Machine learning identifies esophageal cancer better than current methods

Researchers have developed a deep learning model to accurately identify cancerous esophagus tissue on microscopy images instead of the high-cost, time-consuming manual annotation process used by pathologists.

A research team at Dartmouth and Dartmouth-Hitchcock Norris Cotton Cancer Center tested their new machine learning approach for identifying cancerous and precancerous esophagus tissue on high-resolution microscopy images.

Pathology-Philips-CROP.jpg

Whole-slide images were collected from patients who underwent endoscopic esophagus and gastroesophageal junction mucosal biopsy, and an attention-based deep neural network framework was used to classify microscopy images.

Results of the study were published on Wednesday in JAMA Network Open.

“Previous methods for analyzing microscopy images were limited by bounding box annotations and unscalable heuristics,” state the authors. “The model presented here was trained end to end with labels only at the tissue level, thus removing the need for high-cost data annotation and creating new opportunities for applying deep learning in digital pathology.”

"Our new approach outperformed the current state-of-the-art approach that requires these detailed annotations for its training,” says Saeed Hassanpour, assistant professor in the Department of Biomedical Data Science at Geisel School of Medicine at Dartmouth College. "The result is significant because our method is based solely on tissue-level annotations, unlike existing methods that are based on manually annotated regions.”

Going forward, researchers plan to further validate their deep learning model by testing it on data from other institutions and conducting prospective clinical trials. In addition, they intend to apply the model to histological images of other types of tumors and lesions for which training data are scarce or bounding box annotations are not available.

“Our method would facilitate a more extensive range of research on analyzing histopathology images that were previously not possible due to the lack of detailed annotations,” adds Hassanpour. “Clinical deployment of such systems could assist pathologists in reading histopathology slides more accurately and efficiently, which is a critical task for the cancer diagnosis, predicting prognosis and treatment of cancer patients."

For reprint and licensing requests for this article, click here.