The rapid development of the picture archiving and communication system has significantly changed the method of communication and management in medical information science. However, as the extent of the hospital operation increases, large amounts of digital images are transferred to the network mainly for system efficiency. Two servers made up the server cluster that was constructed for this investigation. A total of 1 to 16 workstations were used to move the clinical condition at the same time, including computed radiography (CR), computed tomography (CT), and magnetic resonance (MR) images.
The average transfer rate (ATR) was analyzed between the cluster and the non-classifier server. In the download scenario, AGR, CT, and MRI images increased by 44.3%, 56.6% and 100.9%, respectively, when using server clusters, while ATR-er uploaded the landscape increased by 23.0%, 39.2% and 24.9%. In the mixing landscape, transmission performance increased by 45.2% when you used eight computer units. The server cluster can improve transmission efficiency by maintaining high reliability and continuous accessibility in a healthy environment, making it a strong approach for PACS Optimization.
During the decade, a picture archiving and communication system (PACS) has proven to be an effective platform to increase productivity and patient satisfaction in health facilities. Web-based solutions allow hospitals and spread and administer campuses in all PACS architectures.
It has been accepted as a primary alternative for PACSS on a large scale. However, viewing the central collection and requesting images causes the PACS server to become overloaded, and during peak hours, network traffic increases significantly. Image distribution time is largely longer and cannot be accepted by doctors. Therefore, a strong online PAC is necessary to enhance system efficiency and maintain reliability, making it essential to focus on PACS Optimization.
An online PACS picture archiving time is sensitive. Fast delivery of medical images is one of the most important features of maintaining the workflow for radiology. There are two types of ways to increase the speed of imaging: By using a sharp network connection and distributing a server with high performance. For network connections, it is necessary to transfer an image from PACS Archiver to a view Gigabit Ethernet or higher workstation. In addition, the image model and the PACS server should be connected to a network with a minimum speed of 100 Mbit/s. An increase in the amount of RAM and the number of CPUs can significantly reduce the transfer time.
In addition to improving system efficiency, it is also important to maintain the reliability and validity of PACS. PACS Optimization plays a vital role here, as PACS picture archiving is a simple server error point (SPOF); any obstacle to the services can put the data integrity at risk and prevent daily clinical operation. Therefore, fault tolerance should be measured to maximize the system uptime for the end users.
A simple method is to use the parts, which can achieve an accessibility rate of 99%. An alternative approach is to distribute the cluster server, which can reach an availability of 99.99%. A triple modular redundant (TMR) architecture with three Solaris Unix servers has been used to produce a strong PAC with high operational reliability, further supporting effective PACS Optimization.
A continuous availability of 99,999% was achieved under a number of clinical conditions. Although it is okay to add fruitless equipment, it is expensive, and the image does not improve the speed of the transfer. Other studies have shown backup and extraction of clinical images using Data Grid Architecture and an application service provider (ASP) model, both of which contribute to better PACS Optimization.
When you consider the costs and efficiency, the commercial hardware outside the chest was used to create a server cluster in this study, which has an active/active configuration, which runs replication services for network balance (NLB), distributed file systems (DFS), and Structured Query Language (SQL) servers. As a measure of improvement in efficiency, we compared the average transmission speed between different configurations of PACS server with different scenarios, including image, download, and server error. The purpose of this study was to evaluate the possibility of using the COTS server cluster as a sophisticated PACS collection and controller server to increase the system efficiency and reliability of a health care environment, ultimately supporting better PACS Optimization.
2. Materials and methods
2.1. Server configuration
The hardware configuration of the COTS PACS server included AMD Athlon 64 X2 4200+ Dual-Core Processor (2.2 GHz) and 8 GB DDR RAM (533 MHz), which played a crucial role in supporting PACS Optimization for better performance and reliability.
Two network interface cards were installed; One (Broadcom 5755 Gigabit Ethernet) handled Internet Works Communication, and the other (Intelpro/100 Management Adapter 82559) took network traffic at Ethernet Ridge. Four 250 GB hard drives, 750 GB RAID 5, were connected to the server as a short-term storage device, creating a total usable storage of 1,250 GB. The server is operated with a service package 1 installed on Windows Server 2008 R2. Victory Software version 1416RC2 was used as an image web server (IWS), ensuring system stability and supporting better PACS Optimization.
The road works include image sequencing, image collection, image format conversion, DICOM network access, DICOM image filtration and web view, and image compression. Microsoft SQL Server 2008 was also employed by IWS as it is required for details and stored details, and the collection of the patient’s information, exam study, chain number, photo mode, and upcoming image model. These integrated processes play a vital role in enhancing PACS Optimization for smoother workflow and data management.
2.2. Customer configuration
An ETLON X2 4200+ CPU and individual computers with 1 GB of RAM were distributed as client workstations. A Broadcom 5755 Gigabit Ethernet card and a 1A60 GB Hard disk were installed. Microsoft Windows XP Professional was established as an operating system, and Internet Explorer 6.0 was used as a default browser to access IWS through the DICOM query and rebuild the protocol. When the client computer first logged into IWS, an ActiveX component was loaded for the query/reconstruction, patient management, and image enabling. These configurations were designed to support PACS Optimization by ensuring smoother data access and system performance.
2.3. Server cluster- and non-cluster mode
Two types of server architecture were produced: no cluster and cluster mode. For noncluster mode, a server computer was used to fulfill the requests for image uploads and to download from the model and client work drive, while for cluster mode, two identical server computers were combined to form an active/active server cluster as a PACS Archive server, enhancing overall PACS Optimization in performance and reliability.
2.5.2. Download Landscape
Images were downloaded from PACS picture archiving servers using varying numbers of client PCs, ranging from one to sixteen. IWS was previously logged into the client PC. Finally, the obtained image package was restored and shown on a screen continuously. ATR was measured and compared between the cluster and the non-cluster configuration.
2.5.3. Mixing and disaster landscape
In the mixing landscape, two client computers were classified as devices, in which one computer demonstrated the download process for the image and the other demonstrated the image upload process. The total file size of 160 MB of CT, CT, and MR images was sent. In different numbers from 1 to 8, their functions were used to perform. The purpose of the mixing landscape was to simulate a real health environment and evaluate PACS Optimization under practical conditions.
In addition, the disaster landscape was mimicked with 8 units that carried out the mixing landscape. The connection to a server was removed to follow an error situation after 10, 20, and 40 seconds. The integrity of the transaction of images was analyzed, as well as the ATR. All measurements were repeated in three copies, and average and standard deviations were estimated to assess PACS Optimization in handling error conditions effectively.
2.6. Aid
In addition to the software and services above, clients were required to reduce human errors during the client computer and automatic operation. The NTPCLOCK version 2.1 that supports Network Time Protocol (NTP) is installed on the client and server computers. This software sends a periodic request to the server located in Taiwan, National Standard Time and Frequency Laboratory, to obtain an accuracy of 30 ms, to adjust the clock in the operating system. Automatic version 1.3 was installed to automatically check client computers. Mouse movement, mouse clicks, and keyboard attacks were already recorded for each landscape. Therefore, client computers can perform accurate commands and procedures at the right time, supporting overall PACS Optimization in clinical workflows.
The car part is a time-consuming process. Any extra wait can be unacceptable to doctors, especially when the transfer rate is less than 500 kb/s. Previously, to solve the problem of slow access to medical images during high times, several independent image collection servers and IWs were used to spread the workload. In this aspect, the proposed server cluster architecture in this study can significantly reduce the transfer time using the NLB service to distribute CPU loading, ensuring better PACS Optimization for medical environments.
Parallel treatment increases the efficiency of downloading images by 100.9%. In the clinical state, downloading images and uploading are often done together. The proposed server cluster still has the capacity to increase the transfer rate for MRI images by about 45.2%. These results suggest that the COTS Server Cluster is a viable option for PACS Optimization in online environments.
Small and intermediate hospitals require a minimum of 50 GB/day upload capacity, while large health stations require a minimum of 100 GB/day capacity. Our findings for the upload scenario demonstrate that the suggested server cluster can provide an upload capacity of at least 357 GB per day. In order to satisfy the needs for multiple transfers in contemporary hospitals, such as images produced by multidetector computed tomography (MDCT), server cluster architecture can be employed. However, because compression servers put additional strain on the CPU, they reduce upload capacity. By efficiently converting charge to both nodes, the server cluster architecture can enhance upload performance and contribute to effective PACS Optimization.
It is theoretically possible to expand the number of nodes in the server cluster. However, in order to maintain accessibility and reliability, uploaded images must be compared to each node and duplicate, which reduces the performance of the system. Therefore, the functions of the health care system must adapt to the server configuration by assessing their scales to achieve optimal cost-effectiveness conditions. In addition, the Peer-To-Peer (P2P) protocol can be used to replace the traditional database storage protocol. DFS NAMSPACE technology can be used to group shared folders on different nodes, helping to minimize unnecessary replication and support better PACS Optimization.
In the proposed server cluster, all the nodes are active. Each participating node requires a surveillance script, which repeatedly examines the condition of the system and calls itself from the cluster to call or remove the NLB tool as needed. In the occurrence of an error, the PACS Optimization process ensures that the NLB service automatically detects errors and redirects data flow. The remaining active node carries out additional treatment operations. Therefore, picture archiving system services have no obstacles, and clients are not aware of your computer user error.
When we reduce the periodic test holes on heartbeat messages, communication between server nodes increases, resulting in minimal failure. However, the NLB service requires CPU and network resources to investigate the upcoming packages and make a correct decision on load balance. If the check gap is very low, the message packages can capture all system resources, leading to a reduction in PACS performance. After adaptation, a periodic examination time of five seconds was prescribed, ensuring that data can be redirected to a healthy node in ten seconds, which supports better PACS Optimization.
In the future, a multivode cluster server can be constructed with several active nodes, a primary passive node, and an alternative passive node. The primary passive node is used when one of the active nodes fails or needs to roll up. The alternative passive node is used only when an error event occurs and the primary inactive node is inaccessible. This design can maintain a minimum cost of using COTS hardware and maximize the efficiency, reliability, and availability of PAC, making it a practical approach for effective PACS Optimization.
5. Conclusion
Picture archiving system PACS is a simple server error point; Any failure can put patient care and hospital operations at risk. Using the proposed COTS server cluster as an online PACS improves the download of the image and uploads efficiency and guarantees continuous accessibility in a variety of medical image collections and recovery scenarios. This study suggested the actual transmission rate of the COTS server in a clinical PACS picture archiving environment, which can be used as a reference for the creation of an efficient, online PAC cluster of reliable, scalable active/active COTS servers, ultimately supporting effective PACS Optimization.