Balancing Speed and Accuracy in Cardiac Magnetic Resonance Function Post-Processing: Comparing 2 Levels of Automation in 3 Vendors to Manual Assessment

Optimizing Deep Learning for Cardiac MRI Segmentation: The Impact of Automated Slice Range Classification

RATIONALE AND OBJECTIVES

Cardiac magnetic resonance imaging is crucial for diagnosing cardiovascular diseases, but lengthy postprocessing and manual segmentation can lead to observer bias. Deep learning (DL) has been proposed for automated cardiac segmentation; however, its effectiveness is limited by the slice range selection from base to apex.

MATERIALS AND METHODS

In this study, we integrated an automated slice range classification step to identify basal to apical short-axis slices before DL-based segmentation. We employed publicly available Multi-Disease, Multi-View & Multi-Center Right Ventricular Segmentation in Cardiac MRI data set with short-axis cine data from 160 training, 40 validation, and 160 testing cases. Three classification and seven segmentation DL models were studied. The top-performing segmentation model was assessed with and without the classification model. Model validation to compare automated and manual segmentation was performed using Dice score and Hausdorff distance and clinical indices (correlation score and Bland-Altman plots).

RESULTS

The combined classification (CBAM-integrated 2D-CNN) and segmentation model (2D-UNet with dilated convolution block) demonstrated superior performance, achieving Dice scores of 0.952 for left ventricle (LV), 0.933 for right ventricle (RV), and 0.875 for myocardium, compared to the stand-alone segmentation model (0.949 for LV, 0.925 for RV, and 0.867 for myocardium). Combined classification and segmentation model showed high correlation (0.92-0.99) with manual segmentation for biventricular volumes, ejection fraction, and myocardial mass. The mean absolute difference (2.8-8.3 mL) for clinical parameters between automated and manual segmentation was within the interobserver variability range, indicating comparable performance to manual annotation.

CONCLUSION

Integrating an initial automated slice range classification step into the segmentation process improves the performance of DL-based cardiac chamber segmentation.

Interpretation of pre-morbid cardiac 3T MRI findings in overweight and hypertensive young adults

In young adults, overweight and hypertension possibly already trigger cardiac remodeling as seen in mature adults, potentially overlapping non-ischemic cardiomyopathy findings. To this end, in young overweight and hypertensive adults, we aimed to investigate changes in left ventricular mass (LVM) and cardiac volumes, and the impact of different body scales for indexation. We also aimed to explore the presence of myocardial fibrosis, fat and edema, and changes in cellular mass with extracellular volume (ECV), T1 and T2 tissue characteristics. We prospectively recruited 126 asymptomatic subjects (51% male) aged 27-41 years for 3T cardiac magnetic resonance imaging: 40 controls, 40 overweight, 17 hypertensive and 29 hypertensive overweight. Myocyte mass was calculated as (100%-ECV) * height2.7-indexed LVM. Absolute LVM was significantly increased in overweight, hypertensive and hypertensive overweight groups (104 ± 23, 109 ± 27, 112 ± 26 g) versus controls (87 ± 21 g), with similar volumes. Body surface area (BSA) indexation resulted in LVM normalization in overweights (48 ± 8 g/m2) versus controls (47 ± 9 g/m2), but not in hypertensives (55 ± 9 g/m2) and hypertensive overweights (52 ± 9 g/m2). BSA-indexation overly decreased volumes in overweight versus normal-weight (LV end-diastolic volume; 80 ± 14 versus 92 ± 13 ml/m2), where height2.7-indexation did not. All risk groups had lower ECV (23 ± 2%, 23 ± 2%, 23 ± 3%) than controls (25 ± 2%) (P = 0.006, P = 0.113, P = 0.039), indicating increased myocyte mass (16.9 ± 2.7, 16.5 ± 2.3, 18.1 ± 3.5 versus 14.0 ± 2.9 g/m2.7). Native T1 values were similar. Lower T2 values in the hypertensive overweight group related to heart rate. In conclusion, BSA-indexation masks hypertrophy and causes volume overcorrection in overweight subjects compared to controls, height2.7-indexation therefore seems advisable.

Cardiac Alterations on 3T MRI in Young Adults With Sedentary Lifestyle-Related Risk Factors

Background

Young adult populations with the sedentary lifestyle-related risk factors overweight, hypertension, and type 2 diabetes (T2D) are growing, and associated cardiac alterations could overlap early findings in non-ischemic cardiomyopathy on cardiovascular MRI. We aimed to investigate cardiac morphology, function, and tissue characteristics for these cardiovascular risk factors.

Methods

Non-athletic non-smoking asymptomatic adults aged 18-45 years were prospectively recruited and underwent 3Tesla cardiac MRI. Multivariate linear regression was performed to investigate independent associations of risk factor-related parameters with cardiac MRI values.

Results

We included 311 adults (age, 32 ± 7 years; men, 49%). Of them, 220 subjects had one or multiple risk factors, while 91 subjects were free of risk factors. For overweight, increased body mass index (per SD = 5.3 kg/m2) was associated with increased left ventricular (LV) mass (+7.3 g), biventricular higher end-diastolic (LV, +8.6 ml), and stroke volumes (SV; +5.0 ml), higher native T1 (+7.3 ms), and lower extracellular volume (ECV, -0.38%), whereas the higher waist-hip ratio was associated with lower biventricular volumes. Regarding hypertension, increased systolic blood pressure (per SD = 14 mmHg) was associated with increased LV mass (+6.9 g), higher LV ejection fraction (EF; +1.0%), and lower ECV (-0.48%), whereas increased diastolic blood pressure was associated with lower LV EF. In T2D, increased HbA1c (per SD = 9.0 mmol/mol) was associated with increased LV mass (+2.2 g), higher right ventricular end-diastolic volume (+3.2 ml), and higher ECV (+0.27%). Increased heart rate was linked with decreased LV mass, lower biventricular volumes, and lower T2 values.

Conclusions

Young asymptomatic adults with overweight, hypertension, and T2D show subclinical alterations in cardiac morphology, function, and tissue characteristics. These alterations should be considered in cardiac MRI-based clinical decision making.