Domain generalization in deep learning for contrast-enhanced imaging.

BACKGROUND: The domain generalization problem has been widely investigated in deep learning for non-contrast imaging over the last years, but it received limited attention for contrast-enhanced imaging. However, there are marked differences in contrast imaging protocols across clinical centers, in particular in the time between contrast injection and image acquisition, while access to multi-center contrast-enhanced image data is limited compared to available datasets for non-contrast imaging. This calls for new tools for generalizing single-domain, single-center deep learning models across new unseen domains and clinical centers in contrast-enhanced imaging. METHODS: In this paper, we present an exhaustive evaluation of deep learning techniques to achieve generalizability to unseen clinical centers for contrast-enhanced image segmentation. To this end, several techniques are investigated, optimized and systematically evaluated, including data augmentation, domain mixing, transfer learning and domain adaptation. To demonstrate the potential of domain generalization for contrast-enhanced imaging, the methods are evaluated for ventricular segmentation in contrast-enhanced cardiac magnetic resonance imaging (MRI). RESULTS: The results are obtained based on a multi-center cardiac contrast-enhanced MRI dataset acquired in four hospitals located in three countries (France, Spain and China). They show that the combination of data augmentation and transfer learning can lead to single-center models that generalize well to new clinical centers not included during training. CONCLUSIONS: Single-domain neural networks enriched with suitable generalization procedures can reach and even surpass the performance of multi-center, multi-vendor models in contrast-enhanced imaging, hence eliminating the need for comprehensive multi-center datasets to train generalizable models.

Minimising multi-centre radiomics variability through image normalisation: a pilot study.

Radiomics is an emerging technique for the quantification of imaging data that has recently shown great promise for deeper phenotyping of cardiovascular disease. Thus far, the technique has been mostly applied in single-centre studies. However, one of the main difficulties in multi-centre imaging studies is the inherent variability of image characteristics due to centre differences. In this paper, a comprehensive analysis of radiomics variability under several image- and feature-based normalisation techniques was conducted using a multi-centre cardiovascular magnetic resonance dataset. 218 subjects divided into healthy (n = 112) and hypertrophic cardiomyopathy (n = 106, HCM) groups from five different centres were considered. First and second order texture radiomic features were extracted from three regions of interest, namely the left and right ventricular cavities and the left ventricular myocardium. Two methods were used to assess features' variability. First, feature distributions were compared across centres to obtain a distribution similarity index. Second, two classification tasks were proposed to assess: (1) the amount of centre-related information encoded in normalised features (centre identification) and (2) the generalisation ability for a classification model when trained on these features (healthy versus HCM classification). The results showed that the feature-based harmonisation technique ComBat is able to remove the variability introduced by centre information from radiomic features, at the expense of slightly degrading classification performance. Piecewise linear histogram matching normalisation gave features with greater generalisation ability for classification ( balanced accuracy in between 0.78 ± 0.08 and 0.79 ± 0.09). Models trained with features from images without normalisation showed the worst performance overall ( balanced accuracy in between 0.45 ± 0.28 and 0.60 ± 0.22). In conclusion, centre-related information removal did not imply good generalisation ability for classification.

Usefulness of Cardiac Computed Tomography in Coronary Risk Prediction: A Five-Year Follow-Up of the SPICA Study (Secure Prevention with Imaging of the Coronary Arteries).

Accurate identification of individuals at high coronary risk would reduce acute coronary syndrome incidence and morbi-mortality. We analyzed the effect on coronary risk prediction of adding coronary artery calcification (CAC) and Segment Involvement Score (SIS) to cardiovascular risk factors. This was a prospective cohort study of asymptomatic patients recruited between 2013-2017. All participants underwent a coronary computed tomography angiography to determine CAC and SIS. The cohort was followed-up for a composite endpoint of myocardial infarction, coronary angiography and/or revascularization (median = five years). Discrimination and reclassification of the REGICOR function with CAC/SIS were examined with the Sommer's D index and with the Net reclassification index (NRI). Nine of the 251 individuals included had an event. Of the included participants, 94 had a CAC = 0 and 85 a SIS = 0, none of them had an event. The addition of SIS or of SIS and CAC to the REGICOR risk function significantly increased the discrimination capacity from 0.74 to 0.89. Reclassification improved significantly when SIS or both scores were included. CAC and SIS were associated with five-year coronary event incidence, independently of cardiovascular risk factors. Discrimination and reclassification of the REGICOR risk function were significantly improved by both indexes, but SIS overrode the effect of CAC.

Multi-Centre, Multi-Vendor and Multi-Disease Cardiac Segmentation: The M&Ms Challenge.

The emergence of deep learning has considerably advanced the state-of-the-art in cardiac magnetic resonance (CMR) segmentation. Many techniques have been proposed over the last few years, bringing the accuracy of automated segmentation close to human performance. However, these models have been all too often trained and validated using cardiac imaging samples from single clinical centres or homogeneous imaging protocols. This has prevented the development and validation of models that are generalizable across different clinical centres, imaging conditions or scanner vendors. To promote further research and scientific benchmarking in the field of generalizable deep learning for cardiac segmentation, this paper presents the results of the Multi-Centre, Multi-Vendor and Multi-Disease Cardiac Segmentation (M&Ms) Challenge, which was recently organized as part of the MICCAI 2020 Conference. A total of 14 teams submitted different solutions to the problem, combining various baseline models, data augmentation strategies, and domain adaptation techniques. The obtained results indicate the importance of intensity-driven data augmentation, as well as the need for further research to improve generalizability towards unseen scanner vendors or new imaging protocols. Furthermore, we present a new resource of 375 heterogeneous CMR datasets acquired by using four different scanner vendors in six hospitals and three different countries (Spain, Canada and Germany), which we provide as open-access for the community to enable future research in the field.

Utilidad diagnóstica del mapeo T 1 por RM cardiaca en la hipertrofia ventricular / Diagnostic usefulness of cardiac MRI T 1 mapping in ventricular hypertrophy

No disponible

Relation of ST-segment elevation before and after percutaneous transluminal coronary angioplasty to left ventricular area at risk, myocardial infarct size, and systolic function.

Electrocardiography is an excellent tool for decision making in patients with ST elevation myocardial infarction (STEMI). However, little is known on the correlation between its dynamic changes during primary percutaneous coronary intervention (PCI) and the anatomic information provided by cardiovascular magnetic resonance. The study aimed to assess the predictive value of dynamic ST-segment changes before and after PCI on myocardial area at risk (AAR), infarct size, and left ventricular function in patients with STEMI. Eighty-five consecutive patients with a first STEMI were included. An electrocardiogram was recorded before and after PCI at 1, 24, 48, 72, and 120 hours. Sum of ST elevation (sumSTE), the number of STE, and STE resolution (resSTE) were determined. Complete resSTE was defined as ≥70% resolution, and patients were classified into 3 groups: group 1 (resSTE 1 hour after PCI) n = 39; group 2 (resSTE 120 hour after PCI) n = 27; and group 3, without resSTE (n = 19). Cardiovascular magnetic resonance was performed during hospitalization and at 6 months. Left ventricular volumes, ejection fraction, AAR, infarct size, myocardial salvage index, and microvascular obstruction were determined. Before PCI, the number of STE and sumSTE were best associated with AAR (p <0.001). After PCI, lack of resSTE (group 3) was associated with larger infarct size, MVO, and lower myocardial salvage index. However, sumSTE at 120 hours after PCI best discriminated patients with larger infarct size, ventricular volumes, and lower ejection fraction during hospitalization and at follow-up. In conclusion, admission sumSTE best correlates with AAR, whereas sumSTE at 120 hours rather than early resSTE best correlates with infarct size and left ventricular volumes during hospitalization and at 6 months.

Enfermedad coronaria subclínica por tomografía computarizada multidetector en población asintomática estratificada por nivel de riesgo coronario / Subclinical coronary atherosclerosis identified by coronary computed tomography angiography in asymptomatic population by coronary artery disease risk level

No disponible

Cuantificación del área miocárdica en riesgo: validación de puntuaciones angiográficas coronarias con métodos de resonancia magnética cardiovascular / Quantification of myocardial area at risk: validation of coronary angiographic scores with cardiovascular magnetic resonance methods

Introducción y objetivos. La cuantificación del área miocárdica en riesgo tras el infarto agudo de miocardio tiene repercusiones clínicas importantes y puede determinarse mediante resonancia magnética cardiovascular. Las puntuaciones angiográficas Bypass Angioplasty Revascularization Investigation Myocardial Jeopardy Index (BARI) y Alberta Provincial Project for Outcome Assessment in Coronary Heart Disease (APPROACH) se han utilizado ampliamente para la estimación rápida del área miocárdica en riesgo, pero no han sido validadas de manera directa. Nuestro objetivo es comparar el área miocárdica en riesgo estimada mediante las puntuaciones angiográficas BARI y APPROACH con la determinada mediante resonancia magnética cardiovascular. Métodos. En un estudio prospectivo, en la primera semana siguiente a la intervención coronaria percutánea, se realizaron exploraciones de resonancia magnética cardiovascular a 70 pacientes con un primer infarto agudo de miocardio con elevación del segmento ST reperfundido con éxito. El área miocárdica en riesgo se determinó mediante el análisis de secuencias T2-short tau inversion recovery y el cálculo del área endocárdica con infarto utilizando secuencias de contraste tardío. Estos resultados se compararon con los de las puntuaciones BARI y APPROACH. Resultados. Las puntuaciones BARI y APPROACH mostraron una correlación estadísticamente significativa con el T2-short tau inversion recovery para la estimación del área miocárdica en riesgo (BARI, coeficiente de correlación intraclase=0,72; p<0,001; APPROACH, coeficiente de correlación intraclase=0,69; p<0,001). Se observaron correlaciones mejores para el infarto agudo de miocardio de cara anterior que para otras localizaciones (BARI, coeficiente de correlación intraclase, 0,73 frente a 0,63; APPROACH, coeficiente de correlación intraclase, 0,68 frente a 0,50). El área de superficie endocárdica con infarto mostró buena correlación con ambas puntuaciones angiográficas (con BARI, coeficiente de correlación intraclase=0,72; p<0,001; con APPROACH, coeficiente de correlación intraclase=0,70; p<0,001). Conclusiones. Las puntuaciones angiográficas BARI y APPROACH permiten obtener una estimación fiable del área miocárdica en riesgo en la práctica clínica actual, sobre todo en los infartos de cara anterior (AU)

Introduction and objectives. Quantification of myocardial area-at-risk after acute myocardial infarction has major clinical implications and can be determined by cardiovascular magnetic resonance. The Bypass Angioplasty Revascularization Investigation Myocardial Jeopardy Index (BARI) and Alberta Provincial Project for Outcome Assessment in Coronary Heart Disease (APPROACH) angiographic scores have been widely used for rapid myocardial area-at-risk estimation but have not been directly validated. Our objective was to compare the myocardial area-at-risk estimated by BARI and APPROACH angiographic scores with those determined by cardiovascular magnetic resonance. Methods. In a prospective study, cardiovascular magnetic resonance was performed in 70 patients with a first successfully-reperfused ST-segment elevation acute myocardial infarction in the first week after percutaneous coronary intervention. Myocardial area-at-risk was obtained both by analysis of T2-short tau inversion recovery sequences and calculation of infarct endocardial surface area with late enhancement sequences. These results were compared with those of BARI and APPROACH scores. Results. BARI and APPROACH showed a statistically significant correlation with T2-short tau inversion recovery for myocardial area-at-risk estimation (BARI, intraclass correlation coefficient=0.72; P<.001; APPROACH, intraclass correlation coefficient=0.69; P<.001). Better correlations were observed for anterior acute myocardial infarction than for other locations (BARI, intraclass correlation coefficient=0.73 vs 0.63; APPROACH, intraclass correlation coefficient=0.68 vs 0.50). Infarct endocardial surface area showed a good correlation with both angiographic scores (BARI, intraclass correlation coefficient=0.72; P<.001; with APPROACH, intraclass correlation coefficient=0.70; P<.001). Conclusions. BARI and APPROACH angiographic scores allow reliable estimation of myocardial area-at-risk in current clinical practice, particularly in anterior infarctions (AU)

Quantification of myocardial area at risk: validation of coronary angiographic scores with cardiovascular magnetic resonance methods.

INTRODUCTION AND OBJECTIVES: Quantification of myocardial area-at-risk after acute myocardial infarction has major clinical implications and can be determined by cardiovascular magnetic resonance. The Bypass Angioplasty Revascularization Investigation Myocardial Jeopardy Index (BARI) and Alberta Provincial Project for Outcome Assessment in Coronary Heart Disease (APPROACH) angiographic scores have been widely used for rapid myocardial area-at-risk estimation but have not been directly validated. Our objective was to compare the myocardial area-at-risk estimated by BARI and APPROACH angiographic scores with those determined by cardiovascular magnetic resonance. METHODS: In a prospective study, cardiovascular magnetic resonance was performed in 70 patients with a first successfully-reperfused ST-segment elevation acute myocardial infarction in the first week after percutaneous coronary intervention. Myocardial area-at-risk was obtained both by analysis of T2-short tau inversion recovery sequences and calculation of infarct endocardial surface area with late enhancement sequences. These results were compared with those of BARI and APPROACH scores. RESULTS: BARI and APPROACH showed a statistically significant correlation with T2-short tau inversion recovery for myocardial area-at-risk estimation (BARI, intraclass correlation coefficient=0.72; P<.001; APPROACH, intraclass correlation coefficient=0.69; P<.001). Better correlations were observed for anterior acute myocardial infarction than for other locations (BARI, intraclass correlation coefficient=0.73 vs 0.63; APPROACH, intraclass correlation coefficient=0.68 vs 0.50). Infarct endocardial surface area showed a good correlation with both angiographic scores (BARI, intraclass correlation coefficient=0.72; P<.001; with APPROACH, intraclass correlation coefficient=0.70; P<.001). CONCLUSIONS: BARI and APPROACH angiographic scores allow reliable estimation of myocardial area-at-risk in current clinical practice, particularly in anterior infarctions. Full English text available from:www.revespcardiol.org.

Análisis simultáneo del strain 2D, de la reserva coronaria y de la contractilidad parietal durante el eco estrés con dipiridamol: Resultados comparativos / Simultaneous Analysis of Two-Dimensional Strain, Coronary Flow Reserve and Wall Motion during Dipyridamole Stress Echocardiography: Comparative Outcomes

Introducción El eco estrés con dipiridamol (ECODIP) es una prueba farmacológica aceptada para la evaluación de pacientes con cardiopatía isquémica; está demostrado que es una técnica ideal para combinar la información visual de la contractilidad con la reserva coronaria (RC), pero no se tiene información respecto del uso del strain 2D (deformación bidimensional) para el diagnóstico de isquemia miocárdica, por lo que su evaluación y la comparación de los resultados con otros métodos actuales para la determinación de isquemia regional puede ser de singular interés práctico. Objetivos Comparar, durante el ECODIP los resultados del análisis visual de la motilidad parietal con el valor de la RC de la arteria descendente anterior (ADA) y con el comportamiento del strain 2D longitudinal en el mismo territorio, utilizando la perfusión miocárdica con 99Tc sestamibi con SPECT simultáneo como patrón oro para el diagnóstico de isquemia. Material y métodos Se evaluaron 41 pacientes (16 hombres; edad promedio 68,5 años) derivados para ECODIP. Se promedió el strain 2D longitudinal obtenido en los 9 segmentos correspondientes a la ADA desde las tres vistas apicales tanto en reposo como en estrés, considerando como isquemia a la reducción en la deformación o un incremento menor del 5%. En el pico de efecto de 0,84 mg/kg de dipiridamol se evaluó también la motilidad parietal, la RC de la ADA (normal: > 2) y se inyectó 99Tc sestamibi para SPECT en estrés, completando su vista en reposo a las 24-48 horas. Resultados Se excluyeron 3 pacientes, uno por señal deficiente en la ADA y dos por ventana subóptima para el strain 2D. Doce pacientes con defectos reversibles con SPECT se consideraron isquémicos. Se observó concordancia de los cuatro parámetros evaluados en 6 pacientes con todos los estudios anormales y en 23 con todos los estudios normales. La deformación longitudinal en el territorio de la ADA en pacientes con SPECT normal fue de -19,9% (-18, -22) en reposo y de -22,5% (-21, -26) en el pico del ECODIP (p = 0,0003). En pacientes con SPECT isquémico, el strain 2D fue -19,35% (-17, -22) en reposo y -20,25% (-13, -21) en el pico del ECODIP (p = ns). La RC mostró una sensibilidad del 66,7% (IC 95% 35,4-88,7) y una especificidad del 100%; la contractilidad, una sensibilidad del 50% (IC 95% 22,3-77,7) y una especificidad del 100% y el strain 2D, una sensibilidad del 83,3% (IC 95% 50,9-97) y una especificidad del 88,5% (IC 95% 68,7-97). Conclusiones El strain 2D durante el eco estrés con dipiridamol resultó un método cuantitativo factible con una efectividad similar a la RC y superior al análisis visual de la contractilidad para detectar isquemia en el territorio de la ADA.

Background Dipyridamole stress echocardiography (DSE) is a pharmacological test accepted for the evaluation of patients with ischemic heart disease that has proved to be an ideal method that combines the visual information of wall motion with coronary flow reserve (CFR). However, the usefulness of 2D strain for the diagnosis of myocardial ischemia is still uncertain. In this sense, it seems interesting for daily practice to evalúate the results of this technique and compare them with those of other methods currently used for the evaluation of regional ischemia. Objectives To compare the results of visual analysis of wall motion and the valué of CFR of the left anterior descending (LAD) coronary artery during DSE with 2D longitudinal strain in the same territory using simultaneous 99mTc-sestamibi SPECT sean as the gold standard for the diagnosis of myocardial ischemia. Material and Methods We evaluated 41 patients (16 men; average age 68.5 years) referred for DSE. The average 2D longitudinal strain obtained of the 9 segments corresponding to the LAD coronary artery from the three apical views at rest and during stress was estimated. A drop in 2D strain valué or an average increase < 5% was considered an ischemic response. Wall motion and CFR in the LAD (normal: >2) were evaluated during the peak effect of 0.84 mg/kg of dipyridamole, when intravenous 99Tc sestamibi was injected and stress SPECT images were obtained. Rest images were acquired at 24 and 48 hours. Results Three patients were excluded: one due to poor pulsed Doppler signal in the LAD and two patients presented a suboptimal ultrasound window for 2D strain. Twelve patients presented reversible perfusión defeets in SPECT images (ischemic response). There was a good correlation in the four parameters evaluated in 6 patients in whom all the studies were abnormal and in 23 patients in whom all the studies were normal. In patients with normal SPECT images, 2D longitudinal strain in the territory of the LAD was -19.9% (-18, -22) at rest and -22.5% (-21, -26) during peak infusión of dipyridamole (p=0.0003). In patients with ischemia, 2D strain was -19.35% (-17, -22) at rest and -20.25% (-13, -21) during peak dipyridamole infusión (p=ns). The sensitivities and specificities obtained were: CFR, 66.7% (CI 95% 35.4-88.7) and 100%, respectively; wall motion, 50% (CI 95% 22.3-77.7) and 100%, respectively; and 2D strain, 83.3% (CI 95% 50.9-97) and 88.5% (CI 95% 68.7-97), respectively. Conclusions Simultaneous analysis of 2D strain during dipyridamole stress echocardiography proved to be a feasible quantitative method, as effective as CFR and better than the visual wall motion analysis for the diagnosis of ischemia in the territory of the LAD coronary artery.

Angioplastia coronaria en centros con residencia de cardiología en la Argentina: Estudio CONAREC XIV - Área de Investigación de la SAC / Coronary Angioplasty in Centers of Argentina with Residency in Cardiology: CONAREC XIV Study - Research Area of the Argentine Society of Cardiology (SAC)

Introducción Los nuevos tratamientos médicos y los avances técnicos, junto con la mayor experiencia adquirida en cardiología intervencionista, hicieron necesaria la realización de este nuevo registro, el protocolo CONAREC XIV, sobre empleo de angioplastia coronaria (ATC), un procedimiento que es seguro y eficaz para el tratamiento de la enfermedad coronaria. Objetivo Evaluar las características de los pacientes, las indicaciones y los resultados de la ATC en nuestro país. Material y métodos Se realizó un registro prospectivo y consecutivo durante 6 meses de pacientes tratados con ATC en centros con residencia de cardiología. Se determinaron antecedentes, cuadro clínico de ingreso, tratamiento, resultados y complicaciones intrahospitalarias. Resultados Se registraron 1.500 pacientes. La edad promedio fue de 62,8 ± 10,8 años y el 78,3% eran hombres. Antecedentes: 72% hipertensión arterial, 56,6% dislipidemia, 19,2% diabetes y 22,4% tabaquismo. Los cuadros clínicos de presentación fueron: 20% asintomáticos, 16,2% angina crónica estable, 45% síndrome coronario agudo sin supradesnivel del ST (SCA-SST), 19% síndrome coronario agudo con supradesnivel del ST (IAM-ST). En el 74,7% de los casos se realizó ATC de un vaso. Se utilizaron stents en el 94,5% de los casos y en el 18,7%, stents liberadores de drogas. El uso de pruebas funcionales previas a la ATC en cuadros estables fue del 53,9%, mientras que en el SCA-SST fue del 31,6%. La mediana de tiempo de evolución hasta la ATC en el SCA-SST fue de 1 día con un rango intercuartil 25-75% (RIC) de 0 a 3. En el IAM-ST, el tiempo puerta-balón fue de 60 minutos (RIC 40-105) y la mortalidad fue del 8%. Conclusiones La ATC se utiliza principalmente para el tratamiento de síndromes coronarios agudos. Se evidenció una tasa alta de uso de stents y de stents liberadores de drogas. El empleo de pruebas funcionales fue bajo. La tasa de complicaciones fue similar a la de los registros internacionales.

Introduction New medical therapies and technical advances, as well as the increased experience in interventional cardiology, made it necessary to carry out this new registry, the CONAREC XIV protocol, on the use of percutaneous coronary intervention (PCI), a procedure that is safe and effective for the treatment of coronary disease. Objective To evaluate the characteristics of patients, indications, and results of PCI in our country. Material and Methods A prospective consecutive registry was performed during 6 months in patients that underwent PCI at centers that had a Residency in Cardiology. Background, clinical condition upon admission, therapy, results and in-hospital complications were assessed. Results The recorded patients were 1,500. The average age was 62,8 ± 10,8 years and 78,3% were males. Background: 72% arterial hypertension, 56,6% dyslipemia, 19,2% diabetes and 22,4% smokers. Clinical conditions were: 20% asymptomatic, 16,2% stable chronic angina, 45% non ST segment elevation acute coronary syndrome (NSTE-ACS), 19% ST elevation acute coronary syndrome (STE-ACS). In 74,7% of the cases PCI was performed in one of the vessels. Stents were used in 94,5% of the cases and in 18,7%, drug eluting stents were used. The use of functional tests previous to PCI in stable patients was of 53,9%, whereas in STE-ACS was of 31,6%. The median evolution time up to PCI in the STEACS was 1 day with 25-75% (ICR) interquartile range of 0 to 3. In ST-AMI, the door-to-balloon time was 60 minutes (ICR 40-105) and mortality was 8%. Conclusions PCI is used mainly in the treatment of acute coronary syndromes. High use rate of stents and drug eluting stents was evident. The use of functional tests was low. The complications rate was similar to those shown in international registries.