Cardiovascular disease (CAD) is still the most prevalent disease although there are extensive improvements in mortality and healthcare. More patients who received initial treatment for heart failure suffer greatly from reperfusion injury, resulting in myocardial infarction (MI). Hemorrhagic MI is proven to be the severest form of MI, and there is the calling for comprehensively characterize MI for patient’s long-term survival and quality of life.
Cardiac magnetic resonance (CMR) imaging has been developed as first-line imaging modality for tissue characterization: fibrosis, edema, microvascular obstruction (MVO), intramyocardial hemorrhage (IMH) and its residual iron, and fat deposition. Those substrates are independent predictors of major adverse cardiac events (MACE) and manifests as different severity of MIs. Thus, being able to characterize MVO, IMH, MVO and its fat deposition is of great significance. Current practice of MI characterization continues to demand strong needs and technical challenges. Three needs were identified: a fast-imaging protocol to identifying MI, non-contrast technique to characterize multiple substrates underlying MI, and overlooked influence of spatial resolution on IMH detection.
First, the lengthy CMR protocol in multiple tissue characterization may have impeded its implementation. A fast, comprehensive acquisition to characterize MI, MVO, IMH, cardiac function is on demand. In Chapter 2, we developed a free-breathing, non-ECG gating 3D T1, T2* imaging approach based on low-rank tensor (LRT) framework to shorten the imaging protocol of the same conventional acquisition by factor of 3-4. This framework also overcomes critical image artifacts from undesirable cardiac and respiratory motion and provides additional information in MVO characterization.
Second, native T1 mapping has been extensively studied to provide diagnostic values in identifying acute and chronic MIs by characterizing edema or fibrotic tissue with elevated their T1 values. However, we realized that the co-existence of fibrosis, IMH and/or its induced fat deposition are causing inhomogeneity of T1 values within MI. Instead of treating them as confounding factors, in Chapter 3, we demonstrated native T1 heterogeneity within MIs as a new look for MI characterization and theorized it as a potential application for MI patient risk stratification. Specifically, we probed the heterogeneity of MIs in T1 mapping using entropy analysis, known as T1 entropy. We have shown that T1 entropy is capable of distinguishing MI from remote myocardium in both acute and chronic MI. We found T1 entropy is strongly associated with fat fraction and R2* (iron content) in the high fat fraction and high R2* group. The results were also validated in patient studies. In conclusion, T1 entropy is a promising predictor for heterogeneity of MI, and it is a reliable biomarker for analysis of severity of chronic MI.
Third, the current imaging protocol in T2* mapping for IMH and iron remnants detection is derived from iron overload from thalassemia. However, for iron deposits from chronic MI with wall thinning (< 6 mm), the long-overlooked influence of spatial resolution may play a significant role in its detection. The primary goal of Chapter 4 in this dissertation is to address the limitations by investigating the influence of spatial infarction on T2* mapping in the competing effect of partial volume and SNR with simulation phantoms, ex-vivo heart scans and in-vivo scans in detecting iron residuals from thinned myocardial wall. We found that the spatial resolution can be optimized given SNR greater than 10. However, the current in-vivo CMR protocol may be limited in detecting thinned narrowed iron band (< 1 mm) due to insufficient SNR level.
In general, this dissertation contributed MI tissue characterization with CMR by addressing the spatial, temporal, and contrast challenges. It paves the way for MRI to characterize MI for prognosis and therapeutic care of patients with the capability to characterize severe form of MI such as IMH. These challenges have been addressed in this dissertation.
In future work, our developed fully ungated free-breathing 3D LRT T1, T2* will be investigated for patients for MI characterization. And we will further specialize the technique due to its advantages in MVO characterization for delayed enhancement. The T1 entropy will be validated in post-contrast T1 mapping. We would also employ the free-breathing nature of LRT framework to improve the SNR profile for iron remnants detection in T2* mapping.
