- 學位論文
利用Hadoop平台建立雲端醫學影像資料共享之研究
Study on Using Hadoop Platform for Sharing Cloudy Medical Image
摘要
隨著科技的進步，現在許多產生醫學影像的醫療器材都已經數位化。而數位化的醫學影像是使用醫療數位影像傳輸協定(DICOM)，並透過醫學影像儲存與傳輸系統(PACS)來做儲存、傳輸與管理，可以解決膠片儲存、人力資源的問題，更提供影像調閱的即時性與方便。 雖然醫學影像的數位化帶來了許多好處，但產生的醫學影像愈來愈多，如何有效率的處理大量的資料變成了一個問題。同時高科技醫療儀器檢查被視為是造成醫療費用上漲的重要因素。而因為不同醫院的制度與規範上的差異，在某些情況下會做重複性的檢查，如急診或是病人轉院等，而造成金錢上的浪費，尤其是像核磁共振成像、電腦斷層掃描等檢查費用高昂的項目。此外也會增加診斷決策的時間，而可能會造成治療的延誤。 在本研究中，我們設計了一個基於Hadoop雲端運算平台，作為醫學影像交換共享的系統。雲端運算有著高擴充性、靈活性、低成本且隨時隨地都能存取，使用者只需透過網際網路就能得到雲端運算的服務資源。而Hadoop雲端平台是設計用來部署在低成本的硬體上使用的分散式運算檔案系統，適用於處理像醫學影像這樣的大量資料。在系統中，我們將DICOM醫學影像分成中繼資料與影像資料兩部分做儲存。中繼資料會在中繼資料庫裡永久保存，當只需要搜索與取得病人資訊時，可以有很好的效能，也能減少系統的資料儲存量。而影像資料庫則會儲存半年內產生的醫學影像，提供病人在急診以及轉院需要時使用。若需要查看半年之前的影像，再根據儲存的中繼資料到相對應影像儲存位置的PACS裡取得影像資料。藉由使用我們的系統，達到醫學影像資源的共享，減少重複性的檢查，增加民眾的方便性，以及加快醫師診斷的時間。
並列摘要
Along with the advancement of technology, now much medical equipment which can generate medical images has been digitalized. The digitalized medical images are under the Digital Imaging and Communications in Medicine (DICOM) standard, and it can be stored, transmit and manage through Picture Archiving and Communication System (PACS). Then, film store and human resources problem. In addition, the image can be retrieved more immediately and conveniently. Although the digitalization of medical images can bring us many advantages, more and more medical images have been generated. How efficiently process a large amount of medical images became a problem. At the same time, high-tech medical equipment examination can be regarded as the main cause of health care costs rising. Because of the differences between hospitals’ systems and regulations, sometimes examinations repeat. For example, it may cause money waste when emergency and transferring, especially the Magnetic Resonance Imaging (MRI) and Computed Tomography (CT). Moreover, the time of diagnosis may be increased. In this case, cures may be delayed. In this study, we designed a system based on Hadoop cloud computing platform for exchanging and sharing medical images. Cloud computing has the features of high scalability, flexibility, low cost and it can be accessed at any time and any place. Users can get the services provided by cloud computing through the Internet. Hadoop is designed to be deployed to use distributed computing file system on low-cost hardware, suitable for handling of a large amount of data like medical images. In the system, we have divided DICOM into two parts which include metadata and image data for storing. Our System can permanently store metadata in the metadata database. When users only need to query and retrieve patients’ information, it can achieve high efficiency and reduce the amount of data storing of the system. And the image database can store medical images which generated in the recent half a year. The medical images can be provided to the patients in need when emergency and transferring. If there is a need to view the medical images that generated before the recent half a year, the system will be accorded with metadata to retrieve the corresponding images from PACS. Though using the system, the medical image resources can be shared. Meanwhile, it is more convenient for the patients since the repetitive examinations can be reduced. Furthermore, the time of diagnosis can also be decreased.
參考文獻
延伸閱讀
- 黃元鴻（2011）。基於Hadoop MapReduce與HBase之醫療資訊快速分析平台〔碩士論文，國立臺灣大學〕。華藝線上圖書館。https://doi.org/10.6342/NTU.2011.01863
- 李卓勳（2014）。基於Hadoop叢集之具關聯式規則探勘雲端系統設計與效能之研究〔碩士論文，國立臺中科技大學〕。華藝線上圖書館。https://doi.org/10.6826/NUTC.2014.00114
- 羅子澄 (2012). 運用於Hadoop雲端運算的資料探勘混合編碼演算法 [master's thesis, National Tsing Hua University]. Airiti Library. https://doi.org/10.6843/NTHU.2012.00072
- 王育誠（2015）。巨量資料於Hadoop平台基礎與架構之運用〔碩士論文，國立暨南國際大學〕。華藝線上圖書館。https://www.airitilibrary.com/Article/Detail?DocID=U0020-1106201515450400
- 葉志翔（2018）。Improvement of Computing Performance of Hadoop Map Reduce〔碩士論文，健行科技大學〕。華藝線上圖書館。https://www.airitilibrary.com/Article/Detail?DocID=U0022-1906201818582100