Studies indicate that high-risk patients with early-detected pancreatic ductal adenocarcinoma (PDAC) have a median overall survival of 9.8 years, compared to 1.5 years for those diagnosed late. This was termed as the "king of cancers."
With a survival rate consistently below 10 percent, pancreatic cancer is hightlighted by the tragic losses of Apple's Steve Jobs in 2011 and chief scientist at the Chinese Centre for Disease Control and Prevention, Wu Zunyou.
Early Detection of King Cancer
A groundbreaking achievement in early-stage screening for one of the deadliest cancers has been attained through an artificial intelligence tool created by Chinese scientists. Li Ruijiang, an associate professor of radiation oncology at the Stanford School of Medicine, hailed the work as a significant step forward in pancreatic cancer screening.
South China Morning Post reported that the developed screening model specifically targets pancreatic ductal adenocarcinoma (PDAC), the primary subtype responsible for over 95 percent of pancreatic cancer cases.
With a specificity of 99.9 percent and a sensitivity surpassing mean radiologist performance by 34.1 percent, the model shows promise but requires further regulatory approval for practical implementation.
Despite the importance of early detection, there is a lack of accessible screening technology for the general population. The low prevalence of pancreatic cancer (less than 13 cases per 100,000) makes the use of expensive contrast-enhanced CAT scans economically impractical.
Lead author Cao Kai, from the Shanghai Institution of Pancreatic Diseases, highlighted the inadequacy of existing early screening tools, often leading to misdiagnoses and unnecessary panic. To address this, Cao and Lu Le, the leader of DAMO Academy's medical team, conceived the idea of using AI for early cancer screening.
Creating PANDA
As posted on Nature Medicine, they both initiated a research project with over 10 medical institutions, aiming to develop a technology combining non-contrast CAT scans with AI for large-scale pancreatic cancer screening.
The result of their collaboration was the creation of an algorithm named "Pancreatic Cancer Detection with Artificial Intelligence," abbreviated as PANDA. This innovative system underwent training using over 3,200 image sets from a prominent pancreatic cancer institution in China. Notably, around 70 percent of these sets originated from patients with pancreatic lesions.
Benefiting from the extensive dataset, rigorous data processing, and a pioneering training strategy, News Medical reported that PANDA emerged as a highly perceptive AI imaging expert.
The study revealed PANDA's efficient detection of lesions in a multi-center validation cohort, surpassing a radiologist's average performance by 6.3% in specificity and 34.1% in sensitivity for pancreatic lesions.
PANDA achieved 92.9% sensitivity and 99.9% specificity in a large-scale validation. The researchers demonstrated the effectiveness of PANDA through a process involving the curation of a large dataset, transfer of lesion annotations, and a deep learning approach.
In conclusion, PANDA exhibits high specificity and sensitivity for detecting and diagnosing pancreatic lesions, showcasing its potential for large-scale screening and early detection of pancreatic ductal adenocarcinoma (PDAC).
Related Article : This Implantable Device Promises to Shrink Pancreatic Tumors-How?
