Laboratory Computer Performance in a Digital Pathology Environment: Outcomes from a Single Institution

Abstract

Background: In an effort to provide improved user experience and system reliability at a moderate cost, our department embarked on targeted upgrades of a total of 87 computers over a period of 3 years. Upgrades came in three forms: (i) replacement of the computer with newer architecture, (ii) replacement of the computer's hard drive with a solid-state drive (SSD), or (iii) replacement of the computer with newer architecture and a SSD.

Methods: We measured the impact of each form of upgrade on a set of pathology-relevant tasks that fell into three categories: standard use, whole-slide navigation, and whole-slide analysis. We used time to completion of a task as the primary variable of interest.

Results: We found that for most tasks, the SSD upgrade had a greater impact than the upgrade in architecture. This effect was especially prominent for whole-slide viewing, likely due to the way in which most whole-slide viewers cached image tiles. However, other tasks, such as whole-slide image analysis, often relied less on disk input or output and were instead more sensitive to the computer architecture.

Conclusions: Based on our experience, we suggest that SSD upgrades are viewed in some settings as a viable alternative to complete computer replacement and recommend that computer replacements in a digital pathology setting are accompanied by an upgrade to SSDs.

Keywords: ImageScope; QuPath; Windows system image; solid-state drive; virtual slides.

Figure 1

Schematic depicting whole-slide viewing model. (a) A graphical depiction of a single frame of the 64-tile test demonstrates the spatial arrangement of tiles within the browser. A tile was replaced when a new tile was loaded, producing a scintillating pattern when observed in video form. (b) The experimental paradigm consisted of a series of 800–1500 trials (denoted by a single-line segment), each separated by a 30-s interval. Each trial consisted of a total of 10,000 tiles presented x at a time (where x is a value between 1 and 256 in 2^N increments). The time it took to render all 10,000 tiles was measured

Figure 2

Whole-slide viewing performance. (a) Rendering performance on the whole-slide viewing model was computed by dividing the number of tiles processed by the total processing time for each trial. Ten trials were conducted for each test condition, and the data are shown according to the standard deviation of the measurements. The number of tiles simultaneously rendered serves as the independent variable. (b) The same test was conducted for 27 computers in the field. The parallel performance ratio was computed as the ratio between 64-tile performance and single-tile performance. The black arrow denotes a drive upgrade computer that underperformed on the single-tile test relative to the rest of the group, and the gray arrow denotes a drive upgrade computer that rendered tiles more quickly in the single-tile test than in parallel

Figure 3

Drive and CPU usage during the whole-slide viewing test. Drive usage (top) was monitored and logged using the Windows Performance Monitor before and during the test. The start of the first trial is denoted by the black arrow. Disk transfer/s represents the number of transfers to and from the hard disk drive or solid-state drive/s. Higher values suggest higher data throughput. CPU usage (bottom) was likewise monitored, and the percentage of processor time represents the amount of CPU usage relative to its capabilities. Gray: sample i3-hard disk drive computer. Black: same computer following drive upgrade. The solid-state drive-based computer was able to complete three trials in the time it took the hard disk drive-based computer to complete one. Note that the hard disk drive-based computer never quite reached 100% CPU usage as disk speed appeared to be the bottleneck

Figure 4

Whole-slide image analysis test. (a) Sample whole-slide image. Scale bar: 2 mm. Inset: higher-resolution view before (right) and following (left) segmentation. (b) CPU usage during the test, blue: sample i3-hard disk drive computer; red: sample Core 2 Duo-solid-state drive computer. (c) Average time each group needed to segment the whole-slide image in its entirety. Twenty iterations were applied for each computer under test