A method for filamentous fungal growth and sample preparation aimed at more consistent MALDI-TOF MS spectra despite variations in growth rates and/or incubation times

Deep convolutional neural network: a novel approach for the detection of Aspergillus fungi via stereomicroscopy

Fungi of the genus Aspergillus are ubiquitously distributed in nature, and some cause invasive aspergillosis (IA) infections in immunosuppressed individuals and contamination in agricultural products. Because microscopic observation and molecular detection of Aspergillus species represent the most operator-dependent and time-intensive activities, automated and cost-effective approaches are needed. To address this challenge, a deep convolutional neural network (CNN) was used to investigate the ability to classify various Aspergillus species. Using a dissecting microscopy (DM)/stereomicroscopy platform, colonies on plates were scanned with a 35× objective, generating images of sufficient resolution for classification. A total of 8,995 original colony images from seven Aspergillus species cultured in enrichment medium were gathered and autocut to generate 17,142 image crops as training and test datasets containing the typical representative morphology of conidiophores or colonies of each strain. Encouragingly, the Xception model exhibited a classification accuracy of 99.8% on the training image set. After training, our CNN model achieved a classification accuracy of 99.7% on the test image set. Based on the Xception performance during training and testing, this classification algorithm was further applied to recognize and validate a new set of raw images of these strains, showing a detection accuracy of 98.2%. Thus, our study demonstrated a novel concept for an artificial-intelligence-based and cost-effective detection methodology for Aspergillus organisms, which also has the potential to improve the public's understanding of the fungal kingdom.

Multilaboratory Evaluation of the MALDI-TOF Mass Spectrometry System, MicroIDSys Elite, for the Identification of Medically Important Filamentous Fungi

With the increasing number of fungal infections and immunocompromised patients, rapid and accurate fungal identification is required in clinical microbiology laboratories. We evaluated the applicability of matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) system, MicroIDSys Elite (ASTA Corp., South Korea) for the identification of medically important filamentous fungi. A total of 505 strains comprising 37 genera and 90 species collected from 11 Korean hospitals were sent to the microbiology laboratory of International St. Mary's Hospital. All isolates were tested using MicroIDSys Elite, and data were analyzed using the MoldDB v.1.22 database (ASTA). Correct identification rates were compared with the multigene sequencing results. MicroIDSys Elite correctly identified 86.5% (437/505) and 88.9% (449/505) of all tested isolates at the species and genus level, respectively. About 98.2% of Aspergillus isolates were identified at the species level, including cryptic and rare species of A. calidoustus, A. tamarii, A. lentulus, A. versicolor and A. aculeatus. MicroIDSys Elite identified 75.0% of basidiomycetes, including Schizophyllum commune, and 84.3% of the dermatophytes. It also distinguished Sprothrix globosa at the species level. The mean scores of total isolates corresponding to correct species identification were significantly higher than those obtained for genus-level identification (253.5 ± 50.7 vs. 168.6 ± 30.3, P < 0.001). MicroIDSys Elite showed high accuracy for the identification of filamentous fungi, including cryptic and rare Aspergillus species. It is suitable for use in clinical laboratories as a rapid and efficient tool for clinical mold identification. Further evaluations are recommended for MicroIDSys Elite as a rapid and efficient tool for the identification of medically important filamentous fungi.

Identification of Penicillium species by MALDI-TOF MS analysis of spores collected by dielectrophoresis

In matrix-assisted laser-desorption and ionization mass spectrometry, spectral differences are frequently observed using different growth media on agar plates and/or different growth times in culture, which add undesirable analytical variance. In this article, we explore an approach to the above problem based upon the rationale that, while protein expression in fungal mycelium may well vary under different growth conditions, this might not apply to the same extent in fungal spores. To this end, we have exploited the fact that while mycelium is generally anchored to the fungal-growth substrate, some fungi produce physically-isolated spores which, as such, are amenable to manipulation using dielectrophoresis (the translational motion of charged or uncharged matter caused by polarization effects in a non-uniform electrical field). Such fields can be conveniently generated through the charging of an insulator using the triboelectric effect (the transfer of charge between two objects through friction when they are rubbed together). In this study, polystyrene microbiological inoculating loops were used in combination with nylon-fabric rubbing to harvest fungal spores from five species from within the genus Penicillium, which were grown on agar plates containing two different media over an extended time course. In terms of average Bruker spectral-comparison scores, our method generated higher scores in 80% of cases tested and, in terms of average coefficients of variation, our method generated lower spectral variability in 93% of cases tested. Harvesting of spores using a rapid, inexpensive and simple dielectrophoretic method, therefore, facilitates improved fungal identification for the Penicillium species tested.

MALDI-TOF MS protein fingerprinting of mixed samples

Analytical techniques currently available for the characterization of mixtures of microorganisms are generally based on next-generation sequencing. Motivated to develop practical and less-expensive methods for characterizing such mixtures, we propose, as an alternative or complement, the use of matrix-assisted laser-desorption and ionization time-of-flight mass spectrometry (MALDI-TOF MS), which is capable of high-resolution discrimination between species and even between biotypes within species. Potential approaches employing this technique for such characterization are discussed along with impediments to their successful employment. As a consequence, our rationale has been to capitalize on the powerful algorithms currently available for spectral comparison. Following this rationale, the first priority is to ensure the generation of MALDI-TOF MS spectra from mixtures of microorganisms that contain manageable peak complexities and that can be handled by the existing spectral comparison algorithms, preferably with the option to archive and re-run sample preparations and to pipette replicates of these onto MALDI-TOF MS sample plates. The second priority is to ensure that database entry is comparably facile to sample preparation so that large databases of known microorganism mixture MALDI-TOF MS spectra could be readily prepared for comparison with the spectra of unknown mixtures. In this article, we address the above priorities and generate illustrative MALDI-TOF MS spectra to demonstrate the utility of this approach. In addition, we investigate methods aimed at chemically modulating the peak complexity of the obtained MALDI-TOF MS spectra.