Use of AI for identification and research of orthohantavirus infections

Interview with virologist Tomas Strandin on the use of AI in the quantification of antibodies signaling orthohantavirus infections
Written by Aiforia

Introduction

Orthohantaviruses are globally spread zoonotic pathogens typically causing chronic asymptomatic infection in rodents. However, humans can be infected through contact with rodent urine, saliva, or feces, causing severe health consequences, such as hemorrhagic fever with renal syndrome (HFRS) and hantavirus pulmonary syndrome (HPS). The mechanisms of the virus, though, are unclear.

Immunoglobulins, or antibodies, are a critical part of the immune response by recognizing particular bacteria or viruses and aiding in their destruction. A team of researchers at the University of Helsinki are studying biomarkers that signal an increase in hantavirus infection, particularly circulating κ and λ immunoglobulin light chains in blood serum samples, as outlined in their new publication Hepojoki et al. PLOS path 2021

Digitized images of kidney biopsies of patients with acute HFRS were analysed using Aiforia Create for detections of κ or λ light chain-positive cells in relation to total cell count. The AI training process included ~2500 iterations based on ~500 annotations. Their results suggest an increase in circulating antibodies and that excessive immune responses in humans toward the virus may play a role in the development of orthohantavirus infections. We interviewed principal investigator Tomas Strandin on his experience using Aiforia for image analysis in this publication.

Interview with Tomas Strandin, Principal Investigator, Department of Virology, University of Helsinki

What is your motivation in using AI for this project?

Tomas: AI is the method of choice to analyze multiple samples in a standardized fashion. This is especially important for histological sample materials, which interpretation is inherently subjective due to sample complexity. By making the analysis faster and more objective, AI is helping my work significantly and making it possible to analyze tissue material in a quantitative way.

Our sample materials are renal biopsies from 100 patients with Puumala hantavirus-caused kidney failure or control subjects, in which we analyzed the extent of various immune cell infiltrations with Aiforia to better understand the immunological responses that explain disease pathogenesis

How has it been getting started with Aiforia? Have you had experience with digital image analysis in the past?

Aiforia has a very user-friendly AI platform and coming in terms of using did require only a couple of hours. My past experience with digital image analysis is not very big, but my perception is that this type of analysis can be very hard to grasp to people coming from other fields of research. However, Aiforia goes a long way in proving that this perception is false.

How many images and annotations did you need to train the neural networks of your AI model?

Our project was not on neural networks but rather infiltration of various immune cells in renal tissues. I don’t remember the details of the AI training currently.

What are the biggest benefits AI provides in virology labs?

As I see it the biggest benefit is in the analysis of complex sample material. In a virology lab, this could include screening of several tissue samples from various sources for the presence and quantification of an infectious virus.

What are your next steps after this publication; do you plan to use AI for further research or other projects?

There are actually two publications where we have used Aiforia in a similar fashion (Hepojoki et al. PLOS path 2021 and Vangeti et al. PLOS path 2021). Next time when our research comes up to a point where we need to assess the expression of any given protein or cell in tissue samples of humans or in animals, I will definitely consider Aiforia as the method of choice. 

Read more about the use of Aiforia’s AI models in various infectious diseases here

References