Virtual Gram staining of label-free bacteria using darkfield microscopy and deep learning

Cagatay Isil,Hatice Ceylan Koydemir,Merve Eryilmaz,Kevin de Haan,Nir Pillar,Koray Mentesoglu,Aras Firat Unal,Yair Rivenson,Sukantha Chandrasekaran,Omai B. Garner,Aydogan Ozcan
2024-07-17
Abstract:Gram staining has been one of the most frequently used staining protocols in microbiology for over a century, utilized across various fields, including diagnostics, food safety, and environmental monitoring. Its manual procedures make it vulnerable to staining errors and artifacts due to, e.g., operator inexperience and chemical variations. Here, we introduce virtual Gram staining of label-free bacteria using a trained deep neural network that digitally transforms darkfield images of unstained bacteria into their Gram-stained equivalents matching brightfield image contrast. After a one-time training effort, the virtual Gram staining model processes an axial stack of darkfield microscopy images of label-free bacteria (never seen before) to rapidly generate Gram staining, bypassing several chemical steps involved in the conventional staining process. We demonstrated the success of the virtual Gram staining workflow on label-free bacteria samples containing Escherichia coli and Listeria innocua by quantifying the staining accuracy of the virtual Gram staining model and comparing the chromatic and morphological features of the virtually stained bacteria against their chemically stained counterparts. This virtual bacteria staining framework effectively bypasses the traditional Gram staining protocol and its challenges, including stain standardization, operator errors, and sensitivity to chemical variations.
Quantitative Methods,Machine Learning,Image and Video Processing,Medical Physics,Optics
What problem does this paper attempt to address?
The paper aims to address the issues present in traditional Gram staining techniques, such as the susceptibility to manual errors, inconsistencies due to changes in chemical reagents, and the time-consuming nature of the process. The paper proposes a deep learning-based method that generates virtual Gram staining results from dark-field microscope images, thereby avoiding the cumbersome manual steps and the use of chemical reagents in traditional staining processes. Specifically, the paper introduces a Conditional Generative Adversarial Network (cGAN) model that can convert dark-field microscope images of unlabeled bacteria into Gram-stained images similar to those obtained with traditional bright-field microscopy. This method not only improves the consistency and reproducibility of staining but also reduces errors caused by operator inexperience or variations in chemical reagents. Additionally, this method has been validated on different types of bacterial samples, including Escherichia coli (Gram-negative bacteria) and Listeria monocytogenes (Gram-positive bacteria), demonstrating its effectiveness in staining accuracy. Through various quantitative metrics analyses, the high consistency of the virtual staining method in terms of morphological and color characteristics is showcased, indicating its broad application prospects.