Cardiac Magnetic Resonance Imaging And Early Detection Of Ischemic Heart Disease

Cardiac magnetic resonance imaging(MRI)has become the gold standard for the diagnosis of heart disease because of the advantages of non-invasive,non-ionizing radiation,rich tissue contrast,and flexible imaging plan selection.In addition,cardiac MRI is the only versatile imaging modality that allows different methods to be used to detect myocardial damage at different disease stages.As the disease progress,myocardial injury presents as a series of changes from reduced myocardial perfusion to myocardial necrosis.These changes can be identified using different cardiac MRI methods.However,due to the constant beating of the heart,cardiac MRI has been the most challenging area of technology in the world.First,limited to acquisition time,myocardial first perfusion imaging in MRI lacks whole cardiac coverage compared to PET and SPECT.Second,existing T₂-weighted edema imaging techniques are susceptible to motion,resulting in signal loss.Also,intraventricular stagnation of blood could produce edema mimicking bright signals,which would lead to over estimation of the extent of edema.In addition,as cardiac quantitative magnetic resonance imaging,T₁ and T₂ mapping can also detect the edema area.But,the current quantitative imaging could not provide fast and stable results due to the cardiac motion and scanning time,which limits its clinical application.Magnetic resonance fingerprinting(MRF)solves the above problems,but its quantitative parameters cannot completely match those of the traditional fitting method,which affects its widespread application in the clinic.Therefore,this thesis aims to explore and improve various imaging methods for the early detection of ischemic heart disease through sequence design,parameter optimization,and physical model.The main work and contributions are:1)Myocardial first-pass perfusion with increased anatomic coverage using autocalibration multiband imagingOne of the keys to a comprehensive assessment of myocardial ischemia is to improve the coverage of myocardial perfusion imaging in the left ventricle.In this thesis,an autocalibrated multiband imaging with through-time encoding was included in the myocardial first-pass perfusion imaging sequence to increase the anatomic coverage without the need for reference scans.The proposed method achieves double coverage(proposed method vs conventional method:8 slices vs 4 slices)of the original first-pass perfusion imaging.In addition,a k-t space undersampling acceleration pattern was designed with compress sensing reconstruction,which effectively reduces the image blur and noise-like problems due to the contrast variation over time.2)The study of combining T₁ and T₂ weighted edema imagingImproving the contrast between edema and normal myocardium is one of the keys to improving myocardial edema detection.In this thesis,a magnetization preparation module T₂STIR was proposed to exploit the simultaneous elevation of T₁ and T₂ in edematous tissues for improved differentiation of edematous and normal myocardium while suppressing fat(proposed method vs conventional method:5.88±2.55 vs 2.75±2.00,P<0.05).The module was combined with single-shot b SSFP to make it insensitive to arrhythmia or breathing motion and was combined with segmented b SSFP to improve the spatial resolution of edema imaging.In addition,the parameters of the module were optimized by considering T₂ preparation time,inversion recovery time,heart rates,and flip angles to maximize the signal difference between the edema and the normal myocardium,thereby improving edema imaging.3)Fractional-order magnetic resonance fingerprinting in quantitative imagingOne of the challenges faced by MRF imaging technology is that the obtained quantitative parameter values are different from those of traditional quantitative imaging methods,therefore a unified quantitative value standard cannot be used.Except for the necessity of considering cardiac motion in cardiac MRI,cardiac MRF is essentially an MRF technique and still has the above-mentioned problem.Theoretically,the tissue quantitative parameters measured by different methods should be consistent,but the T₂ quantitative parameters measured by MRF tend to have a large deviation from those obtained by traditional methods.In this thesis,the physical model of magnetization to time was explored.Then,a signal evolution model with fractional-order Bloch equations to describe magnetic resonance relaxation was introduced in the dictionary generation of MRF.In addition,objective-driven pattern recognition was manipulated to estimate the relaxation values,which included the pre-knowledge constrain to alleviate the defects of fractional-order models that were susceptible to noise in highly undersampled scenarios,and ultimately improved the accuracy of T₁and T₂ estimation,especially T₂ values(taking brain tissue as an example,white matter T₂ reference value 63?80 ms,traditional method vs proposed method:53±7.5 ms vs70±4.7 ms;gray matter T₂ reference value 78?117 ms,traditional method vs proposed method:74±8.6 ms vs 95±3.8 ms).In this thesis,a variety of imaging techniques in cardiac magnetic resonance were studied for the early detection of ischemic heart disease.First,the autocalibration multiband combining with compress sensing method was proposed for fast and large-coverage myocardial first-pass perfusion imaging.Then,combined T₁ and T₂ weighted imaging were proposed to improve the detection of myocardial edema.Finally,this work explored the influence of the dictionary generation model on parameter measurement and improved the model to reduce the result that differs from traditional curve fitting methods in MRF.