化結節、細菌性感染與纖維化之類型，與惡性腫瘤之特徵類似，因此導致系統之正確率降低。

關鍵字

電腦斷層影像 ； 曲率計算 ； 紋理分析 ； 倒傳遞類神經網路 ； 肺臟腫瘤

並列摘要

A pulmonary nodule diagnosis system which based on 3D morphological analysis method for CT image of lung was presented in this paper. Image process technologies, which include image enhancement, semi-automated edge selection, Gaussian smoothing, curvatures calculation, circularity, texture count were used to analysis the shapes of lung nodules. Moreover, mean density, circularity, Gaussian curvature and mean curvature of reconstructed 3-dimension images of chest and nodules provided the observation in morphology. Finally, a Back Propagation Neural Network model was implement for the diagnosis of lung nodules. Preliminary results showed that the accuracy was 0.67, sensitivity was 1 and Kappa is 0.378 for 3mm section thickness lung nodules testing cases. For 5mm section thickness lung nodules testing cases, accuracy was 0.833, specificity was 1 and Kappa was 0.515.From results, we also found that the characteristics of the calcification nodes of benign nodules, bacterial affection and fibered types were similar to malignant nodules, hence, this lower accuracy were obtained in this system.

並列關鍵字

CT image, Curvature calculated ； Texture analysis ； Back Propagation Neural Network ； pulmonary nodule