Future Endoscopic Diagnosis
– Work Flow
We believe that changing our focus on endoscopic procedures from individual steps, such as diagnosis and therapy, to overall workflow will lead to improved quality, including efficiency and safety, in endoscopic medicine. Furthermore, we believe this will allow us to discover and address unmet needs.
Looking at Japan’s population demographics, one in four people are at age 65 or older, and the rate of aging will continue to increase.
Population aging is a worldwide issue, but the rate in Japan is particularly high. Notably, this is thought to be a factor in the increasing number of cases involving cancer in most digestive organs.
Endoscopes are playing a big role in the early detection and therapy of digestive organ cancer. An endoscope is a device with a thin tube that is inserted through the mouth, nose, or rectum to display organ interiors on a monitor via a small image sensor on the tube's tip.
Endoscopes have undergone many improvements since the first gastrocamera was invented in 1950. Glass fiber was introduced in the 1960s to make real-time observations possible. Since then, other important improvements such as higher-definition imaging, the use of special spectrum illumination, and wider-angle lenses has advanced. As for operability, improvements such as insertion tubes with flexibility adjustment and mechanisms to fit the insertion tubes into intestinal tracts has been continued.
Advancements to date have centered on improving observation performance and operability, but now that we are in the age of AI and ICT, the direction of innovation needs to change. As AI functionality replicates the skills of doctors, nurses, and paramedics and begins to be used to assist people, we anticipate this will lead to innovative performance.
Let's consider the direction of innovation by looking at three aspects of endoscopic- procedure workflow.
Beginning with computer-aided diagnosis (CAD), the application of AI in diagnostic imaging has a long history of research and development. The goal has been to support the early detection of pathological changes and the differentiation of benign and malignant growths. In recent years, the spotlight has focused on technology known as "deep learning" that uses neural networks. Research using this advanced technology already has been the subject of presentations at many academic conferences.
Using AI for diagnosis is a revolutionary step, but it must be used carefully to improve the value of healthcare. As an advance in CAD, broadening its application to various kind of subjects is anticipated. When a doctor reaches a diagnosis based on imaging, the logic of the conclusion entails the interworking of diverse knowledge. Developing CAD technology that can clarify as well as explain the structure of this knowledge is another target of the evolutionary process involving CAD.
The second key innovation is the use of AI for endoscope insertion, because inserting an endoscope into the large intestine requires advanced skills. An experienced doctor, by virtue of study and practice, begins with a mental image of the digestive tract. This, plus the combined use of images displayed on a monitor and hand sensations while manipulating the device, enable the doctor to insert the endoscope properly. In the case of less experienced doctors, however, they could improve such skills with the help of AI.
In pursuing this approach, however, one of the challenges is that endoscopes are not equipped with insertion related sensors that could otherwise provide information to serve a learning data for AI.
One possible solution is a system that indicates the endoscope's shape during insertion. A coil is installed inside the endoscope, its magnetic field is detected by an antenna outside the patient's body, and then this data is used to estimate the shape of the endoscope. However, this solution doesn't provide information about the surrounding intestinal tract or how much force is being exerted against the intestinal wall, so a better approach must be devised going forward.
The third key innovation for endoscopic- procedure workflow is the use of AI in cleaning and disinfection. Neither process is as simple as placing the endoscope in an automated cleaning device and pressing a button. The workflow includes pre-cleaning immediately after use, followed by brushing, testing, setting the cleaning/disinfection machine, drying, and storage. There also are manual procedures at every turn. This is why the processes depend on human skills.
To ensure proper cleaning and disinfection, a software has been developed to manage the process while the device is connected to a network. We believe that cleaning and disinfection could be made even safer through the incorporation of AI support, including by adding sensors, using ICT for data acquisition and confirmation.
So these are three key ways that innovation could be used to improve endoscopic- procedure workflow. By incorporating these and other innovative technologies, we firmly believe that the value of endoscopic medicine has potential for further improvement.