AI Tool Detects Cancer in Blood Samples In 10 Minutes

Detecting cancer recurrence or spread often depends on identifying rare tumor cells circulating in the bloodstream — a process known as a liquid biopsy. However, current methods rely on trained specialists spending hours combing through images of millions of blood cells to find a few abnormal ones. Researchers have now developed an artificial intelligence (AI) algorithm that automates this painstaking task, enabling the detection of cancer cells in approximately ten minutes with unprecedented speed and accuracy.

The new algorithm, called RED (Rare Event Detection), was developed by scientists from the USC Viterbi School of Engineering (Los Angeles, CA, USA) and the USC Dornsife College of Letters, Arts and Sciences (Los Angeles, CA, USA). Unlike existing liquid biopsy systems that rely on human review or predefined cellular features, RED uses deep learning to autonomously detect “outlier” cells that differ from millions of normal blood cells. It does not need prior knowledge of what a cancer cell looks like.

Instead, it identifies unusual cellular patterns and ranks them by rarity, automatically flagging potential cancer cells for further review. In laboratory tests, the researchers evaluated RED using two datasets: blood samples from patients with advanced breast cancer and simulated samples where cancer cells were added to normal blood. The findings, published in Precision Oncology, show that the algorithm achieved a 99% detection rate for epithelial cancer cells and 97% for endothelial cells, while reducing the data volume for human review by a factor of 1,000.

Compared to traditional methods, RED identified twice as many relevant cells associated with cancer, highlighting its ability to eliminate human bias and uncover subtle biological signals. By combining computational modeling with human expertise, the USC team has built a framework that accelerates the analysis of liquid biopsies and supports ongoing cancer monitoring. The approach is already being applied to study outcomes for breast, pancreatic, and multiple myeloma cancers. RED’s ability to detect rare cancer cells in blood samples could also enhance patient surveillance, enabling earlier detection of recurrence and more effective treatment planning.

“This is one of the really great examples where modern AI is really changing the way we do healthcare research,” said Peter Kuhn, University Professor and Director of the Convergent Science Institute in Cancer at USC. “Our next step is to continue pushing the forefront of AI to radically change our ability to find cancer in the blood of patients early.”

Related Links:
USC Viterbi School of Engineering
USC Dornsife College of Letters, Arts and Sciences