PhenoMATRIX

Performance Metrics

Clinical Evidence

Two bodies of evidence exist, and they do not cover the same product configuration. The first is the FDA clinical validation in the K251511 decision summary: three US laboratories, 5,845 plate images (3,856 body fluid and aerobic cultures at site A, 418 clean-catch urines at site B, 1,571 clean-catch, catheter and paediatric-bag urines at site C), each image classified by the software and then reviewed by the laboratory's microbiologist on the WASPLab screen. Growth detection reached 100% positive percent agreement on blood agar (2,478 of 2,478 plates) with 98.3% negative percent agreement, 99.7% and 99.0% on MacConkey, 100% and 96.4% on CHROMagar Orientation. Morphology labels agreed at 98.1% to 100% by organism group. Rule-based sorting agreed with the microbiologist's classification at 98.3%, 98.1% and 92.6% across the three sites. The weak spots are in the tables rather than the summary: semi-quantification of heavy growth on CHROMagar Orientation agreed in 78.6% of plates (the software under-called 36 of 168), the 10⁴ CFU/mL MacConkey bin at site B fell to 37.9% with a 1 microlitre loop, and the "two isolates" mixed-culture category at site C reached 64.4%. The FDA accepted those figures because every image is reviewed by a trained microbiologist before a result is defined. The second body is peer-reviewed and largely European. Van and colleagues (J Clin Microbiol 2019) enrolled 480 throat swabs from children and read Colorex Strep A agar with the chromogenic detection module: 90.6% sensitivity and 94.0% specificity against a PCR-anchored reference, versus 87.5% and 97.7% for technologist reading. The software found three extra true positives but called more false positives, with a positive predictive value of 79.1% against 90.3%. Baker and colleagues (J Clin Microbiol 2020) tested 676 vaginal-rectal LIM broths on chromID Strepto B. The software missed no culture that technologists called positive, but against the composite reference its sensitivity was 95.5% and its specificity 63.0%, with a positive predictive value of 43.2%. Foschi and colleagues (Diagn Microbiol Infect Dis 2021) analysed 1,068 recto-vaginal swabs prospectively on CHROMagar StrepB and reported the same pattern: 100% sensitivity, 64.5% specificity and a positive predictive value of 32.8%, with viridans streptococci the main false positives. Dauwalder and colleagues (Clin Microbiol Infect 2021) built a tailored urine workflow: a second software version reached 87.5% to 97% agreement with technologists by specimen type and 92% in routine use three months after go-live. Cherkaoui and colleagues at Geneva University Hospitals then published three screening studies, each analysing consecutive routine specimens prospectively with technologists blinded: MRSA on chromogenic media in 17,223 specimens, 99.8% sensitivity and 99.1% specificity for PhenoMATRIX, and 100% and 95.2% for PhenoMATRIX PLUS on a further 1,409 (J Clin Microbiol 2024); Group B Streptococcus on 830 validation plates, 100% sensitivity and 90.2% specificity, negative predictive value 100% and positive predictive value 60.3% (Eur J Clin Microbiol Infect Dis 2025); and a 4,517-specimen rectal screen for carbapenemase producers, ESBL, Acinetobacter and VRE with no false negatives (Front Microbiol 2025). Davidson and colleagues (J Clin Microbiol 2026) compared turnaround before and after implementation at two Canadian laboratories. At the tertiary site the software alone did not shorten reporting: final reports by 18 hours fell from 33.0% to 22.2%. Adding automatic release of negatives (PhenoMATRIX PLUS) raised that to 49.6% and cut mean turnaround by about 1 h 19 min against the baseline. The community laboratory gained about 5 h 18 min, but it also moved its review shift earlier, and the authors say the gain was not due to the software alone. One co-author is a Copan Diagnostics employee. Read together, the pattern is consistent: near-complete sensitivity for growth and for the target morphology, with specificity that varies by medium and by how the laboratory tunes its rules. Most of the European studies used chromogenic screening media (chromID Strepto B, MRSA, ESBL, OXA-48, VRE) and the PLUS auto-release workflow, neither of which is inside the US cleared indication. There is no randomised trial and no patient-outcome study. The evidence grade is prospective on the strength of the Bologna study and the three Geneva studies, which state that routine specimens were analysed prospectively. The AurisMark is one. Approval and evidence would support two, but the rubric asks for a manageable safety record, and the FDA register carried two open Class II recalls at review: one from May 2025 for distribution without a premarket notification, and one from June 2026 in which a WASPLab interface bug could send the software's suggested result to the laboratory system instead of the microbiologist's correction. That second event touches the exact safeguard the clearance rests on. Re-evaluate once the FDA terminates both recalls.

Inside the Auris+ Listing

Regulatory Approvals

Safety Record

The FDA register lists two open Class II recalls. Z-1812-2025, initiated 6 May 2025 and posted 21 May 2025, covered all PhenoMATRIX software versions at 14 US installations across 13 states. The stated reason was that the plate-interpretation software had no premarket approval or premarket notification, so colony counts or morphology from images might not reflect the physical plate; customers were told to check images against the actual agar plate. The 510(k) that resolves the underlying gap was received ten days later, on 16 May 2025, and cleared on 22 January 2026, but the recall record still read "Open, Classified" in September 2026. Z-2854-2026, initiated 16 June 2026 and posted 31 July 2026, is a WASPLab WebApp defect (versions 2.33.1, 2.33.3 and 2.34.2) that under specific conditions prevented the microbiologist's chosen result from overwriting the PhenoMATRIX suggestion, so the suggested result could be transmitted to the laboratory information system instead. The firm released a corrected WebApp on 15 June 2026; one US unit in Arizona was in scope, and the manufacturer filed a matching MAUDE malfunction report (event 9 June 2026). No injury has been reported in either record. The design boundary is the same in every study and in the clearance: the software sorts and suggests, and a trained person defines the result. The 2026 defect matters because it sat on that boundary.

Intended Use & Indications

The FDA indications for use (K251511) read: PhenoMATRIX is a WASPLab optional module intended for the automatic sorting of images of blood-based agar, chocolate agar, MacConkey agar, and CHROMagar Orientation culture media plates according to classification parameters based on Image Analysis Software results and clinical and demographic data. Image Analysis Software performs semi-quantitative and/or qualitative analysis of culture media plates by detecting microbial growth, estimating colony counts and differentiating isolates based on phenotypic colony characteristics. The system determines how the images should be sorted based on image analysis results in addition to patient data according to expert rules defined by the laboratory. All images shall be evaluated by trained personnel for final assessment and result definition. Prescription use only. The FDA decision summary lists the specimen types the software is intended to analyse: urine, upper and lower respiratory tract, genital swabs, vaginal-rectal swabs, body fluids, skin and stool.