Deep learning decodes antibiotic modes from brightfield bacterial images
Researchers trained a convolutional neural network to identify the mode of action of antibiotics from unlabelled brightfield images of Escherichia coli. The model was exposed to eight distinct MoAs and could predict treatment conditions with near‑perfect accuracy, even when only eight images per condition were used. Importantly, it detected drug exposure at sub‑inhibitory concentrations, a…
Key points
- CNN predicts antibiotic MoA from brightfield images with near‑perfect accuracy using only eight images per condition.
- Detects sub‑inhibitory drug exposure and flags novelty in 5/6 MoAs, aiding discovery of new antibiotic classes.
- Validated on *E. coli* and *K. pneumoniae*, showing cross‑species applicability.
When presented with previously unseen compounds, the network correctly assigned them to the appropriate MoA if that class was represented in the training set. It also flagged novelty in five of six MoAs, enabling the identification of new antibiotic classes. The approach was validated on Klebsiella pneumoniae, demonstrating cross‑species applicability.
This imaging‑based, label‑free method complements existing assays and accelerates the discovery of antibiotics with novel mechanisms, addressing the urgent antimicrobial resistance crisis.
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us.
Comments
via GitHub Discussions