Assessing the Reading of Urine Cultures Using Digital Images

21 November 2015

The aim of this preliminary study was to determine if reading digital images of urine culture plates from the Automated Plate Assessment System (APAS®) was equivalent to the traditional manual plate reading method.

Automation has brought new methods for assessing clinical and laboratory materials.In microbiology, plate inoculation, incubation and the imaging of cultures have been combined by a number of the large diagnostic companies into full-line automation systems. These systems produce digital images of agar plates for assessment and reporting: a quite different method from the traditional process where a technologist reads cultures directly from the plate.

The new technology provides consistent plate illumination, magnification and the ability to notate results electronically. However, we have found no existing studies comparing plate reading using digital images with the conventional process and we have therefore performed this preliminary study using 200 urine cultures on two commonly used agars.

As a protocol to validate the reading of digital images of agar plates has not yet been published, this study used the College of American Pathologists’ guidelines (1) for assessing digital images of stained slides to determine if on-screen review was suitable for use.

200 urine samples submitted for routine culture (SA Pathology, Adelaide, South Australia) were inoculated onto TSA sheep blood agar and MacConkey with CV agar (Remel, Lenexa, USA). The inoculation volume was 1µL and incubation was for 18h at 35C. Following incubation, plate pairs were independently assessed by three microbiologists and the findings for quantity of growth and colony types were recorded.

The plates were then imaged using an Automated Plate Assessment System (APAS®, LBT Innovations, Adelaide, South Australia) and the digital images stored for later reading.

Two weeks after the initial reporting, the three microbiologists independently read the digital images on-screen and recorded their findings. This delay was built into the protocol to ensure that the microbiologists did not remember specific cases that could bias their reporting.

The manual plate and on-screen digital image results for each of the 1200 plate readings were then compared.

In this preliminary study of 200 cases and 1200 plate comparisons, it was found that the on-screen reading of digital images did not miss important isolates and was more sensitive to some growth features.

To read the full article click here

EP 125, Monday 27 April
European Congress of Clinical Microbiology and Infectious Disease
Copenhagen, 2015