The National Taiwan University Hospital has recently launched a up-to-date diagnostic diagnostic service for pancreatic cancer.
The ntuh from the end of 2024, offered as an own service in the NTUH medical imaging department, contains the AI system advertised as the first of this kind in the world.
What is it about
The mandrel, which the hospital has developed with the Applied Mathematical Sciences Institute at the National Taiwan University, uses deep learning algorithms and a repeatedly image set of training data for automatic analysis of CT scans, sketching pancreas and flagning of suspicious changes. The system directly integrates with PACS NTUH and is available in all departments, including gastroenterology, surgery and oncology.
In national clinical validation, the AI system showed 80% sensitivity in detecting the early stages of pancreatic tumors less than 2 centimeters and general diagnostic accuracy of over 90%.
The technology received regulatory approval of the Taiwanese Food and Drug Agency, as well as a breakthrough designation of the device from the United States.
Why does it matter
According to NTUH, pancreatic cancer is one of the most complex cancers for treatment and diagnosis, and death cases gradually grow in Taiwan in recent years, and the five -year survival rate remains about 10%. He notices how pancreatic cancer-especially those below 2 cm, which are unnoticed for the human eye-are complex to notice “about 40% of changes”.
Therefore, the Pancreasaver research team developed their AI system to solve this challenge in the diagnosis of pancreatic cancer.
The team is now working on expanding the employ of the system to other conditions related to the pancreas, such as pancreatitis and cystic pancreas. They also take into account more multi -modal imaging and clinical data to further expand the range and accuracy of the diagnostic system.
Greater trend
Fujitsu i Southern tohoku general hospital In Japan, it also developed a similar solution based on AI to early diagnosis of pancreatic cancer. They began to create diagnostic software in 2022, using Fujitsu’s artificial intelligence, which was trained with anonymized images of CT pancreas.
Meanwhile, NTUH is also working on the development of huge languages models. At the end of last year, he acquired two up-to-date supercomputers that allow it Turn on more data types in LLM. The hospital aims to develop multimodal LLM to optimize surgery and improve patient services.