An ultrafast MRI-based radiomics model performs just as well as standard MRI and isn’t dependent on radiologist experience, according to research published January 25 in La Radiologica Medica.
A team led by Bo Kyoung Seo, MD, PhD, from Korea University in Ansan City, South Korea, found that the ultrafast MRI radiomics model successfully classified hormone receptors, HER2 status, and molecular subtypes on par with standard MRI’s performance.
“Our study underscores the potential clinical utility of ultrafast MRI radiomics in breast cancer classification, irrespective of readers' experience levels,” the team wrote.
Ultrafast MRI offers faster temporal resolution than standard MRI, allowing for shorter scan times and better assessment of tumor kinetics. The researchers also highlighted that the modality could improve lesion conspicuity by capturing cancer enhancement before background enhancement. It also provides information on time to enhancement (TTE), maximum slope, and wash-in slope.
Previous research suggests that MRI-based radiomics models perform well in predicting complete pathologic response to breast cancer treatment. Features used for radiomics models can provide systematic analysis of images by overcoming the limitations of subjective analysis and dependence on the experience of radiologists interpreting MR images.
Seo and colleagues compared the performance of ultrafast MRI with standard MRI in classifying histological factors and subtypes of invasive breast cancer among radiologists with varying experience.
The prospective study enrolled 225 women with 233 breast cancer prior to treatment between 2021 and 2022.
Two readers, including a breast radiologist with 24 years of experience in breast imaging (reader 1) and a radiology resident with two years of experience in breast imaging (reader 2), performed tumor segmentation on MRI. From there, the researchers extracted 1,618 radiomic features and four kinetic features from ultrafast and standard MR images. They also performed feature selection by the least absolute shrinkage and selection operator (LASSO) and adopted logistic regression algorithms for prediction modeling.
The ultrafast MRI model outperformed the standard MRI model in predicting HER2 status and classifying subtypes while both modalities performed similarly in predicting hormone receptors.
| Performance of MRI radiomics models, readers in predicting HER2 status | |
|---|---|
| Method | Area under the curve (AUC) |
| Ultrafast MRI model (Reader 1) | 0.87 |
| Ultrafast MRI model (Reader 2) | 0.88 |
| Standard MRI model (Reader 1) | 0.77 |
| Standard MRI model (Reader 2) | 0.77 |
The researchers also reported that ultrafast MRI showed superior classification of the following: luminal subtype for both readers, the HER2-overexpressed subtype for reader 2, and the triple-negative subtype for reader one (p
The study authors highlighted ultrafast MR radiomics imaging as a noninvasive tool in classifying breast cancers according to subtypes and histological factors.
“Our study demonstrates the promise of radiomics approaches with ultrafast MRI that have advantages in terms of scan time and lesion conspicuity, which are problems with standard MRI,” they concluded.
The full study can be accessed here.
Whether you are a professional looking for a new job or a representative of an organization who needs workforce solutions - we are here to help.