Operational departmentwide picture archiving communication system analysis using discrete event-driven block-oriented network simulation

Abstract

The accurate prediction of image throughput is a critical issue in planning for and acquisition of any successful picture archiving and communication system (PACS). Simulation plays an important role in this effort. The PACS image management chain is decomposed into eight subsystems. These subsystems include network transfers over three different networks and five software programs and/or queueing structures. This decomposition is used to create a simulation model that was effectuated using commercially available block-oriented network simulation software. From the PACS database, the traffic generation patterns of the imaging modality devices are used to drive the simulation. The simulation models the image file flow through the PACS for a 24-hour period. The behavior of the simulated traffic generators agreed well with the values derived from the PACS database. The mean delay for the simulated PACS is found to be 225 +/- 59 seconds. The delay time was found to vary during the simulated 24-hour cycle in a consistent manner with observations. This simulation provides estimates on what a radiological department can expect from a PACS in terms of throughput, utilization, and delay. The block-oriented network simulator (BONeS, Comdisco Systems Inc, Foster City, CA) simulation model of the modeled PACS is highly accurate. The models for the imaging modality traffic sources are validated with a high degree of accuracy. The simulation model allows for the study of what happens to the delay time under various loads. In this simulation, the reformatting process was determined to be the bottleneck causing a large increase in delay time under heavy loads.

PACSPulse: A Web-based DICOM Network Traffic Monitor and Analysis Tool

“…Good performance has been a fundamental criterion since the inception of picture archiving and communication systems (PACS) over a decade ago (1)(2)(3)(4)(5)(6)(7)(8)(9). As computers and networks have increased in capacity and performance, so has the ability to generate ever-larger volumes of data and higher-resolution images in radiology.…”