生物醫學資訊在臨床泌尿學研究之建構及應用

TMUIR > College of Medical Science and Technology > > Dissertations/Theses > Item 987654321/53961

Please use this identifier to cite or link to this item: http://libir.tmu.edu.tw/handle/987654321/53961

Title:	生物醫學資訊在臨床泌尿學研究之建構及應用 Development and Application of Biomedical Informatics in Clinical Urology Studies
Authors:	曹智惟 Tsao, Chih-Wei
Keywords:	類神經網路;莢膜侵犯;前列腺腫瘤;磁振造影;葛里森評分;前列腺特異抗原;勃起功能障礙;牙周疾病;Artificial neural network;Capsule invasion;Prostate neoplasm;MRI;Gleason score;PSA;Erectile dysfunction;Periodontal disorder
Date:	2014-07-17
Issue Date:	2018-11-15 13:03:19 (UTC+8)
Abstract:	論文摘要論文名稱：生物醫學資訊在臨床泌尿學研究之建構及應用臺北醫學大學醫學資訊研究所研究生姓名：曹智惟畢業時間： 102 學年度第 2 學期指導教授：徐建業臺北醫學大學醫學資訊研究所教授目的：本篇論文的目的系運用生物資訊來建構一類神經網路模型，用於預測前列腺癌患者預備接受根除性前列腺切除手術之病理期別，以探討前列腺包膜被腫瘤細胞侵犯之可能性：另以生物資訊方法，來輔助臨床骨盆腔磁振造影診斷前列腺癌病理期別之準確性；同時我們也以處理龐大資料庫之原理，將臺灣健保資料庫作與男性勃起功能障礙相關因子之分析研究。方法：此回溯性研究包含了299 位接受恥骨後前列腺根除手術或機器手臂輔助式腹腔鏡前列腺根除手術之病患，利用患者手術前的臨床資料 (如年齡、肥胖度、血液前列腺特異抗原、肛門指診結果、經直腸前列腺超音波檢查結果及前列腺切片組織病理等等) 來建構一類神經網路預測模型用於預測其前列腺包膜被腫瘤細胞侵犯之機率：涵蓋94位前列腺癌接受根除手術之病患，利用血液前列腺特異抗原及葛里森評分，來增進骨盆腔磁振造影診斷術後前列腺癌病理期別之準確性；最後，我們利用台灣健保資料庫，隨機撈取5,105位勃起功能障礙之患者及10,210位對照男性，分析慢性牙周病與臨床勃起功能障礙之相關聯性。結果：類神經網路研究中，所建構之預測模型包含了 7 個預測因子 (輸入因子)。測試結果發現其 ROC 曲線下之面積 (AUC) 是0.795，大於羅吉斯迴歸模型之 AUC (0.746)且有顯著統計差異. 亦遠大於平時應用於臨床預測模型 (Partin tables) 所得之AUC (0.695)；臨床研究發現，在臨床期別二期的前列腺癌患者輔佐以葛里森評分，在臨床期別三期的病患以血液前列腺特異抗原，可大大提昇骨盆腔磁振造影來分析術後前列腺癌病理期別之準確性；在健保資料庫資料研究中，慢性牙周病與勃起功能障礙具有統計意義之關聯性，此關係在小於30歲及大於59歲的男性族群更為顯著，但在接受相關治療之患者則有較低之影響。結論：生物資訊包含建構之類神經網路預測模式針對前列腺癌患者提供較準確之病理期別的預測，及增進現今以骨盆腔磁振造影診斷術後前列腺癌病理期別之準確性，同時更能處理較大資料庫以利於臨床的研究分析及日後醫療方針的確立。 Abstract Title of Thesis：Development and Application of Biomedical Informatics in Clinical Urology Studies Author：Chih-Wei Tsao Thesis advised by ：Chien-Yeh Hsu Taipei Medical University, Graduate Institute of Biomedical Informatics Objective: We developed an artificial neural network (ANN) model to predict prostate cancer pathological staging more effectively than logistic regression (LR) and practice-based Partin tables; To improve the accuracy of MRI diagnosis of the final pathological stage in the current diagnosis process; ED and CPD have been reported to have similar risk factors and associated systemic conditions related to the impairment of endothelial function, the relationship between CPD and ED is unclear. Materials and methods: The study evaluated 299 patients undergoing radical prostatectomy. The results were intended to predict the pathological stage of prostate cancer (T2 or T3). The predictive ability of the ANN was compared with that of LR and validation of the 2007 Partin Tables estimated by the areas under the receiving operating characteristic curve (AUCs); 2nd study included 94 PCa patients receiving eMRI testing during pre-surgical evaluation and undergoing radical prostatectomy. These stage assessments as evaluated through the use of MRI were compared with the final specimen pathological stage after the patients underwent radical prostatectomy; Using a nationwide population-based dataset, we examined the association between ED and CPD, and assessed the effect of dental extraction (DE) on ED prevalence in different aged CPD populations in Taiwan. Results: The ANN overall outperformed LR overall significantly (0.795 ± 0.023 versus 0.746 ± 0.025, p = 0.016). Validation of the current Partin Tables in the subjects of our study was assessed, and the predictive capacity of the AUC for OCD was 0.695; In clinical stage T2 prostate cancer patients, the Gleason score significantly improved the discriminative ability of eMRI to successfully predict prostate cancer at the OCD stage. In cases of clinical stage T3 prostate cancer, accurate determination of PSA levels significantly improved eMRI predictive ability to assess ECE or SVI staging; We identified 5,105 patients with ED and randomly selected 10,210 patients as controls. 2,617 (17.09%) were diagnosed with CPD: 1,196 (23.43%) in the ED group and 1,421 (13.92%) in the control group. Patients with ED were more likely to have been diagnosed with prior CPD than controls (OR = 1.79, 95% CI = 1.64-1.96, p < 0.001). The association was much stronger in the populations aged less than 30 years (OR = 2.13, 95% CI = 1.23-3.70, p < 0.001) and more than 59 years (OR = 2.27, 95% CI = 1.99-2.59, p < 0.001) Conclusions: The bioinformatics including that ANN development for pathologic stage prediction, improvement of eMRI diagnosis accuracy and big data analysis for clinical treatment are practicable and powerful in the urology study field.
Description:	博士指導教授-徐建業委員-查岱龍委員-于大雄委員-楊騰芳委員-張祐誠
Data Type:	thesis
Appears in Collections:	[ ] Dissertations/Theses

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	80	View/Open

View Licence

Loading...