Humans are exposed to bacterial pathogens all the time, which are present not only in infected patients, but also in soil, water, wildlife, domestic animals, and food. With the increase of bacterial infection and drug resistance developed over time of the most infectious diseases caused by pathogenic bacteria, we are facing a major public health problem in the world. It is reported that more than 2.8 million people die from bacterial infections every year (Karbelkar and Furst, 2020). A complete and detailed description of the structure and morphology of different bacterial pathogens plays an important role in many biomedical studies related to the diagnosis and treatment of bacterial infections. The determination of bacterial morphology and spectral information can provide structural and biochemical characteristics of the bacteria, which can be used to distinguish pathogenic from harmless symbiotic bacteria. However, due to resolution limitations, individual bacteria cannot be studied in detail using traditional microscopy techniques (Lucidi et al., 2014). For example, the lateral resolution that can be achieved by using conventional microscopy techniques based on laser excitation is limited by the phenomenon of light diffraction, reaching only half the wavelength of the excitation light (Gahlmann and Moerner, 2014). Therefore, higher resolution is needed to describe the basic structure of bacteria at the subcellular level. Optical nanotechnology based on superresolved fluorescence overcomes the diffraction barrier and can provide typical resolution of 30–100 nm (Rust et al., 2006). However, their lack of chemical sensitivity and reliance on tailor-made fluorescent probes limit their wide applicability. For biological samples, the contrast agents used in fluorescence technology can seriously affect the morphology, metabolism and movement of organisms, and may lead to cytotoxicity and phototoxicity (Cosentino et al., 2019). These limitations and concerns have spurred interest in actively developing label-free optical imaging technologies.

Terahertz (THz) radiation refers to the frequency band of 0.1–10 THz in the electromagnetic spectrum, corresponding to the wavelength of 3 mm–30 μm (Zhang et al., 2018). Due to the high sensitivity of THz spectral imaging technology to biomacromolecule, for example, proteins and nucleic acids exhibit unique vibration, rotation and conformational responses to THz radiation (Mittleman, 2018). Therefore, rich physical and chemical information inside biological samples can be provided in a label-free, non-invasive and non-ionizing manner (Chen et al., 2020; Zhou et al., 2021). At present, significant progress has been made in imaging biological samples using THz imaging technology (Wan et al., 2020; Chen et al., 2021; Ke et al., 2021). For example, THz imaging has successfully distinguished burnt skin tissue from normal skin tissue, and cancerous brain tissue from normal brain tissue (Bowman et al., 2017; Geng et al., 2019; Wu et al., 2019). Furthermore, the dehydration state of a single plant cell (~100 μm) was monitored and characterized by the THz near-field imaging technology based on the microstructure photo-conductive antenna (PCA; Li et al., 2020, 2022). Unfortunately, due to the inherent structural limitations of the PCA antenna micro-probes, it is difficult to further improve the resolution to the sub-micron or even nanometer level for the detection of a single organelle (~1 μm). At present, THz scattering-type scanning near-field optical microscope (THz s-SNOM) has the highest resolution in the field of THz imaging and is most likely to achieve a fine description of bacteria without damage. THz s-SNOM is an innovative ultra-high resolution optical microscope and spectral system based on atomic force microscopy (AFM) technology. The THz s-SNOM system relies on light scattering from metallized probes of nanoscale AFM to achieve near-field optical imaging and spectroscopy at subwavelength spatial resolution, independent of the wavelength of light. This method has been widely used to characterize nanoscale semiconductor devices (Yao et al., 2019), and free carrier concentration (Huber et al., 2008). A recent report has also shown that a single protein (~10 nm) can be imaged using a graphene-based THz imaging technique based on the scattering principle of nano-metal tips (Yang et al., 2021). Therefore, there is great potential to study single bacteria using the THz s-SNOM system.

This paper develops a new method for studying single bacteria using THz s-SNOM. Through the coupling effect of nanoscale radius of platinum probe and gold base, greatly enhanced THz near-field signal of biological samples, combined with the image processing method to remove the background noise and performing segmentation to obtain the subject target, has successfully demonstrated the morphological structure of individual bacteria.

Figure 2 is the result of AFM imaging of single E. coli and S. aureus, where Figures 2A,C are topographic maps, which clearly show the morphological information of single bacterium. Figures 2B,D are amplitude images, which describe the surface topography of the bacteria. A rod form of the bacterium can be clearly observed in Figures 2A,B, it is bluntly rounded at both ends and has flagella at the tail, which is clear that E. coli has been detected. The single E. coli is about 1.7 μm long and 0.5 μm wide. Meanwhile, it becomes higher from the edge to the middle, and the highest point is about 280 nm. Figures 2C,D depict a spherical bacterium, which is approximately 1.8 μm in diameter and 300 nm at its highest. It is clearly S. aureus. The results in this paper are in good agreement with the previous results of bacterial morphology of E. coli and S. aureus, and the specific results can be referred to the book (Ghalem, 2015) and reference (Klainer and Perkins, 1970).

Figures 3, 4 describe the two different bacteria by using THz near-field imaging, where (A–D) and (E–H) are THz amplitude and phase images at different harmonic demodulation (from 2 Ω to 5 Ω), respectively. Refer to Figure 2, THz near-field amplitude and phase imaging of the bacterium are consistent with topographic maps, and both can accurately describe the shape and size of the individual bacterium. From the results shown in Figures 3, 4, a clear regional division between the biological samples and the substrate can be easily identified based on the THz near field images. It is generally believed that the resolution of s-SNOM is closely related to the harmonic frequency of signal extraction (Mastel et al., 2018; Dai et al., 2019). To test the detailed resolution of biological samples at different harmonics, we select a line on the amplitude image that crosses the boundary between the sample and the substrate (the black line in Figure 4A) to derive the resolution, along which signal values are obtained as a function of position. These are shown in Figure 5, for the harmonics demodulation frequencies from 2 Ω to 5 Ω. In Figure 5, according to the 10–90% rule, the spatial resolution of the near-field image demodulated at the fifth harmonic of the tip modulation frequency is obtained: corresponding to 2 Ω to 5 Ω, the resolution is 0.63 μm, 0.49 μm, 0.24 μm and 0.11 μm, respectively. Our experimental results are consistent with reported findings previously, that is, the resolution of higher harmonics is better than that of lower harmonics (Fikri et al., 2004). On the other hand, as the order increases, the demodulated phase image showing the outline of the bacteria becomes clearer, which can be seen in Figures 3E–H, 4E–H. In our example, the resolution obtained at 5 Ω is the best of all harmonics. Obviously, if we need to obtain sample information at the highest possible resolution, the demodulation signal at 5 Ω would be a good choice. In addition, we found that the intensity of the near-field signal decreased sharply as the demodulation frequency increased from 2 Ω to 3 Ω, which can be attributed to the high frequency attenuation suffered by the demodulation signal during the lock-in process (Wurtz et al., 1999; Mastel et al., 2018). Except for the almost invisible differences, all four amplitude images have good qualities and describe morphology of bacteria clearly, which also indicates that signal strength alone does not always make better image quality, and the intensity contrast between biological samples and substrate is also a key factor in evaluating the quality of near-field signals (Dai et al., 2021).

In this study, we have demonstrated that THz spectral images can describe the morphology of bacteria as well as the AFM images. More importantly, THz waves can penetrate into the bacteria, and demonstrate the distribution of substances inside the bacteria according to the differences in the spatial variation of composition and dielectric properties interior of the bacteria. The THz spectral images of bacterial morphology have been denoised and segmented by the image processing method presented in Section 3.1, as shown in Figure 6. In addition to the morphology of bacteria, the internal structure of bacteria can also be observed. In particular, pseudonuclei made up of loosely coiled circular DNA molecules can be observed in E. coli (Figure 6A), and the structure of E. coli can be referred to the literature (Verma et al., 2019). By counting the number of pixels in the bacterial regions, the calculated areas of E. coli and S. aureus were about 0.747 μm² and 1.296 μm², respectively. As shown in Figure 6C, the THz near-field signals on bacterial bodies at different positions have been extracted from the THz spectra image in Figures 6A,B. The near-field signals of E. coli and S. aureus are obviously different. The near field signals of E. coli are more stable, while that of S. aureus fluctuate with increasing frequency. The two types of bacteria can also be easily identified by differences in their THz spectral information.

In this paper, an advanced THz super-resolution imaging method is proposed. Specifically, the THz near-field signal is enhanced by the coupling effect between the probe and the gold film substrate, and the THz spectral image of a single bacterium is successfully obtained by using the THz s-SNOM system. Further, combining with image processing method to remove background noise and obtain biological image segmentation, we were able to clearly show the internal nucleoid structure of E. coli. The method presented in this paper has been applied to the monomer identification of E. coli and S. aureus. Super-resolution of THz morphological imaging and THz spectral differences can be used to distinguish the two monomers easily. Hence, this method paves a new avenue for the label-free study of individual bacteria.

The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author.

All authors listed have made a substantial, direct, and intellectual contribution to the work, and approved it for publication.

This work is supported by the Major Instrumentation Development Program of the Chinese Academy of Sciences of China (ZDKYYQ20220008), the National Natural Science Foundation of China (NSFC) under grants 61875051 and 62271439, and the Natural Science Foundation of Zhejiang Province (ZJNSF) under grant LR21F010002.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.