TL;DR
This paper introduces a novel free-breathing 3D cardiac ECV mapping technique that overcomes the limitations of existing 2D methods. By employing a sophisticated image reconstruction model (LTSA) and accounting for time-varying T1 signals after contrast injection, the method enables comprehensive whole-heart ECV assessment in a clinically feasible timeframe.
What Problem Does This Paper Address?
Extracellular volume (ECV) is a crucial biomarker for assessing diffuse myocardial fibrosis, a condition difficult to detect with traditional imaging techniques. However, current ECV mapping methods are largely confined to 2D acquisitions and often require breath-holding, limiting their ability to capture the entire heart and potentially causing patient discomfort. Furthermore, the dynamic changes in T1 values after contrast agent injection pose a significant challenge for accurate ECV quantification, especially in 3D settings. This paper aims to bridge this gap by developing a free-breathing, 3D ECV mapping method that provides whole-heart coverage with improved accuracy and practicality.
Methods
The core of this work lies in a two-pronged approach: advanced image reconstruction and a refined T1 mapping model.
- Image Reconstruction: The researchers utilized a Linear Tangent Space Alignment (LTSA) model-based framework for reconstructing dynamic MR images. This method leverages the intrinsic low-dimensional manifold structure of the spatiotemporal signals to recover high-frame-rate images from sparsely sampled (k,t)-space data. This is a significant advancement over traditional low-rank (LR) methods, as LTSA can capture more complex, non-linear relationships in the data, leading to improved image quality and reduced artifacts.
- T1 Mapping and ECV Calculation: Cardiac T1 mapping was performed using a free-breathing, ECG-gated inversion-recovery sequence. A key innovation is the introduction of a linearly time-varying T1 model for post-contrast T1 mapping. This model explicitly accounts for the changes in T1 values over time after contrast agent administration, which is crucial for accurate ECV estimation. Joint estimation of T1 and transmit B1 (B1+) was also performed to correct for B1 inhomogeneity, a common issue at 3T. Finally, ECV maps were generated by aligning pre- and post-contrast T1 maps using affine transformation and applying a standard ECV formula.
Datasets
The study utilized data from six healthy volunteers scanned on a 3T MRI scanner. No new benchmark datasets were introduced; instead, the proposed method was evaluated against the established 2D MOLLI method for comparison.
Key Insights and Ideas
- LTSA for Dynamic MR Reconstruction: The application of LTSA for reconstructing dynamic cardiac MR images from undersampled data proved highly effective. It significantly improved image quality, particularly in delineating the myocardium and reducing artifacts compared to simpler low-rank methods.
- Time-Varying T1 Model is Crucial: Explicitly modeling the time-varying nature of T1 after contrast injection is essential for accurate post-contrast T1 and, consequently, ECV mapping. This approach addresses a fundamental challenge in dynamic contrast-enhanced imaging.
- Free-Breathing 3D ECV is Achievable: The study demonstrates that comprehensive, whole-heart 3D ECV mapping can be achieved without breath-holding, enhancing patient comfort and potentially improving data quality by eliminating breath-hold-induced motion.
- B1+ Correction Matters (but maybe not for ECV alone): While B1+ correction is important for accurate T1 estimation, preliminary findings suggest that for ECV estimation alone, its necessity might be reduced if the bias from B1+ inhomogeneity is consistent between pre- and post-contrast measurements.
Key Results
- Image Quality: The LTSA-based reconstruction yielded visually superior images with clearer myocardial wall delineation and fewer artifacts compared to LR methods.
- T1 and ECV Accuracy: The proposed method demonstrated good qualitative and quantitative agreement with the reference 2D MOLLI method for both pre- and post-contrast T1 maps and ECV maps.
- Quantitative Agreement: Average pre-contrast myocardial T1 values were 1250 ms for the proposed method and 1253 ms for MOLLI. Average myocardial ECV was 31.7% for the proposed method and 33.08% for MOLLI.
- Scan Time: The total imaging time for pre- and post-contrast T1 mapping was approximately 10.1 ± 1.4 minutes and 10.6 ± 1.6 minutes, respectively, which is considered practically feasible.
Limitations and Future Work
- Misregistration Errors: The study acknowledges limitations due to misregistration errors between pre- and post-contrast T1 maps, even with a non-rigid registration method. Future work could explore more advanced deformable image registration techniques.
- Computation Time: The computation time for T1 fitting, especially for the post-contrast case, was noted as relatively long. Optimizations in hardware or algorithms could address this.
- Bias Similar to MOLLI: The data acquisition method might introduce biases similar to MOLLI due to blood flow and magnetization transfer effects.
- Healthy Volunteers Only: The study was conducted on healthy volunteers. Future work should validate the method in patients with cardiovascular diseases, particularly those with diffuse fibrosis.
- Histological Validation: Histological validation through surgery or biopsy is suggested for further confirmation of accuracy.
- Contrast Agent Administration: The assumption of linearly varying T1 after bolus injection might not hold in all circumstances. Exploring alternative contrast administration strategies (e.g., continuous infusion) or static T1 modeling for ECV estimation with sufficient delay is suggested.
Summary updated