Earlier Identification of Pancreatic Cancer

May 27, 2019

Johns Hopkins uses deep learning algorithms to identify pancreatic cancer

Today, only about 15-20% of patients diagnosed with pancreatic cancer receive their diagnosis early enough to qualify for surgical treatment options, and only 7% of pancreatic cancer patients live for five years after being diagnosed with the disease. Researchers at Johns Hopkins are developing deep learning algorithms to review CT scans with the aim of finding pancreatic tumors sooner.

To develop the algorithms, researchers have supercomputers review hundreds of abdominal CT scans for changes in texture and pattern in the pancreas and nearby organs. Using a deep learning system to review these scans could help radiologists spot tumors even in cases where pancreatic cancer isn’t suspected, potentially changing the outcome for pancreatic cancer patients.