Techniques for cardiac tissue characterization using magnetic resonance imaging

The characterization of different cardiac tissues is important for therapeutic decision-making in patients with heart disease. Magnetic resonance imaging (MRI) is playing a significant role in cardiac imaging, and different MRI pulse sequences have been developed to enable functional and viability imaging of the heart. In the proposed work, different techniques are introduced for acquiring better cardiac functional and viability MR images. Inversion recovery (IR) is the standard MRI technique for acquiring delayed-enhancement (DE) cardiac viability images. However, the resulting images have poor infarct-blood contrast. Stimulated-echo acquisition mode (STEAM) is an MRI technique that has many advantages including black-blood property, and has successfully been used for myocardial imaging. In the proposed work, a new technique, based on the STEAM pulse sequence, is introduced to obtain a black-blood viability image of the heart. The resulting image has sharp infarct-blood border and shows good agreement with standard IR-DE viability images. Strain-encoding (SENC) imaging is an MRI technique that has successfully been implemented for myocardial functional imaging. In the proposed work, a new technique, based on SENC imaging, is introduced to result in both functional and viability images without increasing scan time. Because acquired at the same cardiac phase, the resulting functional and viability images can be used to construct one composite image of the heart without misregistration problems. In addition, an unsupervised fuzzy clustering technique is proposed for identifying different tissue types from the resulting SENC functional and viability images. We showed that the clustering technique is robust and fast in identifying blood, normal myocardium, infarcted myocardium, and background. In addition, the technique can identify non-contracting myocardium adjacent to the infarction, which may represent hibernating or stunned myocardium that constitutes a major predictor of patient recovery after surgical operation. The SENC functional images are improved one step further by introducing a technique that corrects for tissue through-plane motion while reducing scan time to a single heartbeat. We showed that the results are significantly better than those acquired without tissue through-plane correction. Despite the aforementioned advantages, SENC images have low signal-to-noise ratio (SNR). One solution to this problem is to implement coherent steady-state pulse sequences, like balanced steady-state free precession (bSSFP) pulse sequence. bSSFP had recently been implemented for acquiring cardiac tagged images with high SNR and fast acquisition. However, the signal from tagged myocardium fades at later cardiac phases due to magnetization relaxation. Thus, as a final addition to the aforementioned contributions, a new technique is proposed for enhancing the SNR of tagged myocardium at the later cardiac phases, which results in constant tagging contrast throughout the whole cardiac cycle, and allows for better image analysis.