應用資料探勘於肝癌治療傳統栓塞及微球栓塞療效評估之研究

THUIR > College of Engineering > Department of Industrial Engineering and Enterprise Information > Theses and Dissertations > Item 310901/30758

Please use this identifier to cite or link to this item: http://140.128.103.80:8080/handle/310901/30758

Title:	應用資料探勘於肝癌治療傳統栓塞及微球栓塞療效評估之研究
Other Titles:	Applying Data Mining to Efficacy Assessment of Conventional Transcatheter Arterial Chemoembolization and Drug-eluting Micro-bead Embolization for Hepatoma Treatment
Authors:	郭承定 KUO, CHENG-DING
Contributors:	洪堯勳 HON, JAU-SHIN 工業工程與經營資訊學系
Keywords:	肝癌;肝動脈栓塞;栓塞預後;資料採礦;決策樹;邏輯斯迴歸 Liver Cancer;Hepatic Artery Embolization;Embolization Prognosis;Data Mining;Decision Tree;Logistic Regression
Date:	2018
Issue Date:	2018-11-07T04:19:02Z (UTC)
Abstract:	肝癌是亞洲地區重要面對的議題，台灣肝癌罹患人數有逐年下降的趨勢，但仍舊維持癌症之首，不易被察覺的特質導致發現時往往已是中晚末期。目前療法之中以肝癌動脈栓塞較為平和，且適用較廣，一直是不斷研究改良的重點標的，栓塞目前應用於臨床醫學有兩種療法，分別為傳統栓塞及微球栓塞，有部分研究認為兩種療法各有所適用之情況，故本研究期望在現今具備較大量完整資料的背景下採用資料採礦技術對肝癌栓塞兩種療法做更進一步的分類探討。本研究以台中榮總接受肝動脈栓塞之肝癌病患資料為研究對象。以卡方檢定比較療法療效的差異，邏輯斯迴歸計算勝算比比較差異程度，再分別對兩種療法以決策樹C4.5更進一步分類各自適用領域，最後以卡方及邏輯斯迴歸驗證分類結果。研究結果顯示，卡方檢定顯示傳統療法顯著優於微球療法，平衡兩方法比較基礎，僅保留癌症分級A、B級患者再次比較，卡方結果差異不大，以計算勝算比得知傳統療法有效的機率是微球療法的2.57倍。進一步以決策樹C4.5做為分類器，結果發現，傳統療法受限於腫瘤大小，小於等於9.3公分有效，大於則無效，均值檢定有療效的腫瘤大小平均顯著較小，卡方檢定腫瘤小於等於9.3公分顯著有效於大於9.3公分的，勝算比顯示腫瘤小治療有效機率是腫瘤大的8.475倍。微球療法受制於腫瘤數量，多顆無效，單顆腫瘤100-300(µm)有效，卡方檢定單顆腫瘤顯著有效於多顆腫瘤，勝算比發現使用微球療法治療單顆腫瘤成功機率會比多顆腫瘤高3.39倍的。單顆腫瘤中，卡方檢定100-300(µm)顯著有效於300-500(µm)，勝算比結果為100-300(µm)治療成功機率是300-500(µm)的3.412倍。以此研究結果提供肝癌栓塞選擇療法上的一項參考。 Liver cancer is an important issue of cancer death in the Asian region. The number of liver cancer patients in Taiwan has been declining year by year, but it still maintains the top of cancer problem. The symptoms of liver cancer are not easy to detect, so it may be found in intermediate staged or late staged cancer. At present, trans-arterial chemoembolization (TACE) is relatively calm and widely used, and has become a key target for continuous research and improvement. Clinical medicine currently uses two embolization therapies, traditional embolization and drug eluting microsphere embolization. Some studies have shown that each of the two therapies has their own suitable situation, so our study hopes to use the current large amount of complete data, using data mining technology to further classify the two treatment methods for liver cancer embolization.Our study collected data of liver cancer patients at Taichung Veterans General Hospital. Using chi-square test to determine the difference in the efficacy of the comparison therapy. Using logistic regression to calculate the odds ratio. The two treatments were further classified into the applicable fields by decision tree C4.5. Finally, the classification results were verified by chi-square and logistic regression.The results of our study showed that the chi-square test showed that traditional therapy was significantly better than microsphere therapy. To balance the basis of the two methods, the patients who retained the cancer grades A and B were compared again. The chi-square results were not significantly different from the original ones. The effective rate of calculating traditional treatments was 2.57 times than microsphere therapy. Furthermore, using decision tree C4.5 as a classifier, it was found that conventional treatment is limited by tumor size, tumors less than or equal to 9.3 cm were effective, but if they were larger than 9.3 cm, they were not effective. The result of the T test, the tumor size with a curative effect was significantly smaller on average. The chi-square test results showed that the tumor was less than or equal to 9.3 cm was significantly effective at more than 9.3 cm. The odds ratio shows that the effective rate of small tumor treatment was 8.475 times larger than that of tumor. Microsphere therapy was subject to the number of tumors, multiple ineffective, single tumor 100-300 (μm) was effective. The chi-square test for a single tumor was significantly more effective than multiple tumors. The odds ratio found that the use of microsphere therapy, the probability of success in treating a single tumor was 3.39 times higher than that of multiple tumors. In a single tumor, the chi-square test bead size 100-300 (μm) was significantly more effective than size of 300-500 (μm). The odds ratio showed the success rate of bead size100-300 (μm) treatment was 3.412 times than the size of 300-500 (μm). This study provides a reference for the selection of liver cancer embolization therapy.
Appears in Collections:	[Department of Industrial Engineering and Enterprise Information] Theses and Dissertations

Files in This Item:

File	Description	Size	Format
106THU00030016-001.pdf		3205Kb	Adobe PDF	44	View/Open

Loading...