How to solve problems with sparse modeling (Part 4)

How To Solve Problems With Sparse Modeling (Part 4)

Hello everyone, My name is Kenshin Fujiwara and I am the CEO and founder of HACARUS Inc. 

Through this series of blogs, I will discuss a wide range of topics of AI, from the history of AI to practical tips for the successful application of AI projects. I hope my blog posts will help you better understand AI and solve your business issues.

In the last blog, we studied the case of an electronic equipment manufacturer which applied AI solutions to the inspection system. Today, we will look at another case study of AI implementation for diagnosis support at Kobe University.

Case Study 3: How AI Can Support Medical Professionals – Liver Cancer Diagnosis AI Research at Kobe University

Each year, around 25,000 people die from liver cancer in Japan. For men 18 years old and older, liver cancer is the third leading cause of death behind stomach cancer and colorectal cancer. 

At Kobe University’s Graduate School of Medicine, the Department of Diagnostic Radiology has been conducting research on MRI image analysis and diagnostic support AI focusing on Hepatocellular Carcinoma. 

Challenges Prior to Introducing AI

Before introducing AI, medical professionals were facing the following issues:

  • Manual inspection placed a high burden on diagnostic imaging physicians
  • Interpretability is a requirement for AI systems, something that has been previously impossible with conventional deep learning methods

When physicians use MRI machines to diagnose liver cancer, they need to review hundreds of images. These images are sliced along the Z-axis, the central axis of the MRI machine cylinder. Since each lesion needs to be carefully analyzed, it takes a lot of time just to help one patient. This high-stress work, combined with a high volume of patients, places a lot of stress on the physicians. 

Medical staff who look at images and make diagnoses are called radiologists. These individuals require specialized knowledge and skills, and there is currently a shortage of skilled radiologists. 

Reduce Diagnostic Burden on Radiologists

The main goal of this case study was to reduce the diagnostic burden on these radiologists and reduce the time it takes them to diagnose liver cancer. This was done by implementing AI to detect the risk areas of liver cancer in MRI images. 

While working on this AI project, it wasn’t easy to carry out in practice. Currently, the law prohibits AI from actually diagnosing diseases. Article 3 of the Medical Practitioners Act stipulates that “A physician shall not give medical treatment without examining the patient himself/herself or issue a medical certificate or prescription. In addition, they shall not issue a birth certificate or stillbirth certificate without attending the birth.”

Complying with these requirements, the AI solutions were implemented as a form of support, where the AI was positioned to provide support for the human physician’s diagnosis. The general flow of the support system started with the AI suggesting areas in the MRI image that were suspected of liver cancer. The doctor then performs their own check and confirms or rejects the diagnosis. 

While AI cannot make a diagnosis, simply having AI to detect high-risk areas in advance could reduce the burden on physicians by a considerable amount. By simply identifying suspicious images in advance, AI could save a lot of time and labor, which used to be necessary to check dozens of images for each patient.

Interpretability of AI Solutions

Since the AI would be assisting the physicians who were the ones to make the final diagnosis of a patient’s illness in the end, the interpretability of the results was also extremely important. Therefore, the AI system could not be used unless it provided clear evidence of the areas it identified as high risk and how it came to this conclusion. 

In order to make this AI available to other medical institutions, it had to be reviewed and approved by the Pharmaceuticals and Medical Devices Agency (PMDA), an independent administrative agency under the jurisdiction of the Ministry of Health, Labour, and Welfare.

Since these programs involve human life, a large number of materials had to be submitted and a rigorous review process is taking place. Even though this is a tedious process, it is a necessary step. If the effort isn’t taken now to push this technology forward, it will never be introduced in medical institutions across Japan. 

It is also important to note that liver cancer is not unique to Japan. After obtaining approval from PMDA, the next step is to go global with this technology. Hopefully, the AI developed through this joint research will be used in medical practice across the globe. 

In the next blog, I will introduce another case of AI implementation for AI diagnosis support at Kyoto University. For your updated knowledge and insight about AI technology, subscribe to our newsletter or visit HACARUS website


Subscribe to our newsletter

Click here to sign up