Amid a groundbreaking advancement in medical technology, Chinese researchers have introduced an artificial intelligence (AI) system that can detect early stage gastric cancer using standard CT scans. This breakthrough has the potential to revolutionize cancer detection worldwide.
The breakthrough pushes Chinese scientists and tech companies to take global lead in AI-powered healthcare, with a widespread impact on millions at risk of fatal diseases, News.az reports citing Xinhua.
The Grape model, developed through a collaboration between Zhijiang Cancer Hospital and Alibaba's Damo Academy, represents a major advance in early cancer detection.
The study, published last week in the Nature Medicine Journal, shows how AI can identify gastric tumors from non-contrast CT imaging, a method previously thought to be relatively ineffective for gastric cancer screening.
Gastric cancer is one of the most fatal malignant tumors, especially in Asia. According to the China National Cancer Center, the country records 360,000 new cases and 260,000 deaths per year, with late-stage diagnosis contributing to a disastrous survival rate of less than 30%, with five-year survival rates.
In contrast, patients diagnosed at stage 1 have a survival rate of >90%. However, traditional screening methods rely on invasive endoscopy, which is expensive, uncomfortable, and often prevent participation.
Grapes address these challenges by turning everyday CT scans, already widely used in hospitals and clinics, into powerful screening tools.
“Identifying gastric cancer via AI algorithms is extremely difficult due to the swelling of the stomach and the significant variation in shape,” said Xia Yingda, algorithm expert for the Damo Academy Medical AI team.
Trained with around 100,000 scans from 20 medical centers across China, AI achieved 85.1% sensitivity on early tumor spots, surpassing human radiologists by a significant margin.
“We registered a contrast-enhanced CT scan with non-contrast CT. This allowed the AI to acquire intense precision information and accurately identify the location of the stomach cancer,” explained Xia, co-first author of the paper.
In actual trials at two hospitals, the system increased detection rate to 24.5%. In particular, approximately 40% of the cases detected were patients with symptoms-free gastric cancer.
“AI models allow for image-based gastric cancer screening for the first time,” said Cheng Xiangdong, principal investigator at Zhejiang Cancer Hospital.
“After all, even the most advanced and accurate surgical treatments cannot match the effectiveness of early diagnosis and treatment,” Chen added.
The researchers retrospectively analyzed CT images of 11 patients prior to the diagnosis of gastric cancer and found that the AI model could detect gastric cancer 2-10 months ago.
A 45-year-old patient diagnosed with locally advanced gastric cancer had undergone a non-controlled CT scan 6 months ago for another disease with no gastrointestinal symptoms, and the AI model was shown to be able to detect signs of gastric cancer at that time.
Grapes have already been deployed in hospitals in Zhijiang and Anhui provinces in eastern China, with plans for wider promotions both nationally and internationally, Cheng said.
The breakthrough followed China's previous success with Panda, an AI model designed to detect pancreatic cancer, one of the most elusive killers in medicine, through regular CT imaging.
Pandas showed 92.9% sensitivity and 99.9% specificity for cancer detection in clinical trials. Its adoption is particularly transformative in low-income regions where advanced screening methods are not available, and significantly reduces costs.
news.az
