Evin Ina Papalini1, Christian Polte2, and Kerstin Magdalena Lagerstrand1
1Institute of Clinical Sciences, Sahlgrenska Academy, University of Gothenburg, Gothenburg, Sweden, Gothenburg, Sweden, 2Institute of Medicine, Sahlgrenska Academy, University of Gothenburg, Gothenburg, Sweden, Gothenburg, Sweden
Synopsis
Myocarditis
is a common inflammatory disease in the myocardium, associated with acute heart
failure, chronic dilated cardiomyopathy and sudden cardiac death. Current
clinical diagnosis is based on magnetic resonance imaging, including
administration of gadolinium-based contrast agents. We propose that conventional balanced steady-state-free-precession magnetic resonance imaging images reveals
quantitative diagnostic features based on texture analysis. Our results showed
that the texture features, in specific Variance, Gradient Mean and
Sum Average, were able to significantly separate patients with and
without myocarditis using conventional balanced steady-state-free-precession
magnetic resonance imaging images.
Introduction
Myocarditis is a common inflammatory disease in the myocardium [1]. The diagnosis
remains challenging due to the large spectrum of underlying etiologies and a highly
varying clinical presentation. As
a result, the disease may clinically mimic many other cardiac diseases.
Although patients in the early stages of the disease usually resolves spontaneously,
the disease is associated with acute heart failure, chronic dilated
cardiomyopathy and sudden cardiac death. Thus, it is important to find reliable
diagnostic markers that can identify patients at risk.
The diagnosis relies on the combination of different methods including
clinical assessment and imaging techniques, such as magnetic resonance imaging
(MRI) [1] and is currently based on the Lake Louise Criteria (LLC), where one
of the conditions includes delayed contrast-enhanced MRI. However, intravenous
administration of gadolinium-based contrast agents in patients with renal
impairment are associated with risk of developing nephrogenic systemic fibrosis
[2], making non-contrast techniques for evaluation of myocarditis desirable.Purpose
This proof-of-concept study aimed to assess the diagnostic value of balanced
steady-state-free-precession (bSSFP) MR images using texture analysis (TA) in
patients with clinically suspected myocarditis.Methods
Twenty patients who had undergone a comprehensive cardiac MRI
examination due to clinically suspected myocarditis between 2013 and 2018 at
the Sahlgrenska University Hospital were included, where 10 myocarditis
patients (25±6 years) had clinical signs and positive myocardial biomarkers,
i.e. Troponin T, indicating myocardial injury, as well as positive MRI findings
according to LLC and 10 patients (45±15 years) had clinical signs but both
negative myocardial biomarkers as well as cardiac MRI findings according to LLC,
here called controls. TA was performed on regions of interest encompassing the
left ventricle, delineated on short axis bSSFP images (Figure 1) using a freely
available software package, MaZda [3-5]. 12 features were selected to assess
their diagnostic potential (Table 1).Results
When comparing the myocarditis patients and the controls, three texture
features in the bSSFP images showed statistically significant differences
between the groups (Variance: 236±185 vs 95.9±96.5, p=0.004; Gradient
Mean: 2.74±0.55 vs 3.43±0.37, p<0.001; Sum Average: 64.5±1.30 vs
62.8±0.88, p<0.001; Table 2, Figure 2).Discussion
Our results suggest that TA enable automated detection of myocarditis from
conventional bSSFP cardiac MRI, without the need of contrast agents. To our
knowledge, no other study has previously investigated the value of bSSFP
imaging with TA for the diagnosis of myocarditis.
Texture features that are characteristically for visible pathology, i.e.
Variance, Gradient Mean and Sum Average, were able to differ
between groups of patients with myocarditis and controls and showed high
separation between the individuals in the group, suggesting that these features
should be able to detect myocarditis on an individual level.
The Variance was derived from the histogram of an image and, thus,
measures the differences in tissue heterogeneity of the myocardium. Gradient
Mean was calculated as the spatial variation of the grey-level values in an
image and, thus, will change when pathology is visible or not. At last, Sum
Average was derived from the co-occurrence matrix that contains information
about the grey-level distribution of pairs of pixels and is therefore also is a
measure of heterogeneity, but more sensitive to focal changes.
Present study was limited in the number of included patients
with variation in demographic characteristics between the groups. We plan to
confirm these findings in a large cohort study including more texture features.Conclusion
Conventional bSSFP cardiac MRI reveals
clinically suspected myocarditis, where TA automatically can extract features
not always visible for the naked eye. Hence, this study emphasizes not only the
value of bSSFP imaging as a promising non-contrast-based tissue
characterization technique but also data-driven decision-based diagnostics for improved
sensitivity and specificity. However, this remains to be confirmed in a large
cohort study.Acknowledgements
References
- Cooper, L. T. Jr. Clinical
manifestations and diagnosis of myocarditis in adults. UpToDate. 2018.
Available at: https://www.uptodate.com/contents/clinical-manifestations-and-diagnosis-of-myocarditis-in-adults (Assessed 2020-09-15).
- Bettina, B. et al. Subacute
and Chronic Left Ventricular Myocardial Scar: Accuracy of Texture Analysis on
Nonenhanced Cine MR Images. Radiol. 2007; 286(1):103-112.
- Szczypinski, P., Strzelecki, M.,
Materka, A. & Klepaczko, A. MaZda-a software package for image texture
analysis. Comput Methods Programs Biomed. 2009; 94(1): 66-76.
- Szczypinski
P., Strzelecki M. & Materka A. MaZda - a software for texture analysis. Proc. of ISITC 2007. 2007; 245-249.
- Strzelecki, M., Szczypinski, P.,
Materka, A. & Klepaczko, A. A software tool for automatic classification
and segmentation of 2D/3D medical images. Nucl Instrum Methods Phys Res A.
2013; 702: 137-140.