Automated multimodal segmentation of acute ischemic stroke lesions on clinical MR images.


Magnetic resonance (MR) imaging (MRI) is commonly used to diagnose, assess and monitor stroke. Accurate and timely segmentation of stroke lesions provides the anatomico-structural information that can aid physicians in predicting prognosis, as well as in decision making and triaging for various rehabilitation strategies. To segment stroke lesions, MR protocols, including diffusion-weighted imaging (DWI) and T2-weighted fluid attenuated inversion recovery (FLAIR) are often utilized. These imaging sequences are usually acquired with different spatial resolutions due to time constraints. Within the same image, voxels may be anisotropic, with reduced resolution along slice direction for diffusion scans in particular. In this study, we evaluate the ability of 2D and 3D U-Net Convolutional Neural Network (CNN) architectures to segment ischemic stroke lesions using single contrast (DWI) and dual contrast images (T2w FLAIR and DWI). The predicted segmentations correlate with post-stroke motor outcome measured by the National Institutes of Health Stroke Scale (NIHSS) and Fugl-Meyer Upper Extremity (FM-UE) index based on the lesion loads overlapping the corticospinal tracts (CST), which is a neural substrate for motor movement and function. Although the four methods performed similarly, the 2D multimodal U-Net achieved the best results with a mean Dice of 0.737 (95% CI: 0.705, 0.769) and a relatively high correlation between the weighted lesion load and the NIHSS scores (both at baseline and at 90 days). A monotonically constrained quintic polynomial regression yielded R2 = 0.784 and 0.875 for weighted lesion load versus baseline and 90-Days NIHSS respectively, and better corrected Akaike information criterion (AICc) scores than those of the linear regression. In addition, using the quintic polynomial regression model to regress the weighted lesion load to the 90-Days FM-UE score results in an R2 of 0.570 with a better AICc score than that of the linear regression. Our results suggest that the multi-contrast information enhanced the accuracy of the segmentation and the prediction accuracy for upper extremity motor outcomes. Expanding the training dataset to include different types of stroke lesions and more data points will help add a temporal longitudinal aspect and increase the accuracy. Furthermore, adding patient-specific data may improve the inference about the relationship between imaging metrics and functional outcomes.





Published Version (Please cite this version)


Publication Info

Moon, Hae Sol, Lindsay Heffron, Ali Mahzarnia, Barnabas Obeng-Gyasi, Matthew Holbrook, Cristian T Badea, Wuwei Feng, Alexandra Badea, et al. (2022). Automated multimodal segmentation of acute ischemic stroke lesions on clinical MR images. Magnetic resonance imaging, 92. pp. 45–57. 10.1016/j.mri.2022.06.001 Retrieved from

This is constructed from limited available data and may be imprecise. To cite this article, please review & use the official citation provided by the journal.



Cristian Tudorel Badea

Professor in Radiology

  • Our lab's research focus lies primarily in developing novel quantitative imaging systems, reconstruction algorithms and analysis methods.  My major expertise is in preclinical CT.
  • Currently, we are particularly interested in developing novel strategies for spectral CT imaging using nanoparticle-based contrast agents for theranostics (i.e. therapy and diagnostics).
  • We are also engaged in developing new approaches for multidimensional CT image reconstruction suitable to address difficult undersampling cases in cardiac and spectral CT (dual energy and photon counting) using compressed sensing and/or deep learning.


Alexandra Badea

Associate Professor in Radiology

I have a joint appointment in Radiology and Neurology and my research focuses on neurological conditions like Alzheimer’s disease. I work on imaging and analysis to provide a comprehensive characterization of the brain. MRI is particularly suitable for brain imaging, and diffusion tensor imaging is an important tool for studying brain microstructure, and the connectivity amongst gray matter regions.  

I am interested in image segmentation, morphometry and shape analysis, as well as in integrating information from MRI with genetics, and behavior. Our approaches  target: 1) phenotyping the neuroanatomy using imaging; 2) uncovering the link between structural and functional changes, the genetic bases, and environmental factors. I am interested in generating methods and tools for comprehensive phenotyping.

We use high-performance cluster computing to accelerate our image analysis. We use compressed sensing image reconstruction, and process large image arrays using deformable registration, perform segmentation based on multiple image contrasts including diffusion tensor imaging, as well as voxel, and graph analysis for connectomics.

At BIAC  my efforts focus on developing multivariate biomarkers and identifying vulnerable networks based on genetic risk for Alzheimer's disease.

My enthusiasm comes from the possibility to extend from single to integrative multivariate and network based analyses to obtain a comprehensive picture of normal development and aging, stages of disease, and the effects of treatments.  I am working on multivariate image analysis and predictive modeling approaches to help better understand early biomarkers for human disease indirectly through mouse models, as well as directly in human studies. 

I am dedicated to supporting an increase in female presence in STEM fields, and love working with students. The Bass Connections teams involve undergraduate students in research, providing them the opportunity to do independent research studies and get involved with the community. These students have for example takes classes such as:

BME 394: Projects in Biomedical Engineering (GE)
BME 493: Projects in Biomedical Engineering (GE)
ECE 899: Special Readings in Electrical Engineering
NEUROSCI 493: Research Independent Study 1

Unless otherwise indicated, scholarly articles published by Duke faculty members are made available here with a CC-BY-NC (Creative Commons Attribution Non-Commercial) license, as enabled by the Duke Open Access Policy. If you wish to use the materials in ways not already permitted under CC-BY-NC, please consult the copyright owner. Other materials are made available here through the author’s grant of a non-exclusive license to make their work openly accessible.