Machine learning algorithms applied to biopsy images can shorten the time for diagnosing a gut disease that often causes permanent physical and cognitive damage in children.
The disease affects 20 percent of children under the age of 5 in low- and middle-income countries, such as Bangladesh, Zambia and Pakistan, but it also affects some children in rural Virginia.
Syed and Brown are using a deep learning approach called “convolutional neural networks” to train computers to read thousands of images of biopsies.
Pathologists can then learn from the algorithms how to more effectively screen patients based on where the neural network is looking for differences and where it is focusing its analysis to get results.
“These are the same types of algorithms Google is using in facial recognition, but we’re using them to aid in the diagnosis of disease through biopsy images,” said Brown.
The machine learning algorithm can provide insights that have evaded human eyes, validate pathologists’ diagnoses and shorten the time between imaging and diagnosis, and from a technical engineering perspective, might be able to offer a look into data science’s “black boxes” by giving clues into the thinking mechanism of the machine.