The more we attempt to solve the mysteries of the human body and life, the more we learn of their wonders and preciousness. While the heart is one of the most difficult organs to understand, R&D on the "Heart Simulator," which simulates and visualizes the heart's complex movements on a computer, is underway. We interviewed Masahiro Watanabe, Heart Simulator R&D Leader at Fujitsu, about the history of this revolutionary technological development and innovations that contribute to medicine and society.
Facing the Complexity of Cardiology
"Many die from heart disease, which is the second leading cause of death in Japan. In the U.S., it is the number one cause of death. Heart disease ranks high as the cause of death in developed countries worldwide. At the same time, medical expenses are soaring in Japan, exceeding 40 trillion yen.
By simulating the patient's heart on a computer so that treatment and operation results can be verified in advance, we may be able to decrease the number of deaths due to heart disease. We may also be able to reduce medical expenses. Hoping to develop such an application, we launched a heart simulator development project with the University of Tokyo in 2008."
Wanting to Contribute to Society Using Supercomputers
"When the K computer development project was launched, Yoshimasa Kadooka, who was responsible for developing high-efficiency applications for supercomputers back in 2007, sought to develop a simulator that could contribute to society. He focused on a heart simulator as something to put into practical use. As members with advanced degrees gathered together to develop this simulator, I joined as someone researching blood flow simulation and virtual reality (VR) related to the human main artery at JAIST (the Japan Advanced Institute of Science and Technology).
Research Should Not Be Gated by Akamon
"Prior to my joining Fujitsu, Professor Toshiaki Hisada (currently, Professor Emeritus) and Professor Seiryo Sugiura (currently, Professor Emeritus) of the University of Tokyo were developing a simulator of the left ventricle. At that time, it was extremely difficult to jointly simulate blood flow, which is a fluid, and heart muscles, which are structural objects. I was very surprised when I first saw their research that achieved this.
When I joined the company, Profs. Hisada and Sugiura had already managed to simulate the entire heart. Kadooka strongly felt that this technology should not be gated behind Akamon (the red gate at the University of Tokyo). The members who shared the same idea including myself were assigned to three sections, led respectively by Prof. Hisada, Prof. Sugiura, and Kadooka—creating the model shape based on medical images, using that model shape to perform simulation, and visualizing the simulation results. We built a prototype of a simulation system that replicates the hearts of individual patients. Each individual member contributed their expertise to accelerate development."
Simulating Every Heart Movement from the Root
"There are thought to be more than six billion cells in our hearts, where mechanical, chemical, and electrophysiological phenomena occur in simultaneous, complex ways. The heart's physiology consists of communicating various information scaling from the molecular level to the organ level. One beat pumps approximately a single coffee cup's worth of blood. The R&D policy was to precisely simulate the heart's movement by modeling and introducing all these coordinated phenomena at each scale from the root.
This was quite a challenging idea that would have required infinite time if the professors and project team had pursued the research alone. However, for individual fields such as cells and proteins, heart researchers around the world had already done successful modeling cases and clinical studies. By verifying and incorporating each of these studies while improving any questionable areas, a highly precise heart simulator that can be considered to be a compilation of heart research worldwide was born.
This was an excellent strategy on the part of the professors. They made this choice because they believed that their technology could be useful to society and they wanted to explore ways to introduce it to the real world as quickly as possible."
The Excitement When the K Computer Simulated the Movement of "1.5 Heartbeats"
"We repeatedly carried out trial and error with the professors to simulate the phenomena that occur at different intervals among different levels, such as heart muscle movements and the impact of cells and proteins at a lower level, in order to express them as a heartbeat. In 2011, with the number of heart cells narrowed down from six billion to 60,000, researchers took shifts day and night for a week to repeatedly execute and make corrections at a computation speed of 2.3 petaflops (PFLOPS: 1 petaflop can perform 1,000 trillion floating-point calculations per second) using the K computer. As a result, by continuously and exclusively operating nearly all K systems for approximately 17 hours, we successfully simulated 1.5 heartbeats.
I was in charge of visualization processing and was truly surprised when I saw all the data. This was considered to be the limit of what humans were capable of at that time in terms of simulating a complex organ such as the heart. I believe it was not recognized in the screening for the Gordon Bell Prize, which is awarded to achievements in the field of high-performance computing technology, because it was beyond exceptional."
Creating a Heart Simulator for Patients
"The initial goal of the joint research was to first create a detailed simulation of a standard heart, not that of a specific individual. The other goal was to envision simulating the hearts of individual patients and to develop technologies for that process from start to end. The heart's shape was generated and the functionality was configured. Electrocardiograms were incorporated and conditions for joining with surrounding blood vessels were defined because the heart is not a stand-alone organ. The research team exchanged ideas and rapidly created the basic parts.
For example, CTs and MRIs barely show the heart's outer boundaries in particular. Members struggled about how to simulate the shape from image data obtained from modalities as well as how much precision to expect. Foundational technologies essential for contributing to medicine were developed by discussing and deciding upon the color distribution, representation method, and so on with actual doctors. This ensured processing is quick and easy, and images output by the visualization tool after simulation can easily be read by doctors. It took quite some time, from 2008 to around 2012. The final technology we produced is composed of parts that can be employed separately in other fields."
Ten years have passed since this joint project started. In Part 2, we will describe the current state and future vision of research on heart simulators, which are beginning to be used in the real world.
Heart Simulator R&D Leader
Solutions Development Dept. I
Medical Solutions Div. III
Healthcare Solutions Unit II
The affiliation and title above are as of the interview held on January 11, 2019
2004: Received a PhD (Computer Science) for research on lesions resulting from aortosclerosis and mechanical stress due to blood flow.
Later, took part in MEXT's IT program "VizGrid Project" as a researcher, and promoted the development of a virtual conference system and research on flow analysis in the human main artery.
2007: Joined Fujitsu. Took part in the joint heart simulator development project within the K computer development project.
2015: Assigned as Manager of the Biosimulation Development Department at the Center for the Future of Medical Care. Has been in his present post since 2017.
Promotes commercialization of a heart viewer for education using VR technology.
Awarded a prize with two others for "Clinical research using a heart simulator" in the Research Category of the Prizes for Science and Technology at the 2018 Commendations for Science and Technology by MEXT.