Alibaba DAMO Academy Develops AI Model to Detect Micro Liver Tumors
Alibaba DAMO Academy, in collaboration with Shengjing Hospital and other institutions, has developed DAMO LiON, a liver cancer diagnosis AI model published in Nature Medicine. During a two-month clinical trial involving over 10,000 patients, the model successfully identified 15 micro-tumors (around 1 cm in size) that were previously missed by doctors. Early detection of liver tumors is notoriously difficult due to their small size and interference from conditions like fatty liver or cirrhosis. By improving diagnostic sensitivity by 11.5% and reducing reading time by 27%, this AI helps prevent missed diagnoses and enables junior doctors to perform at the level of senior specialists. DAMO LiON utilizes an improved network architecture that captures the relationship between lesions and the entire liver while preserving local textures and boundaries. It also performs iterative fusion of multi-phase contrast-enhanced CT (CE-CT) scans to detect pixel-level differences in transient micro-lesions.
## BACKGROUND
Contrast-enhanced computed tomography (CE-CT) is a common imaging technique used to diagnose liver malignancies, but small lesions (around 1 cm) are easily overlooked, especially when they are obscured by complex liver structures or when doctors are distracted by primary tumors elsewhere. DAMO Academy has previously developed other medical AI models, such as DAMO PANDA for pancreatic cancer and DAMO GRAPE for gastric cancer, to assist in early cancer screening.