FUJIFILM Holdings

CSR activity report│Fighting Against COVID-19

Aiming at AI Diagnosis of COVID-19 Lung Infection

Applying interstitial pneumonia quantitative technology to analyze and assist treatment outcome decision

GGO identification results are shown as blue on different lung images (A to D), and the appearance frequency in different areas is shown as a graph (E).

Ahead of others, Fujifilm commenced development of a medical AI technology and later launched a new product under the brand name REiLI. We are also participating in a number of projects related to medical AI technology collaborating with universities and research institutions. To fight against COVID-19, we started developing a lung diagnostic technology to support the assessment of treatment effects and recovery status based on algorithms to quantify interstitial pneumonia*1 lesions that we jointly developed with Kyoto University. Collaborating with medical institutions in Japan that accommodate COVID-19 patients, such as the Kanagawa Cardiovascular and Respiratory Center, we are striving to make further practical progress with this technology.

To make a pathological assessment of the pneumonia, a physician has to visually go through a number of chest CTscan images per patient. This is highly demanding even for an experienced specialist. Our interstitial pneumonia quantitative technology produces quantified data of the interstitial pneumonia status by automatically identifying, categorizing, and measuring seven different characteristics of lung lesions.

Physician can then make a diagnosis based on this quantified data, which reduces the workload and irregularities in diagnosis results. This in turn makes it easier to estimate the chronological changes, such as recovery status, a feature that can also be used for pharmaceutical development and assessment. We aim to commercialize this technology as soon as possible with help from a number of physicians who are providing training data for the AI system.

  • *1 Interstitial pneumonia: A group of lung diseases in which the organ hardens due to inflammation and fibrosis. Ground glass opacity (GGO), a typical imaging feature of COVID-19, is similar to that of interstitial pneumonia. GGO is a hazy opacity that appears on an X-ray image indicating inflammation in the interstices of bronchial structures or alveoli.
  • * This page contains some excerpts from the Sustainability Report.