scholarly journals Deep Learning Supplants Visual Analysis by Experienced Operators for the Diagnosis of Cardiac Amyloidosis by Cine-CMR

Diagnostics ◽  
2021 ◽  
Vol 12 (1) ◽  
pp. 69
Author(s):  
Philippe Germain ◽  
Armine Vardazaryan ◽  
Nicolas Padoy ◽  
Aissam Labani ◽  
Catherine Roy ◽  
...  
Keyword(s):  
Cardiac Amyloidosis ◽  
Visual Analysis ◽  
Area Under The Curve ◽  
Left Ventricular ◽  
Human Operators ◽  
Visual Diagnosis ◽  
Cine Cmr ◽  
Better Than ◽  
Unique Capability ◽  
Visual Geometry

Background: Diagnosing cardiac amyloidosis (CA) from cine-CMR (cardiac magnetic resonance) alone is not reliable. In this study, we tested if a convolutional neural network (CNN) could outperform the visual diagnosis of experienced operators. Method: 119 patients with cardiac amyloidosis and 122 patients with left ventricular hypertrophy (LVH) of other origins were retrospectively selected. Diastolic and systolic cine-CMR images were preprocessed and labeled. A dual-input visual geometry group (VGG ) model was used for binary image classification. All images belonging to the same patient were distributed in the same set. Accuracy and area under the curve (AUC) were calculated per frame and per patient from a 40% held-out test set. Results were compared to a visual analysis assessed by three experienced operators. Results: frame-based comparisons between humans and a CNN provided an accuracy of 0.605 vs. 0.746 (p < 0.0008) and an AUC of 0.630 vs. 0.824 (p < 0.0001). Patient-based comparisons provided an accuracy of 0.660 vs. 0.825 (p < 0.008) and an AUC of 0.727 vs. 0.895 (p < 0.002). Conclusion: based on cine-CMR images alone, a CNN is able to discriminate cardiac amyloidosis from LVH of other origins better than experienced human operators (15 to 20 points more in absolute value for accuracy and AUC), demonstrating a unique capability to identify what the eyes cannot see through classical radiological analysis.

scholarly journals Can myocardial work indices contribute to the exploration of patients with cardiac amyloidosis?

Open Heart ◽  
2020 ◽  
Vol 7 (2) ◽  
pp. e001346
Author(s):  
Aénora Roger-Rollé ◽  
Eve Cariou ◽  
Khailène Rguez ◽  
Pauline Fournier ◽  
Yoan Lavie-Badie ◽  
...  
Keyword(s):  
Cardiac Amyloidosis ◽  
Characteristic Curve ◽  
Global Longitudinal Strain ◽  
Left Ventricular ◽  
Peak Oxygen Consumption ◽  
Left Ventricular Ejection ◽  
Ventricular Ejection Fraction ◽  
Echocardiographic Parameters ◽  
Global Work ◽  
Better Than

BackgroundCardiac amyloidosis (CA) is a life-threatening restrictive cardiomyopathy. Identifying patients with a poor prognosis is essential to ensure appropriate care. The aim of this study was to compare myocardial work (MW) indices with standard echocardiographic parameters in predicting mortality among patients with CA.MethodsClinical, biological and transthoracic echocardiographic parameters were retrospectively compared among 118 patients with CA. Global work index (GWI) was calculated as the area of left ventricular pressure–strain loop. Global work efficiency (GWE) was defined as percentage ratio of constructive work to sum of constructive and wasted works. Sixty-one (52%) patients performed a cardiopulmonary exercise.ResultsGWI, GWE, global longitudinal strain (GLS), left ventricular ejection fraction (LVEF) and myocardial contraction fraction (MCF) were correlated with N-terminal prohormone brain natriuretic peptide (R=−0.518, R=−0.383, R=−0.553, R=−0.382 and R=−0.336, respectively; p<0.001). GWI and GLS were correlated with peak oxygen consumption (R=0.359 and R=0.313, respectively; p<0.05). Twenty-eight (24%) patients died during a median follow-up of 11 (4–19) months. The best cut-off values to predict all-cause mortality for GWI, GWE, GLS, LVEF and MCF were 937 mm Hg/%, 89%, 10%, 52% and 15%, respectively. The area under the receiver operator characteristic curve of GWE, GLS, GWI, LVEF and MCF were 0.689, 0.631, 0.626, 0.511 and 0.504, respectively.ConclusionIn CA population, MW indices are well correlated with known prognosis markers and are better than LVEF and MCF in predicting mortality. However, MW does not perform better than GLS.

Phasic left atrial strain analysis to discriminate cardiac amyloidosis in patients with unclear thick heart pathology

2020 ◽  
Author(s):  
Anna Brand ◽  
David Frumkin ◽  
Anne Hübscher ◽  
Henryk Dreger ◽  
Karl Stangl ◽  
...  
Keyword(s):  
Diagnostic Accuracy ◽  
Left Atrial ◽  
Cardiac Amyloidosis ◽  
Area Under The Curve ◽  
Strain Analysis ◽  
Left Ventricular ◽  
Wall Thickening ◽  
Left Atrial Strain ◽  
Heart Pathology ◽  
Atrial Strain

Abstract Aims  Traditional echocardiographic parameters for the assessment of suspected cardiac amyloidosis (CA) are of limited diagnostic accuracy. We sought to explore differences and the discriminative value of phasic left atrial strain (LAS) reductions and of regional longitudinal left ventricular (LV) strain alterations (relative apical sparing; RELAPS) in CA and other causes of LV wall thickening (LVH). Methods and results  We included 54 patients with unclear LVH (mean septal diastolic wall thickness 17.8 ± 3.5 mm); CA was bioptically confirmed in 35 patients (8 mATTR, 6 wtATTR, 20 AL, and 1 AA amyloidosis) and LVH in 19 subjects. We analysed RELAPS as well as LA reservoir (LASr), conduit (LAScd), and contraction strain (LASct) using 2D speckle tracking echocardiography (EchoPAC software, GE). RELAPS was higher (1.37 ± 0.94 vs. 0.86 ± 0.29, P &lt; 0.007), whereas atrial mechanics were significantly reduced in CA (LASr, LAScd, and LASct: 9.7 ± 5.2%, −6.5 ± 3.5%, and −5.0 ± 4.1% in CA; and 22.7 ± 7.8%, −13.9 ± 5.2%, and −13.0 ± 5.5% in LVH, respectively; P &lt; 0.001 each). With an area under the curve (AUC) of 0.91 [95% confidence interval (CI) 0.82–0.99], LASr showed a higher diagnostic accuracy in discriminating CA than RELAPS (AUC 0.74, 95% CI 0.59–0.88). LASr and LAScd remained significantly associated with CA in a multivariate regression model. Conclusion  Phasic LAS was significantly reduced in patients with CA and showed a higher diagnostic accuracy in discriminating CA than RELAPS. The additional assessment of phasic LAS may be useful to rule in the possible diagnosis of CA in patients with unclear LVH.

scholarly journals 70 Deep learning to diagnose cardiac amyloidosis from cardiac magnetic resonance findings

2020 ◽  
Vol 22 (Supplement_N) ◽  
pp. N116-N130
Author(s):  
Alberto Aimo ◽  
Nicola Martini ◽  
Andrea Barison ◽  
Daniele Della Latta ◽  
Giuseppe Vergaro ◽  
...  
Keyword(s):  
Deep Learning ◽  
Cardiac Magnetic Resonance ◽  
Magnetic Resonance ◽  
Cardiac Amyloidosis ◽  
Diagnostic Yield ◽  
Area Under The Curve ◽  
Blood Pool ◽  
Left Ventricular ◽  
Blood Dyscrasia ◽  
Test Accuracy

Abstract Aims Cardiac magnetic resonance (CMR) is part of the diagnostic work-up for cardiac amyloidosis (CA). Deep learning (DL) is an application of artificial intelligence that may allow to automatically analyze CMR findings and establish the likelihood of CA. Methods and results 1.5 T CMR was performed in 187 subjects with suspected CA (n = 92, 49% with unexplained left ventricular—LV—hypertrophy; n = 95, 51% with blood dyscrasia and suspected light-chain amyloidosis). Patients were randomly assigned to the training (n = 121, 65%), validation (n = 28, 15%), and testing subgroups (n = 38, 20%). Short axis (SA), 2-chamber (2 C), 4-chamber (4 C) late gadolinium enhancement (LGE) images were evaluated by 3 networks (DL algorithms). The tags “amyloidosis present” or “absent” were attributed when the average probability of CA from the 3 networks was ≥50% or &lt; 50%, respectively. The DL strategy was compared to a machine learning (ML) algorithm considering all manually extracted features (LV volumes, mass and function, LGE pattern, early blood-pool darkening, pericardial and pleural effusion, etc.), to reproduce exam reading by an experienced operator. The DL strategy displayed good diagnostic accuracy (84%), with an area under the curve (AUC) of 0.96. The precision (positive predictive value), recall score (sensitivity), and F1 score (a measure of test accuracy) were 78%, 94%, and 86% respectively. A ML algorithm considering all CMR features had a similar diagnostic yield to DL strategy (AUC 0.93 vs. 0.96; p = 0.45). Conclusion A DL approach evaluating LGE acquisitions displayed a similar diagnostic performance for CA to a ML-based approach, which simulates CMR reading by experienced operators.

3078Usefulness of a semi-quantitative and layer-specific assessment of the relative apical sparing pattern of longitudinal strain for the identification of cardiac amyloidosis

2019 ◽  
Vol 40 (Supplement_1) ◽  
Author(s):  
M Saito ◽  
M Imai ◽  
D Wake ◽  
R Higaki ◽  
T Sumimoto ◽  
...  
Keyword(s):  
Cardiac Amyloidosis ◽  
Longitudinal Strain ◽  
Global Longitudinal Strain ◽  
Hypertensive Heart Disease ◽  
Left Ventricular ◽  
Discriminative Ability ◽  
Specific Analysis ◽  
Definition Of ◽  
Myocardial Layer ◽  
Better Than

Abstract Background The relative apical sparing pattern (RASP) of left ventricular longitudinal strain (LS) is determined using a strain polar map, while global longitudinal strain is measured using speckle-tracking echocardiography, and it is frequently associated with cardiac amyloidosis (CA). However, the definition of visual RASP is ambiguous, and this leads to insufficient reproducibility, whereas quantitative RASP takes time and leads to difficulty in the clinical application. Generally, amyloid predominantly accumulates in the endo-myocardial layer. As such, layer-specific analysis of RASP may more accurately identify CA. Therefore, the aims of this study were to explore the reproducible and easy definition of RASP for identifying CA and investigate the effect of layer-specific analysis on the assessment. Methods A total of 40 patients with CA diagnosed by biopsy and technetium pyrophosphate scintigraphy were compared with 120 control patients matched for mean left ventricular wall thickness (40 aortic stenosis, 40 hypertrophic cardiomyopathy, and 40 hypertensive heart disease). We compared the discriminative abilities of three definitions of RASP (visual, quantitative, and semi-quantitative). According to a previous paper, visual RASP was defined as visual reduction of LS in the basal and middle LS segments (light red or blue) relative to the apical LS (red). Quantitative RASP was calculated using the following formula: average apical LS/(average basal LS + average mid-ventricle LS), then binarized by the optimal cut-off value for predicting CA. Semi-quantitative RASP was defined as reduction of LS (≥-10%) in five or more segments out of the basal six segments, relative to apical LS (≤-15%). Sample cases are shown in Figure (left). Visual and semi-quantitative RASP were independently assessed by two blinded sonographers. The RASP at the endo-myocardial and all layers was evaluated using customized software. The concordance was assessed using the kappa statistic, whereas the discriminative ability was assessed using receiver operating characteristic curve analysis. Results The concordance of visual RASP was modest but its semi-quantitative RASP was perfect (Table right). The discriminative ability of semi-quantitative RASP at each layer was significantly better than that of visual RASP and close to that of the binary quantitative RASP. Additionally, the discriminative abilities of visual (p=0.10) and semi-quantitative (p=0.11) RASP at the endo-myocardial layer appeared to be better than those at all layers. Conclusions The assessment method of semi-quantitative RASP is easy and highly reproducible. Furthermore, it accurately discriminates CA. In addition, assessment at the endo-myocardial layer potentially improves the discriminative ability.

scholarly journals Deep learning to diagnose cardiac amyloidosis from cardiovascular magnetic resonance

2020 ◽  
Vol 22 (1) ◽  
Author(s):  
Nicola Martini ◽  
Alberto Aimo ◽  
Andrea Barison ◽  
Daniele Della Latta ◽  
Giuseppe Vergaro ◽  
...  
Keyword(s):  
Cardiovascular Magnetic Resonance ◽  
Deep Learning ◽  
Magnetic Resonance ◽  
Cardiac Amyloidosis ◽  
Diagnostic Yield ◽  
Area Under The Curve ◽  
Blood Pool ◽  
Left Ventricular ◽  
Blood Dyscrasia ◽  
Test Accuracy

Abstract Background Cardiovascular magnetic resonance (CMR) is part of the diagnostic work-up for cardiac amyloidosis (CA). Deep learning (DL) is an application of artificial intelligence that may allow to automatically analyze CMR findings and establish the likelihood of CA. Methods 1.5 T CMR was performed in 206 subjects with suspected CA (n = 100, 49% with unexplained left ventricular (LV) hypertrophy; n = 106, 51% with blood dyscrasia and suspected light-chain amyloidosis). Patients were randomly assigned to the training (n = 134, 65%), validation (n = 30, 15%), and testing subgroups (n = 42, 20%). Short axis, 2-chamber, 4-chamber late gadolinium enhancement (LGE) images were evaluated by 3 networks (DL algorithms). The tags “amyloidosis present” or “absent” were attributed when the average probability of CA from the 3 networks was ≥ 50% or < 50%, respectively. The DL strategy was compared to a machine learning (ML) algorithm considering all manually extracted features (LV volumes, mass and function, LGE pattern, early blood-pool darkening, pericardial and pleural effusion, etc.), to reproduce exam reading by an experienced operator. Results The DL strategy displayed good diagnostic accuracy (88%), with an area under the curve (AUC) of 0.982. The precision (positive predictive value), recall score (sensitivity), and F1 score (a measure of test accuracy) were 83%, 95%, and 89% respectively. A ML algorithm considering all CMR features had a similar diagnostic yield to DL strategy (AUC 0.952 vs. 0.982; p = 0.39). Conclusions A DL approach evaluating LGE acquisitions displayed a similar diagnostic performance for CA to a ML-based approach, which simulates CMR reading by experienced operators.

scholarly journals Are there predictor variables for the diagnosis of transthyretin cardiac amyloidosis?

2021 ◽  
Vol 42 (Supplement_1) ◽  
Author(s):  
C Goena ◽  
X Arana ◽  
I Villanueva ◽  
I Solla ◽  
A Rengel ◽  
...  
Keyword(s):  
Logistic Regression ◽  
Cardiac Amyloidosis ◽  
Antihypertensive Drugs ◽  
Area Under The Curve ◽  
Left Ventricular ◽  
Receiver Operating Curve ◽  
Multivariable Logistic Regression Analysis ◽  
Al Amyloidosis ◽  
Funding Sources ◽  
Multivariable Logistic Regression

Abstract Introduction Transthyretin cardiac amyloidosis (ATTR) can be reliably diagnosed in the absence of histology if grade 2 or 3 cardiac uptake is demonstrated on 99mTc-DPD scan (DPD) in the absence of a detectable monoclonal component. Diagnosis requires a high degree of clinical suspicion in the presence of often non-specific findings and that it may be one of the reasons to under-diagnose ATTR. The aim of the study is to identify clinical, analytical and ECG variables that best predict a positive DPD result. Methods This is a multicentre retrospective study including all patients undergoing consecutive 99mTc-DPD scintigraphy in a reference area of 750,000 inhabitants between January 2016 and January 2021 for suspected ATTR. AL amyloidosis patients were excluded. Clinical, analytical, ECG and echocardiographic data were analyzed. We identified variables that independently predicted a positive DPD study using a multivariable logistic regression analysis. Receiver Operating Curve (ROC) analysis and the Area under the Curve (AUC) were calculated to assess the discrimination capacity of the model to predict a positive DPD study. Results DPD scans from a total of 181 patients were analyzed. Mean age of the sample: 78 years (42–96), 100% caucasians, 77% male. 54.7% (N=99) had a positive DPD study (defined as grade 2 or 3 Perugini uptake) and 45.3% (N=82) were negative. Independent predictors of a positive study were age, male gender, left ventricular septum thickness, any grade of atrioventricular block, low QRS voltage, Carpal tunnel syndrome, history of hypotension or need to lower antihypertensive drugs and a NT-proBNP value above 1800 pg/ml (See Table 1). The diagnostic accuracy of the model was excellent, with an AUC of 0.92 (IC 95% 0.87–0.96) (see Figure 1). Conclusions There are clinical-analytical factors and ECG and echocardiogram findings accessible in daily clinical practice that are able to predict a positive result on cardiac scintigraphy requested for suspected ATTR. Identifying these factors may improve the non-invasive diagnosis of ATTR. FUNDunding Acknowledgement Type of funding sources: None. Table 1. Multivariable logistic regression analys Figure 1. ROC curve

scholarly journals Cardiac Amyloidosis with Discordant QRS Voltage between Frontal and Precordial Leads

Medicina ◽  
2021 ◽  
Vol 57 (7) ◽  
pp. 660
Author(s):  
Csilla-Andrea Eötvös ◽  
Roxana-Daiana Lazar ◽  
Iulia-Georgiana Zehan ◽  
Erna-Brigitta Lévay-Hail ◽  
Giorgia Pastiu ◽  
...  
Keyword(s):  
Left Ventricular Hypertrophy ◽  
Cardiac Amyloidosis ◽  
Ventricular Hypertrophy ◽  
Low Voltage ◽  
Disease Process ◽  
Left Ventricular ◽  
Clinical Suspicion ◽  
Immunoglobulin Light Chain ◽  
Different Types

Among the different types, immunoglobulin light chain (AL) cardiac amyloidosis is associated with the highest morbidity and mortality. The outcome, however, is significantly better when an early diagnosis is made and treatment initiated promptly. We present a case of cardiac amyloidosis with left ventricular hypertrophy criteria on the electrocardiogram. After 9 months of follow-up, the patient developed low voltage in the limb leads, while still maintaining the Cornell criteria for left ventricular hypertrophy as well. The relative apical sparing by the disease process, as well as decreased cancellation of the opposing left ventricular walls could be responsible for this phenomenon. The discordance between the voltage in the frontal leads and precordial leads, when present in conjunction with other findings, may be helpful in raising the clinical suspicion of cardiac amyloidosis.

scholarly journals External assessment of the EUROMACS right-sided heart failure risk score

2021 ◽  
Vol 11 (1) ◽  
Author(s):  
Hirak Shah ◽  
Thomas Murray ◽  
Jessica Schultz ◽  
Ranjit John ◽  
Cindy M. Martin ◽  
...  
Keyword(s):  
Heart Failure ◽  
Right Ventricular ◽  
Risk Score ◽  
Pressure Ratio ◽  
Area Under The Curve ◽  
Left Ventricular ◽  
Right Atrial ◽  
Risk Scores ◽  
External Assessment ◽  
Failure Risk

AbstractThe EUROMACS Right-Sided Heart Failure Risk Score was developed to predict right ventricular failure (RVF) after left ventricular assist device (LVAD) placement. The predictive ability of the EUROMACS score has not been tested in other cohorts. We performed a single center analysis of a continuous-flow (CF) LVAD cohort (n = 254) where we calculated EUROMACS risk scores and assessed for right ventricular heart failure after LVAD implantation. Thirty-nine percent of patients (100/254) had post-operative RVF, of which 9% (23/254) required prolonged inotropic support and 5% (12/254) required RVAD placement. For patients who developed RVF after LVAD implantation, there was a 45% increase in the hazards of death on LVAD support (HR 1.45, 95% CI 0.98–2.2, p = 0.066). Two variables in the EUROMACS score (Hemoglobin and Right Atrial Pressure to Pulmonary Capillary Wedge Pressure ratio) were not predictive of RVF in our cohort. Overall, the EUROMACS score had poor external discrimination in our cohort with area under the curve of 58% (95% CI 52–66%). Further work is necessary to enhance our ability to predict RVF after LVAD implantation.

scholarly journals Study of NT-pro-BNP and Hs-Troponin I biomarkers for early detection of children’s heart function of protein-energy malnutrition

2019 ◽  
Vol 11 (2) ◽  
Author(s):  
Herlina Dimiati ◽  
Abdus Samik Wahab ◽  
Mohammad Juffrie ◽  
Madarina Julia ◽  
Basri A. Gani
Keyword(s):  
Troponin I ◽  
Heart Function ◽  
Protein Energy Malnutrition ◽  
Area Under The Curve ◽  
Left Ventricular ◽  
The Body ◽  
Chromatography Method ◽  
Blood Electrolytes ◽  
Ventricular Mass ◽  
Protein Energy

The Protein Energy Malnutrition (PEM) is the condition of a lack of carbohydrate and protein stores in the body that trigger chronic failure nutrient intake and body maintenance function caused to impact the heart functions. The NT-pro-BNP and Hs- Troponin I proteins were found as the indicator of cardiac dysfunction. The sixty subjects of PEM, analyzed by standard of Indonesia Healt Ministry as well as nutritional status. The blood electrolytes examined by laboratory assay and the levels of Hs-Troponin 1 and NT-Pro-BNP were analyzed by Immune-Chromatography method. Assessing of the ventricular mass with the seeing the peak of the diastolic flow rate of left ventricular that estimated by the curve of the receiver operating characteristic and the area under the curve (P<0.05). The result has shown that the PEM decreased in the left ventricular mass for impaired heart function and systolic disorder. The Hs- Troponin I (90.9%) has better sensitivity than NT-pro-BNP (85.5%) if the merger of those markers possesses the lowest sensitivity (81.8%). These proteins have good biomarkers in heart function, mainly in cases where PEM is present.

Sign in / Sign up

Export Citation Format

Share Document