Highly sensitive antigen test system developed using table-top scanning electron microscope (SEM), with potential for contributing to a wide range of fields and areas, not limited to medical care

I joined Hitachi attracted by the wide range of its business areas

Miwa: Studying in university as an applied physics major, I conducted research on materials for use in organic semiconductor devices and on crystal growth for making inorganic semiconductors. Recognizing the need to observe or measure what I had made, during job hunting I became interested in the kinds of research being carried out in Hitachi, including SEM in particular. Feeling also that I would be able to become engaged in various matters thanks to the broad range of business areas, I joined Hitachi in 2010 upon completing my master’s degree.

Ever since, I have continued with R&D on SEM. My initial research was about understanding the generation of secondary electrons or backscattered electrons as physical phenomena, such as those emitted from a specimen irradiated with an electron beam, and interpreting the images formed by those electrons. After that, I became involved in work on bio applications, and on developing a highly reliable SEM system specifically for semiconductors.

For example, I would measure specimens shared by my colleagues, including with microscope technology other than SEM, such as a scanning probe microscope capable of measuring mechanical and electrical properties of a sample. What motivated me was the desire to compile new insights, and to broaden my own scope of involvement looking several years into the future. In the course of these research activities, I was approached by Tandou-san, in the same department; and that’s what led to the technology we will be talking about here today.

Tandou: Through graduate school, I majored in mechanical engineering, conducting research on a stage mechanism for moving things like semiconductor wafers, capable of positioning with nanometer-order precision. I also researched heat engines out of my own curiosity. For me, as well, it was the breadth of its business scope that led me to choose Hitachi, with the expectation that by tossing various things from here and there into the mix, I would be able to come up with something new and fun to do. I wanted to create many new technologies, products, and services one after another. Among the things I was involved in after joining the company was R&D on plasma etching (nanofabrication) equipment for the semiconductor manufacturing field. During times when I was not working on that project, I would take up subthemes out of my own free interest, and that’s how the technology we are introducing here today came about.

Automation of antigen testing using SEM, in joint research with Hamamatsu University School of Medicine researchers

Tandou: This R&D started out from thinking that if we were to use a SEM system like that for observing semiconductor nanofabrication shapes, viewing objects at the nanometer level, perhaps we would be able to see a virus. I first had this thought back around 2016, when my own child came down with influenza. Then in 2017, when Hitachi High-Tech Corporation came out with a new table-top SEM, I wondered if we could use it for simple and rapid testing. With a conventional antigen test kit, since a virus cannot be detected until it multiplies enough in the patient’s body to come out, there are times when a patient comes to the hospital suffering with a fever but is told to return home and wait for a certain period after the fever first appears. Wouldn’t it be nice if we could detect a virus about as simply as taking a temperature with a thermometer? Knowing that Miwa-san would surely find this research interesting, I approached him about it, and as I expected, he got on board [laughs]. During the times when Miwa-san’s colleagues were not using the equipment he is in charge of, we started experimenting with specimens we had obtained. Since this was in 2017, it was still well before the COVID-19 epidemic, of course.

Miwa: Tandou-san is the innovative type who likes to get involved in many new things. At first I was astonished at his willingness to take action, as he would email researchers he had never met and try one thing after another. I wish I could be more like that. I consider him my teacher when it comes to technology development, and also as a member working together with me. I like that we can have enjoyable discussions on the same level. We have also been able to get over difficult times together, so it has been a good experience.

Tandou: When we began our studies, first we searched diligently through the scientific literature to see if there was anyone else out there already doing the same kind of research. We discovered that Dr. Hideya Kawasaki, Associate Professor at the Hamamatsu University School of Medicine, was engaged in research on using SEM to observe viruses. After contacting him by DM, we visited him to consult about what would be needed for simple and highly accurate testing for viruses using SEM.

Miwa: After consulting with Dr. Kawasaki a number of times, we ended up doing joint research with him. I remember being really impressed when he showed us his SEM images.

In conventional antigen test, the presence of an infectious disease is determined by visually confirming coloration of the test line. If the amount of virus in the specimen is too small, however, the coloration may be hard to detect. This is a reason for the drop in accuracy of such an antigen test. Dr. Kawasaki and his colleagues came up with a method for determining the presence of an infection from SEM images based on backscattered electrons, emitted when SEM imaging is carried out by focusing an electron beam on metal marker particles with antibodies attached, to which viruses bind. This method enables detection of infectious material even when the amount of virus in the specimen is small.

This is basically Dr. Kawasaki’s accomplishment, but let me explain in a bit more detail.

An antigen test kit contains marker particles coated with the antibody corresponding to the target virus; that is, the antibodies are attached to the surface of the markers. When an anterior nasal swab specimen mixed in a buffer solution is applied to the sample well of an antigen test kit, if the specimen contains virus material, the virus binds to the antibodies and is captured along with the marker particles near the test line, forming a visible colored line (positive). If the virus is not present, only the antibody-modified marker particles flow, so that only the control line changes color (negative).

When there is plenty of virus material present in the specimen, large numbers of marker participles to which the virus has bound will be trapped by the surface antibodies at the test line, making it easy to confirm the test line coloration. If the amount of virus is small, however, there will be times when the color change is missed (false negative).

The figure below explains the principle of how metal marker particles in the antigen test kit are detected by SEM. The SEM scans a sample with a focused beam of electrons and produces images of the sample from electrons emitted from its surface (backscattered electrons; photo on left). The brightness of a SEM image is determined by the number of these backscattered electrons. Due to the large quantity of backscattered electrons emitted from metal marker particles, these are seen in the images as bright spots, whereas those coming from the carbon material of the substrate or from vacant space are few, showing up as relatively dark areas (photo on right)

Dr. Kawasaki reported that, compared with existing antigen tests of an actual patient suspected of COVID-19 infection, the new testing technology is from 500 to 1,000 times more sensitive, comparable to the sensitivity of a PCR test.
https://www.mdpi.com/2227-9059/10/2/447

Even so, Dr. Kawasaki and his colleagues had to manually count the particles while viewing the images from the table-top SEM one by one. Moreover, it took time and effort to complete the process from putting specimens in the SEM to imaging. Our role, therefore, was to create software for testing the principle of a system automating the processes from imaging to counting the particles and outputting the assessment results.

High-resolution images are obtained by repeatedly moving the stage with specimen mounted on it in the SEM and imaging the specimen. While so doing, the imaging position in the antigen test kit is also automatically recognized, and a calculation is performed to determine whether there is a statistically significant difference between the number of metal marker particles outside the test line as background and the number of marker particles at the test line. In this way, signals are distinguished from noise, and it can be confirmed whether or not the specimen has flowed properly in the antigen test kit and a virus has been detected.

With a SEM, the shorter the imaging time, the worse will be the image quality, making it difficult to distinguish signals from noise. Considering the test flow, however, we wanted the test time to be as short as practicable. In other words, our challenge was whether we could accurately detect particles from images produced in a shorter time. By testing automatically and with the same algorithm, detection can be made stably, even if short of 100% accuracy, with less variability than occurs when detecting particles by the human eye. We estimate that a decision should be possible within 5 to 15 minutes or less per specimen.

In the 2022 report by Dr. Kawasaki and colleagues, the obtaining of 12 images, 6 of the test line and 6 of background noise, along with their analysis, were carried out manually. In verification testing with the automation software, the time for obtaining images, even with the same protocol, was significantly reduced, and I believe the analysis was also more uniform.

I wanted to continue with R&D, even outside my specialty and even without a budget

Tandou: Since we are of course amateurs when it comes to development of testing technologies in the bio field, such as for viruses, Hitachi had no expectations that our work would lead to a business. Initially, we used up a certain portion of the research expenses within the allotted time; but once the term ended, it was no more research budget and back to self-supported research.

Miwa: When observing a sample by SEM, the sample chamber has to be evacuated, which is not ideal for bio materials containing water, so it is understandable that some people had doubts about the kind of study we were carrying out. If someone else had said the same thing, I probably would have agreed with them.

Tandou: There was already an existing antigen test, and during the COVID-19 epidemic the PCR test came into wide use; so the two of us diligently carried out our main jobs, while quietly going ahead with this R&D theme under the radar. When a press release came out in 2020 about the start of verification testing at the Hamamatsu University School of Medicine, at last we were able to make public the fruits of those efforts, which we had carried on without giving up. Dr. Kawasaki was delighted, and I also was happy to have somehow made it this far.

Miwa: My specialty was optimization of SEM observation conditions and effectively extracting the desired information, but at the time I had no experience with image analysis or with developing techniques for automating procedures. Programming, while thinking about which of the many available technologies to choose, what algorithms would be best to use, and the various trade-offs, took a long time. Today, of course, it should be possible to narrow down the choices to a certain extent by asking ChatGPT. The image analysis techniques we acquired by studying on our own are today helping with observation of other kinds of specimens and with research on automatic analysis.

Tandou: When automation of this detection method becomes more advanced and simpler, my hope is that it can be used as a health checker.

Miwa: Conventional antigen tests are used for a wide range of testing, such as for pesticide residues, shellfish poison, and hoof-and-mouth disease in livestock. By applying SEM technology, I believe it may become possible to perform this testing also as additional testing. Especially now that I have a balanced understanding of specimens, equipment, and applications, I feel capable of providing stable test data across various fields. I won’t know without trying, but while thinking from a business standpoint, I would like to consider uses for this technology.