臺灣博碩士論文加值系統

在本研究中，我們針對前列腺癌的臨床風險分級進行了深入探討，特別是透過結合多個深度學習模型 (Resnet34) 來處理不同的影像數據類型。這種方法允許我們從各種影像數據中提取豐富的特徵，並進行更為精確的風險評估。特別地，我們專注於前列腺及其周圍脂肪的影像數據，並且探討棕色脂肪與前列腺癌的關係，這些區域的詳細分析對於準確判定癌症風險具有重要意義。
透過結合多個Resnet34模型，我們能夠從各個角度分析影像數據，這種多模態融合策略顯著提高了模型在識別前列腺癌的敏感性和特異性。實驗結果表明，使用這種綜合模型框架能夠有效提升診斷性能，尤其是在處理不平衡數據集時，Youden's index達到0.585，證明了其在臨床應用中的可靠性。此外，我們的方法避免了侵入式檢測手段，大幅減少了患者的身體和心理負擔，同時降低了誤診的風險。非侵入性的診斷方法提供了一種更安全、更人性化的醫療選擇，對於提高患者的生活質量具有重要價值。
本研究通過創新地結合多個深度學習模型和精確分析前列腺及其周圍脂肪的影像數據，展示了深度學習技術在前列腺癌風險分級中的強大潛力。

In this study, we conducted an in-depth exploration of clinical risk stratification for prostate cancer, particularly through the integration of multiple deep learning models (Resnet34) to process various types of imaging data. This approach allowed us to extract rich features from various imaging data and perform more precise risk assessments. Specifically, we focused on imaging data of the prostate and its surrounding adipose tissue, exploring the relationship between brown fat and prostate cancer. Detailed analysis of these areas is crucial for accurately determining cancer risk.
By combining multiple Resnet34 models, we were able to analyze imaging data from various perspectives. This multimodal fusion strategy significantly enhanced the model's sensitivity and specificity in identifying prostate cancer. The experimental results demonstrated that using this integrated model framework could effectively improve diagnostic performance, especially in handling imbalanced datasets, achieving a Youden's index of 0.585, which confirms its reliability in clinical applications. Moreover, our approach avoided invasive testing methods, substantially reducing the physical and psychological burden on patients, while also lowering the risk of misdiagnosis. The non-invasive diagnostic method provides a safer, more humane medical option, which is crucial for enhancing patients' quality of life.
This research demonstrates the strong potential of deep learning technologies in prostate cancer risk stratification by innovatively combining multiple deep learning models and accurately analyzing imaging data of the prostate and its surrounding fat.

摘要 ............................................................................................................................................. i
Abstract.......................................................................................................................................ii
誌謝 ...........................................................................................................................................iii
目錄 ........................................................................................................................................... iv
表目錄 ....................................................................................................................................... vi
圖目錄 ......................................................................................................................................vii
第一章 介紹..............................................................................................................................1
第二章 方法..............................................................................................................................9
...2.1 前處理........................................................................................................................9
......2.1.1. One-Hot Encoding...........................................................................................9
......2.1.2. FOV匹配...........................................................................................10
......2.1.3. A區塊前處理...........................................................................................11
......2.1.4. B區塊前處理...........................................................................................13
......2.1.5. 臨床風險分級...........................................................................................15
第三章 實驗設置...........................................................................................16
...3.1. 病患資料集...........................................................................................16
...3.2. 效能指標...........................................................................................17
......3.2.1. 留一驗證...........................................................................................17
......3.2.2. 混淆矩陣即其評估指標....................................................................17
...3.3. 實驗...........................................................................................18
......3.3.1. 訓練設置.................................................................................18
......3.3.2. 實驗設計................................................................................19
第四章 結果............................................................................................................................22
第五章 結論............................................................................................................................25
參考文獻 .................................................................................................................................. 26
Extended Abstract.....................................................................................................................29

[1] International Agency for Research on Cancer. (2023). Data source: GLOBOCAN 2020. Graph production: Global Cancer Observatory. Retrieved from https://gco.iarc.fr/.
[2]International Agency for Research on Cancer. (2020, November). Latest global cancer data: Cancer burden rises to 19.3 million new cases and 10.0 million cancer deaths in 2020. Retrieved from https://www.iarc.who.int ..
[3]Taiwan Cancer Registry. (2023). Top 10 cancers in Taiwan, 2020. Retrieved from https://twcr.tw/wp-content/uploads/2023/04/Top-10-cancers-in-Taiwan-2020.pdf .
[4]De Angelis, R., et al. (2014). Cancer survival in Europe 1999–2007 by country and age: results of EUROCARE-5—a population-based study. Lancet Oncology, 15(1), 23-34.
[5]Siegel, R. L., Miller, K. D., & Jemal, A. (2020). Cancer statistics, 2020. CA: A Cancer Journal for Clinicians, 70(1), 7–30.
[6]Arif, M., Schoots, I. G., Tovar, J. C., Bangma, C. H., Krestin, G. P., Roobol, M. J., Niessen, W., & Veenland, J. F. (2020). Clinically significant prostate cancer detection and segmentation in low-risk patients using a convolutional neural network on multi-parametric MRI. European Radiology, 30, 6582–6592. https://doi.org/10.1007/s00330-020-07008-z
[7]Aldoj, N., Lukas, S., Dewey, M., & Penzkofer, T. (2020). Semi-automatic classification of prostate cancer on multi-parametric MR imaging using a multi-channel 3D convolutional neural network. European Radiology, 30, 1243–1253. https://doi.org/10.1007/s00330-019-06417-z
[8]Zhuang, H., Chatterjee, A., Fan, X., Qi, S., Qian, W., & He, D. (2023). A radiomics based method for prediction of prostate cancer Gleason score using enlarged region of interest. BMC Medical Imaging, 23, 205. https://doi.org/10.1186/s12880-023-01167-3
[9]Zhong, X., Cao, R., Shakeri, S., Scalzo, F., Lee, Y., Enzmann, D. R., Wu, H. H., Raman, S. S., & Sung, K. (2019). Deep transfer learning-based prostate cancer classification using 3 Tesla multi-parametric MRI. Abdominal Radiology, 44, 2030–2039. https://doi.org/10.1007/s00261-018-1824-5
[10]Kuroiwa, K., Uchino, H., Yokomizo, A., & Naito, S. (2009). Impact of reporting rules of biopsy Gleason score for prostate cancer. Journal of Clinical Pathology, 62(3), 260-263.
[11]Karimi, D., Nir, G., Fazli, L., Black, P. C., Goldenberg, L., & Salcudean, S. E. (2020). Deep learning-based Gleason grading of prostate cancer from histopathology images—Role of multiscale decision aggregation and data augmentation. IEEE Journal of Biomedical and Health Informatics, 24(5).
[12]Müller, D., Meyer, P., Rentschler, L., Manz, R., Bäcker, J., Cramer, S., Wengenmayr, C., Märkl, B., Huss, R., Soto-Rey, I., & Raffler, J. (2024). DeepGleason: a System for Automated Gleason Grading of Prostate Cancer using Deep Neural Networks. arXiv preprint arXiv:2403.16678. https://doi.org/10.48550/arXiv.2403.16678
[13]Laurent, V., Toulet, A., Attané, C., Milhas, D., Dauvillier, S., Zaidi, F., Clement, E., Cinato, M., Le Gonidec, S., Guérard, A., Lehuédé, C., Garandeau, D., Nieto, L., Renaud-Gabardos, E., Prats, A.-C., Valet, P., Malavaud, B., & Muller, C. (2019). Periprostatic adipose tissue favors prostate cancer cell invasion in an obesity-dependent manner: Role of oxidative stress. Molecular Cancer Research, 17(3), 821-835. https://doi.org/10.1158/1541-7786.MCR-18-0748
[14]Blue Ridge Radiation Oncology.. Prostate cancer screening. Retrieved from https://blueridgeradonc.com/cancers-we-treat/prostate-cancer/prostate-cancer-screening/
[15]Mehta, P., Antonelli, M., Ahmed, H. U., Emberton, M., Punwani, S., & Ourselin, S. (2021). Computer-aided diagnosis of prostate cancer using multiparametric MRI and clinical features: A patient-level classification framework. Medical Image Analysis, 73, 102153.
[16]Zhang, H., Ji, J., Liu, Z., Lu, H., Qian, C., Wei, C., Chen, S., Lu, W., Wang, C., Xu, H., Xu, Y., Chen, X., He, X., Wang, Z., Zhao, X., Cheng, W., Chen, X., Pang, G., Yu, G., Gu, Y., Jiang, K., Xu, B., Chen, J., Xu, B., Wei, X., Chen, M., Chen, R., Cheng, J., & Wang, F. (2023). Artificial intelligence for the diagnosis of clinically significant prostate cancer based on multimodal data: A multicenter study. BMC Medicine, 21, 270. https://doi.org/10.1186/s12916-023-02964-x.
[17]Chen, Z., Zhang, J., Jin, D., Wei, X., Qiu, F., Wang, X., Zhao, X., Pu, J., Hou, J., Huang, Y., & Huang, C. (2023). A novel clinically significant prostate cancer prediction system with multiparametric MRI and PSA: P.Z.A. score. BMC Cancer, 23, 1138. https://doi.org/10.1186/s12885-023-11306-2.
[18]Chen, G., Dai, X., Zhang, M., Tian, Z., Jin, X., Mei, K., Huang, H., & Wu, Z. (2023). Machine learning-based prediction model and visual interpretation for prostate cancer. BMC Urology, 23, 164. https://doi.org/10.1186/s12894-023-01316-4.
[19]Laurent, V., Toulet, A., Attané, C., Milhas, D., Dauvillier, S., Zaidi, F., Clement, E., Cinato, M., Le Gonidec, S., Guérard, A., Lehuédé, C., Garandeau, D., Nieto, L., Renaud-Gabardos, E., Prats, A.-C., Valet, P., Malavaud, B., & Muller, C. (2019). Periprostatic adipose tissue favors prostate cancer cell invasion in an obesity-dependent manner: Role of oxidative stress. Molecular Cancer Research, 17(3), 821-835. https://doi.org/10.1158/1541-7786.MCR-18-0748.
[20]Ribeiro, R., Monteiro, C., Cunha, V., Oliveira, M. J., Freitas, M., Fraga, A., Príncipe, P., Lobato, C., Lobo, F., Morais, A., Silva, V., Sanches-Magalhães, J., Oliveira, J., Pina, F., Mota-Pinto, A., Lopes, C., & Medeiros, R. (2012). Human periprostatic adipose tissue promotes prostate cancer aggressiveness in vitro. Journal of Experimental & Clinical Cancer Research, 31, 32. http://www.jeccr.com/content/31/1/32.
[21]He, K., Zhang, X., Ren, S., & Sun, J. (2016). Deep Residual Learning for Image Recognition. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 770-778.
[22]Álvarez-Artime, A., García-Soler, B., Sainz, R. M., & Mayo, J. C. (2021). Emerging Roles for Browning of White Adipose Tissue in Prostate Cancer Malignant Behaviour. International Journal of Molecular Sciences, 22(11), 5560. https://doi.org/10.3390/ijms22115560.
[23]Saha, A., Hamilton-Reeves, J., & DiGiovanni, J. (2022). White Adipose Tissue Derived Factors and Prostate Cancer Progression: Mechanisms and Targets for Interventions. Cancer Metastasis Reviews. https://doi.org/10.1007/s10555-022-10056-0.
[24]Chu, K., Bos, S. A., Gill, C. M., Torriani, M., & Bredella, M. A. (2019). Brown adipose tissue and cancer progression. Skeletal Radiology. https://doi.org/10.1007/s00256-019-03322-w.
[25]Chen, Y.-C. I., Cypess, A. M., Chen, Y.-C., Palmer, M., Kolodny, G., Kahn, C. R., & Kwong, K. K. (2013). Measurement of Human Brown Adipose Tissue Volume and Activity Using Anatomic MR Imaging and Functional MR Imaging. The Journal of Nuclear Medicine, 54(9), 1584–1587. https://doi.org/10.2967/jnumed.112.117275.