Related Experiment Video
Updated: Jun 9, 2026
How to Measure Cortical Folding from MR Images: a Step-by-Step Tutorial to Compute Local Gyrification Index
Published on: January 2, 2012
Mitigating the Impact of MR Sequence Parameters: Increasing the Robustness of DL-Based Cortical Thickness Estimates
Timo Blattner1, David Romascano1, Richard McKinley1
1Support Center for Advanced Neuroimaging (SCAN), University Institute of Diagnostic and Interventional Neuroradiology, University of Bern, Inselspital, Bern University Hospital, Bern, Switzerland.
Abstract:
Cortical thickness measurements from MRI are increasingly used as biomarkers for neurodegenerative disease progression. However, variations in MRI acquisition parameters, such as inversion time (TI) and repetition time (TR), which are common in clinical settings, can compromise the reliability and sensitivity of these measurements. We fine-tuned a deep-learning-based segmentation tool (DL+DiReCT) to reduce its dependence to image contrast variations by training it on simulated MPRAGE images derived from quantitative relaxation maps. Fine-tuning markedly reduced contrast sensitivity, with the Pearson correlation coefficient decreasing from to . Evaluation on a synthetic atrophy dataset demonstrated that our model accurately replicated atrophy trends with minimal underestimation, outperforming FreeSurfer and SynthSeg. When applied to a dataset of relapsing-remitting multiple sclerosis (RRMS) patients, the fine-tuned model showed a substantial reduction in contrast sensitivity and maintained stable performance after controlling for covariates such as age, sex, field strength, and Expanded Disability Status Scale (EDSS) score. Overall, the proposed approach achieves robust contrast invariance without sacrificing sensitivity to cortical atrophy, offering a practical improvement for longitudinal and multi-center clinical studies.
Insights
This study fine-tuned a deep learning tool to make MRI-based cortical thickness measurements more reliable for tracking neurodegenerative diseases. The improved method reduces sensitivity to MRI contrast variations, enhancing accuracy in clinical studies.
Area of Science:
- Neuroimaging
- Medical Image Analysis
- Deep Learning
Background:
- Cortical thickness from MRI is a key biomarker for neurodegenerative diseases.
- Variations in MRI acquisition parameters (e.g., TI, TR) affect measurement reliability.
- Existing segmentation tools can be sensitive to image contrast differences.
Purpose of the Study:
- To develop a deep learning model (DL+DiReCT) robust to MRI contrast variations for accurate cortical thickness measurement.
- To improve the reliability and sensitivity of neuroimaging biomarkers in clinical settings.
- To enhance the performance of automated segmentation tools for neurodegenerative disease research.
Main Methods:
- Fine-tuning a deep learning segmentation tool (DL+DiReCT) using simulated MPRAGE images from quantitative relaxation maps.
- Training the model to reduce dependence on image contrast variations.
- Evaluating the model on synthetic atrophy datasets and a cohort of relapsing-remitting multiple sclerosis (RRMS) patients.
Main Results:
- Fine-tuning significantly reduced contrast sensitivity (Pearson correlation from -0.644 to 0.094).
- The model accurately replicated atrophy trends on synthetic data, outperforming FreeSurfer and SynthSeg.
- The fine-tuned model demonstrated stable performance in RRMS patients, even after controlling for covariates.
Conclusions:
- The proposed approach achieves robust contrast invariance in cortical thickness measurements.
- This method offers a practical improvement for longitudinal and multi-center neuroimaging studies.
- The contrast-invariant tool enhances the reliability of MRI biomarkers for neurodegenerative disease progression.
