Extracting radiomic features from MR images helps with breast lesion classification

Extracting radiomic features from MR images can help radiologists distinguish between benign breast lesions and luminal A breast cancers, according to a new study published by Academic Radiology.

“In 2012, 74 percent of diagnosed breast cancers were type luminal A, the most of any molecular subtype,” wrote lead author Heather M. Whitney, PhD, department of radiology at the University of Chicago, and colleagues. “Therefore, it is of particular interest to identify radiomic signatures that aid in the diagnosis and prognosis of luminal A breast cancers.”

The authors tested the effectiveness of this machine learning method by studying data from more than 500 breast lesions imaged with MRI at a single medical center. The lesions were classified by using maximum linear size alone and by extracting 39 radiomic features—volume, entropy, signal-to-enhancement ratio, and so on—from the MR images.

Using maximum linear size alone to classify the lesions, the area under the receiver operating characteristic curve (AUC) was 0.797. Classification using all radiomic features had an AUC of 0.846, while classification using radiomic features “except those related to size” had an AUC of 0.848.

“This work demonstrated that in the clinical task of distinguishing between benign lesions and luminal A breast cancers, a radiomic signature using the features described here, quantitatively extracted from MR images, significantly improved the ability to classify the lesions,” Whitney and colleagues wrote.

The authors added that one limitation of their study was that the images were acquired at two different magnetic field strengths. “Our group is currently investigating the impact of field strength difference on this particular classification task,” they wrote. “At the same time, our inclusion of features extracted from images at both field strengths enabled us to maximize statistical power, and our aim for this work was to describe classification performance for clinical populations, for which imaging can be conducted at the two field strengths used here.” 

Michael Walter
Michael Walter, Managing Editor

Michael has more than 16 years of experience as a professional writer and editor. He has written at length about cardiology, radiology, artificial intelligence and other key healthcare topics.

Trimed Popup
Trimed Popup