Classifying MRI motion severity using a stacked ensemble approach
MohammadReza Mohebbian1, Ekta Walia2, Mohammad Habibullah1
1Department of Electrical and Computer Engineering, University of Saskatchewan S7N 5A9, Saskatoon, Saskatchewan, Canada.
Abstract:
Motion artifacts are a common occurrence in Magnetic Resonance Imaging exam. Motion during acquisition has a profound impact on workflow efficiency, often requiring a repeat of sequences. Furthermore, motion artifacts may escape notice by technologists, only to be revealed at the time of reading by the radiologists, affecting their diagnostic quality. There is a paucity of clinical tools to identify and quantitatively assess the severity of motion artifacts in MRI. An image with subtle motion may still have diagnostic value, while severe motion may be uninterpretable by radiologists and requires the exam to be repeated. Therefore, a tool for the automatic identification of motion artifacts would aid in maintaining diagnostic quality, while potentially driving workflow efficiencies. Here we aim to quantify the severity of motion artifacts from MRI images using deep learning. Impact of subject movement parameters like displacement and rotation on image quality is also studied. A state-of-the-art, stacked ensemble model was developed to classify motion artifacts into five levels (no motion, slight, mild, moderate and severe) in brain scans. The stacked ensemble model is able to robustly predict rigid-body motion severity across different acquisition parameters, including T1-weighted and T2-weighted slices acquired in different anatomical planes. The ensemble model with XGBoost metalearner achieves 91.6% accuracy, 94.8% area under the curve, 90% Cohen's Kappa, and is observed to be more accurate and robust than the individual base learners.
Insights
This study introduces a deep learning tool to automatically detect and quantify motion artifacts in Magnetic Resonance Imaging (MRI) scans. The model accurately classifies artifact severity, improving diagnostic quality and workflow efficiency.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Radiology
Background:
- Motion artifacts are a frequent issue in Magnetic Resonance Imaging (MRI), often necessitating repeat scans and impacting diagnostic accuracy.
- Current clinical tools for identifying and quantifying MRI motion artifacts are limited.
- Subtle motion can retain diagnostic value, while severe motion may render images uninterpretable, requiring re-examination.
Purpose of the Study:
- To develop and validate a deep learning-based tool for automatic quantification of motion artifact severity in MRI brain scans.
- To assess the impact of subject movement parameters (displacement, rotation) on image quality.
- To enhance diagnostic quality and workflow efficiency in MRI examinations.
Main Methods:
- Development of a state-of-the-art stacked ensemble model utilizing deep learning.
- Classification of motion artifacts into five severity levels: no motion, slight, mild, moderate, and severe.
- Evaluation of the model's robustness across different acquisition parameters (T1-weighted, T2-weighted slices, various anatomical planes) and rigid-body motion.
Main Results:
- The stacked ensemble model achieved high performance metrics: 91.6% accuracy, 94.8% area under the curve, and 90% Cohen's Kappa.
- The model demonstrated robust prediction of rigid-body motion severity across diverse MRI acquisition parameters.
- The ensemble approach proved more accurate and robust compared to individual base learning models.
Conclusions:
- Deep learning offers a powerful method for automatic identification and quantification of motion artifacts in MRI.
- The developed stacked ensemble model can significantly aid in maintaining diagnostic quality and improving MRI workflow efficiency.
- This tool has the potential to reduce unnecessary repeat scans and support radiologists in image interpretation.
