- 學位論文
智慧型眼科影像資訊系統之設計與製作
Design and Implementation of an Intelligent PACS for Ophthalmology
摘要
影像歸檔與傳輸系統(picture archiving and communication system, PACS)建置的訴求，是提供醫療人員便捷的診斷資料，但美中不足的是，在中大型醫院之就診高峰時段中，巨大的資料流會破壞系統建置初衷的美意；因此在現有的影像歸檔與傳輸系統架構上，融入智慧系統調控的概念是本研究之根本立意。 本研究透過系統資源調控工作站的運作，實現一個虛擬的系統管理者，監視系統運作效能，在PACS伺服器面臨負載過重時，進行系統資源調整，以維持伺服器穩定的效能表現。研究中依照各系統資源參數對於伺服器工作效能的影響力，定義出整體負載指數，以之量化伺服器效能的變化，作為評估伺服器工作狀態的依據。一般而言，PACS的影像傳輸途徑是經由影像擷取工作站傳送至影像伺服器中，再透過伺服器的分送服務，把影像資料傳遞到指定影像顯示工作站；因此，系統資源調整之負載分散機制的驗證，將在臨床負載分佈趨勢之模擬環境中，分別對上述兩階段的影像傳輸效能表現予以探討。 研究結果顯示，使用負載分散機制啟動閥值為60之高負載啟動情況下，兩階段的傳輸速率與未啟動機制前相較之下，分別提升了62.88%與73.92%；相同於前述實驗，當啟動閥值為15之低負載啟動情況下，兩階段的傳輸速率分別提升了62.39%與80.57%。根據以上的研究結果，吾人認為本研究所設計的反應機制對於臨床上應有所助益。預期它不但可以提升高負載狀況下系統的整體運作效能，並可改善臨床負載所造成的效能下滑趨勢。
並列摘要
The goal of equipping picture archiving and communication system (PACS) in clinics is to efficiently deliver digital patient demographic information and the associated images to physicians. This objective is often jeopardized by huge image data flow burst in the system during the busy hours. An intelligent mechanism is introduced to solve the aforementioned problem. Our approach is to create a virtual system manager monitoring the PACS operational efficiency. Whenever a system’s service loading is over a pre-defined workload index indicating newly generated images exploded, the manager then initiates a system resource re-assignment mechanism. Till the raging data flow dies down, a designed system recovery protocol launches. This intelligent framework was evaluated by a test bed containing a PACS server, imaging systems, and image display stations. The test bed imitates services subjected in a clinical working schedule that pertains to a common workload pattern having loading peaks. The study centered on the PACS server. The evaluations were on image transfer speed recorded in image communications between the server and the imaging systems and the image display stations. For the workload index set at 60, the study results showed that image transfer rates increase 62.88% and 73.92% from the imaging systems and to the image displays stations, respectively when the system resource re-assignment mechanism applied. The image transfer performance was slightly different when the index was set at 15. The rate increments were 62.39% and 80.57% corresponding to data received from the imaging systems and data sent to the image displays stations. These conclude that the system resource re-assignment mechanism is capable of reducing heavy workload, consequently remains the PACS service efficiency.