Purpose
To develop an accurate and precise myocardial T1 mapping technique using an inversion recovery spoiled gradient echo readout at 3.0T.
Materials and Methods
The modified Look-Locker inversion-recovery (MOLLI) sequence was modified to use fast low angle shot (FLASH) readout, incorporating a BLESSPC (Bloch Equation Simulation with Slice Profile Correction) T1 estimation algorithm, for accurate myocardial T1 mapping. The FLASH-MOLLI with BLESSPC fitting was compared to different approaches and sequences with regards to T1 estimation accuracy, precision and image artifact based on simulation, phantom studies, and in vivo studies of 10 healthy volunteers and 3 patients at 3.0T.
Results
The FLASH-MOLLI with BLESSPC fitting yields accurate T1 estimation (average error = −5.4±15.1 ms, percentage error = −0.5%±1.2%) for T1 from 236–1852 ms and heart rate from 40–100 bpm in phantom studies. The FLASH-MOLLI sequence prevented off-resonance artifacts in all 10 healthy volunteers at 3.0T. In vivo, there was no significant difference between FLASH-MOLLI-derived myocardial T1 values and “ShMOLLI+IE” derived values (1458.9±20.9 ms vs. 1464.1±6.8 ms, p=0.50); However, the average precision by FLASH-MOLLI was significantly better than that generated by “ShMOLLI+IE” (1.84±0.36% variance vs. 3.57±0.94%, p<0.001).
Conclusion
The FLASH-MOLLI with BLESSPC fitting yields accurate and precise T1 estimation, and eliminates banding artifacts associated with bSSFP at 3.0T.
Background: To review and evaluate approaches to convolutional neural network (CNN) reconstruction for accelerated cardiac MR imaging in the real clinical context. Methods: Two CNN architectures, Unet and residual network (Resnet) were evaluated using quantitative and qualitative assessment by radiologist. Four different loss functions were also considered: pixel-wise (L1 and L2), patch-wise structural dissimilarity (Dssim) and feature-wise (perceptual loss). The networks were evaluated using retrospectively and prospectively under-sampled cardiac MR data.Results: Based on our assessments, we find that Resnet and Unet achieve similar image quality but that former requires only 100,000 parameters compared to 1.3 million parameters for the latter. The perceptual loss function performed significantly better than L1, L2 or Dssim loss functions as determined by the radiologist scores.Conclusions: CNN image reconstruction using Resnet yields comparable image quality to Unet with 10X the number of parameters. This has implications for training with significantly lower data requirements.Network training using the perceptual loss function was found to better agree with radiologist scoring compared to L1, L2 or Dssim loss functions.
Purpose: To explore radiomics features from longitudinal diffusion-weighted MRI (DWI) for pathologic treatment effect prediction in patients with localized soft tissue sarcoma (STS) undergoing hypofractionated preoperative radiotherapy (RT). Methods: Thirty patients with localized STS treated with preoperative hypofractionated RT were recruited to this longitudinal imaging study. DWI were acquired at three time points using a 0.35T MRI-guided radiotherapy system. Treatment effect score (TES) was obtained from the post-surgery pathology as a surrogate of treatment outcome. Patients were divided into two groups based on TES. Response prediction was first performed using support vector machine (SVM) with only mean apparent diffusion coefficient (ADC) or delta ADC to serve as the benchmark. Radiomics features were then extracted from tumor ADC maps at each of the three time points. Logistic regression and SVM were constructed to predict the TES group using features selected by univariate analysis and sequential forward selection. Classification performance using SVM with features from different time points and with or without delta radiomics were evaluated. Results: Prediction performance using only mean ADC or delta ADC was poor (AUC<0.7). For the radiomics study using features from all time points and corresponding delta radiomics, SVM significantly outperformed logistic regression (AUC of 0.91±0.05 v.s. 0.85±0.06). Prediction AUC using single or multiple time points without delta radiomics were all below 0.74. Including delta radiomics of mid-or post-treatment relative to the baseline drastically boosted the prediction. Conclusion: An SVM model was built to predict treatment effect score using radiomics features from longitudinal DWI. Based on this study, we found that mean ADC, or delta ADC, or radiomics features alone was not sufficient for response prediction, and including delta radiomics features of mid-or post-treatment relative to the baseline can optimize the prediction of treatment effect score, a pathologic and clinical endpoint.
Purpose
To compare the accuracy and precision of four different T1 estimation algorithms for MOLLI.
Methods
Four T1 estimation algorithms, including the original fit, Inversion group(IG) fit, Instantaneous signal loss simulation(InSiL) and Bloch equation simulation with slice profile correction(BLESSPC) were studied. T1 estimation accuracy, precision, reproducibility and sensitivity to heart rate(HR), flip angle(FA) and acquisition scheme(AcS) variations were compared in simulation, phantom, and volunteer studies.
Results
T1 estimation accuracy of IG (−2.4%±3.9%) and original fit (−3.2%±1.4%) were worse than BLESSPC (0.2%±1.5%) and InSiL (−0.7%±2.1%). The original fit had the best precision for T1 from 409ms-1884ms for the same FA (0.67%±0.16% vs. 0.90%±0.23% using IG, 0.78%±0.11% using InSiL, 0.77%±0.12% using BLESSPC). BLESSPC generated the most consistent in vivo T1 values over different FAs and AcS, and the T1 estimation reproducibility was similar (p>0.3) among the four methods when FA=35°. When using FA=50°, the reproducibility was significantly improved only when using BLESSPC (1.6%±0.9 vs. 2.6%±1.9%, p<0.05).
Conclusion
BLESSPC has superior accuracy and is the least sensitive to FA, HR, and AcS variations. T1 estimation using BLESSPC and FA=50° is superior to conventional MOLLI with FA=35° in accuracy and precision. Further clinical studies in varying pathological conditions are warranted to confirm our findings.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.