Multi-network approach for image segmentation in non-contrast enhanced cardiac 3D MRI of arrhythmic patients

Left atrial appendage (LAA) is the source of thrombi formation in more than 90% of strokes in patients with nonvalvular atrial fibrillation. Catheter-based LAA occlusion is being increasingly applied as a treatment strategy to prevent stroke. Anatomical complexity of LAA makes percutaneous occlusion commonly performed under transesophageal echocardiography (TEE) and X-ray (XR) guidance especially challenging. Image fusion techniques integrating 3D anatomical models derived from pre-procedural imaging into the live XR fluoroscopy can be applied to guide each step of the LAA closure. Cardiac magnetic resonance (CMR) imaging gains in importance for radiation-free evaluation of cardiac morphology as alternative to gold-standard TEE or computed tomography angiography (CTA). Manual delineation of cardiac structures from non-contrast enhanced CMR is, however, labor-intensive, tedious, and challenging due to the rather low contrast. Additionally, arrhythmia often impairs the image quality in ECG synchronized acquisitions causing blurring and motion artifacts. Thus, for cardiac segmentation in arrhythmic patients, there is a strong need for an automated image segmentation method. Deep learning-based methods have shown great promise in medical image analysis achieving superior performance in various imaging modalities and different clinical applications. Fully-convolutional neural networks (CNNs), especially U-Net, have become the method of choice for cardiac segmentation. In this paper, we propose an approach for automatic segmentation of cardiac structures from non-contrast enhanced CMR images of arrhythmic patients based on CNNs implemented in a multi-stage pipeline. Two-stage implementation allows subdividing the task into localization of the relevant cardiac structures and segmentation of these structures from the cropped sub-regions obtained from previous step leading to efficient and effective way of automated cardiac segmentation.

Distance between the Left Atrial Appendage and Mitral Annulus Evaluated by CARTO 3 Integrated Computed Tomography Imaging

OBJECTIVE

To measure distances between pulmonary veins (PV) and mitral annulus (MA) using angiographic computed tomography (CT) and to compare them with the left atrial appendage-MA (LAA-MA) line.

MATERIALS AND METHODS

Data from 46 catheter ablation procedures for atrial fibrillation involving 36 males, mean age 53 years, range 27-78 years, were analyzed. Three types of mitral isthmus lines were measured using angiographic CT images integrated in the CARTO 3 system (Biosense Webster): the distance between the right superior PV and MA (RSPV-MA), the right inferior PV and MA (RIPV-MA), and the left inferior PV and MA (LIPV-MA). They were compared with the length of the LAA-MA line.

RESULTS

The mean value of LIPV-MA was 29 ± 11.2 mm, RIPV-MA 39 ± 8.2 mm, and RSPV-MA 48 ± 8.2 mm. The circumflex artery (CxA) and the coronary sinus (CS) were closest to the LIPV-MA line. Compared with the three isthmus lines, the LAA-MA was the shortest (24.7 ± 15.6 mm), and the difference was statistically significant (p < 0.05).

CONCLUSION

The angiographic CT provided detailed information regarding the anatomy of the left atrium and distances between atrial structures. The LAA-MA was shorter than the other three lines with the CxA and CS situated at a distance.

Comparison of imaging techniques to assess appendage anatomy and measurements for left atrial appendage closure device selection

BACKGROUND

The adequate device size selection for left atrial appendage closure is crucial to ensuring adequate implantation and for avoiding the need for multiple attempts that increase the risk of complications. Our aim was to evaluate the information obtained using different imaging techniques to select the size of the closure device in a clinical environment.

METHODS

Thirty-seven patients who consecutively underwent implantation of Amplatzer cardiac plug (ACP) devices were studied. All patients were examined using computed tomography (CT) prior to intervention. Measurements were compared to those obtained using intraoperative transesophageal echocardiography (IOTEE) and angiography. Size was determined by the longest axis of the appendage ostium. The influence of all techniques on the correct selection of final size was assessed.

RESULTS

The measurements taken using the three techniques agreed in only 21.6% of the cases, leading to accurate selection of device size. Two techniques coincided as follows: IOTEE-CT in 45.9%, angiography-CT in 35.13%, and angiography-IOTEE in 24.3%. Measurements using CT were definitive for ACP selection in 75.7% of cases, angiography in 48.6%, and echocardiography in 51.4%. Device size was undermeasured with angiography in 35.1% of cases, and with IOTEE in 24.3%; CT overmeasured 21.6% of cases. The combination of angiography-CT was the most accurate for selection of device size.

CONCLUSION

CT most often predicts the appropriate device size. If it fails, it usually overestimates the size. Agreement of measurements with all three techniques is the most accurate situation; when two agree, the most accurate combination is angiography and CT.

Anatomical classification of left atrial appendages in specimens applicable to CT imaging techniques for implantation of amplatzer cardiac plug

BACKGROUND

Certain anatomical characteristics of the left atrial appendage (LAA) are associated with complexity in the implantation of occluder devices.

OBJECTIVE

The aim was to define characteristics measurable by three-dimensional imaging that would predict complexities both in the implantation procedure and the selection of the appropriate device size.

METHODS

An anatomical study was performed of 50 postmortem hearts, of which 15 had a history of atrial fibrillation, and of 30 consecutive patients undergoing LAA occlusion with the Amplatzer cardiac plug (ACP). The specimens were classified according to variables that can be visualized using computerized tomography (CT). The CT scans of 30 consecutive patients were classified according to the level of the LAA ostium, the left lateral ridge (LLR), the LAA limbus and distance from LAA to the mitral annulus before undergoing LAA occlusion, and the results were correlated.

RESULTS

Three types of LAA orifice were defined: type I, with a usually higher, anterior LAA ostium, a short, flattened and wide LLR and almost nonexistent limbus; type II, presenting a long, pointed and narrow LLR, and a longer, more defined limbus; type III, with a lower LAA ostium, close to the left atrium floor and the mitral annulus, a marked separation from the left pulmonary vein orifices and a limbus of intermediate length.

CONCLUSION

LAA with lower ostia are more difficult to occlude. Types II and III have very prominent LLRs with longer limbi, which may increase the difficulty of inserting the guide and making measurements for selection of the right ACP size.

'Coumadin ridge' in the left atrium demonstrated on three dimensional transthoracic echocardiography

An echogenic band like structure was seen in the left atrium on two dimensional transthoracic echocardiography (2D TTE). Full volume three dimensional (3D) TTE and colour Doppler established the surrounding anatomical landmarks, and demonstrated the absence of obstruction related to this band. 3D TTE confirmed that this band like structure was consistent with the ridge between the left atrial appendage and left superior pulmonary vein ('warfarin/coumadin ridge').

Implantation of Left Atrial Appendage Occlusion Devices and Complex Appendage Anatomy: The Importance of Transesophageal Echocardiography

Percutaneous delivery of left atrial appendage (LAA) occluding devices represents a novel approach for stroke prevention in patients with atrial fibrillation. Transesophageal echocardiography (TEE) has a pivotal role throughout these procedures, facilitating device size selection and ensuring optimal deployment of the device. We report a case of an LAA occluding device implantation in which apparent proper positioning on fluoroscopy was determined by TEE to be malpositioning with a nonocclusive, perpendicular orientation to the plane of the LAA ostium. This problem appeared to be related to a complex, multilobed LAA anatomy and was readily resolved by repositioning of the device under TEE guidance.

Morphologic Characteristics of the Left Atrial Appendage, Roof, and Septum: Implications for the Ablation of Atrial Fibrillation

INTRODUCTION

The left atrium (LA) ablation in different regions, including LA appendage (LAA), LA roof, and LA septum, has recently been proposed to improve the success rate of treating patients with atrial fibrillation (AF). The purpose of this study was to investigate the anatomy of LAA, LA roof, and LA septum, using computed tomography (CT).

METHODS AND RESULTS

Multidetector CT scan was used to depict the LA in 47 patients with drug-refractory paroxysmal AF (39 males, age = 50 +/- 12 years) and 49 control subjects (34 males, age = 54 +/- 11 years). The area of LAA orifice, neck, and the length of roof line were greater in AF group than in control subjects. Three types of LAA locations and two types of LAA ridges were observed. Higher incidence of inferior LAA was noted in AF patients. The different morphologies of LA roof were described. Roof pouches were revealed in 15% of AF and 14% of controls. Moreover, we found septal ridge in 32% of AF and 23% of controls.

CONCLUSIONS

Considerable variations of LAA and LA roof morphologies were demonstrated. Peculiar structures, including roof pouches and septal ridges, were delineated by CT imaging. These findings were important for determining the strategy of AF ablation and avoiding the procedure-related complications.

The Left Atrial Appendage, a Small, Blind-Ended Structure

The increasing prevalence of stroke and atrial fibrillation is a stimulus for new therapeutic strategies and also warrants a review of imaging modalities of the most important source of cardiac systemic embolic events: the left atrial appendage (LAA). This blind-ended, complex structure is embryologically distinct from the body of the left atrium and is sometimes regarded as just a minor extension of the atrium. However, it should routinely be analyzed as part of a transesophageal echocardiographic (TEE) examination. A pulsed Doppler TEE analysis of LAA emptying flow should supplement a two-dimensional (2-D) analysis; these examinations have proven to be highly reproducible and to help assess thromboembolic risk. In 2-D imaging, potential thrombus and spontaneous echo contrast should be sought. In addition, LAA plays a hemodynamic role that participates in atrial function and is influenced by various hemodynamic conditions. In view of the embolic risks from a dysfunctional appendage, the LAA is often ligated during cardiac valve surgery. New devices are under evaluation for percutaneous closure of the LAA, and further studies should improve the definition, understanding, and treatment of LAA dysfunction.

Elimination of the Left Atrial Appendage To Prevent Stroke or Embolism?

The "elimination" of the left atrial appendage (LAA) seems to be an attractive alternative to oral anticoagulation in the treatment of atrial fibrillation, especially in patients with contraindications to oral anticoagulation therapy. The LAA, however, plays an important role in the maintenance and regulation of the cardiac function, in arterial hypertension, atrial fibrillation, coronary heart disease, valvular heart disease, and heart failure. Data, mainly from animal studies, indicate that elimination of the LAA may impede thirst in patients with hypovolemia, may impair hemodynamic response to volume or pressure overload, may decrease cardiac output, and may promote heart failure. It may have adverse effects in humans as well. Further studies on the hemodynamic and neurohumoral consequences of left atrial appendage elimination are required to advance our understanding of LAA physiology and pathophysiology.

Percutaneous left atrial appendage occlusion (PLAATO) for preventing cardioembolism: first experience in canine model

BACKGROUND

Atrial fibrillation is associated with a high risk for cardioembolic stroke. The left atrial appendage (LAA) is the source of the vast majority of these thromboemboli. A novel implanted device for percutaneous LAA transcatheter occlusion (PLAATO) has been designed to seal the LAA. The purpose of this study was to test the feasibility and safety of transcatheter LAA occlusion in dogs.

METHODS AND RESULTS

A PLAATO implant was delivered to the LAA through a 12F transseptal catheter in 25 dogs. The PLAATO device was repositioned until occlusion was seen, or it was recaptured and replaced with a different size. LAA sealing was confirmed by intracardiac echocardiography and contrast fluoroscopy. Follow-up was performed 2 days to 6 months after implantation. After imaging assessment, dogs were euthanized and LAA was examined for device healing, migration, perforation, and any thrombosis, both grossly and histologically. The LAA was occluded in all cases. No mobile thrombi associated with the implantation were seen. Healing on the atrial-facing surface was 90% at 1 month and was complete by 3 months, which was confirmed by gross and histological examination. Light microscopic examination of brain, kidney, and spleen showed no evidence of emboli or infarct.

CONCLUSIONS

Transcatheter LAA occlusion is simple and feasible. At the follow-up study, the device remained in the LAA, with benign healing and no evidence of new thrombus or damage to surrounding structures. This new strategy may provide an alternative treatment for patients with nonvalvular atrial fibrillation who are less than optimal candidates for warfarin.

Nonobstructive membranes of the left atrial appendage cavity: report of three cases

A membranous structure causing functional stenosis at the mouth of the left atrial appendage (LAA) has been reported. In this study we describe the presence of nonobstructive membranes traversing the cavity of the LAA found incidentally on transesophageal echocardiography (TEE).

Anatomic correlation of left atrial appendage by 3-dimensional echocardiography

The adequate study of the left atrial appendage is an echocardiographic challenge. The purpose of this study was to assess the ability of 3-dimensional echocardiography in reconstructing this potentially complex structure.

Prognostic value of left atrial appendage function in patients with dilated cardiomyopathy

The purpose of the present study was to determine whether parameters of left atrial appendage (LAA) function, assessed by transesophageal echocardiography (TEE), could predict the clinical outcome in patients with dilated cardiomyopathy (DCM). Fifty-five patients (20 had ischemic cardiomyopathy; mean age, 56+/-14 years) who underwent TEE to evaluate LAA function from 1992 to 1996 were studied. After a mean follow-up period of 34+/-13 months, 16 patients died; the cause was cardiac in 14 and noncardiac in 2. Patients who died of cardiac cause had a lower LAA emptying velocity than survivors (38+/-18 vs 54+/-18 cm/s, p=0.01). There were, however, no significant differences between survivors and nonsurvivors with regard to the maximal LAA area (4.3+/-1.3 vs 4.5+/-0.9 cm2, p=0.55), minimal LAA area (2.4+/-1.1 vs 2.9+1.1 cm2, p=0.13), and LAA ejection fraction (46+/-16 vs 36+/-18%, p=0.05). On the Cox proportional hazards model analysis, LAA emptying velocity <50 cm/s (chi-square 5.9, p=0.02), LAA ejection fraction <43% (chi-square 5.6, p=0.02), female gender (chi-square 5.2, p=0.02), pulmonary artery wedge pressure > or =14 mmHg (chi-square 4.8, p=0.03), E/A ratio > or =1.3 (chi-square 4.6, p=0.03), deceleration time <148 ms (chi-square 4.6, p=0.03), and cardiothoracic ratio > or =54% (chi-square 4.3, p=0.04) were significantly related to cardiac death. The stepwise multivariate analysis revealed that LAA emptying velocity (chi-square 6.1, p=0.01) and gender (chi-square 5.4, p=0.02) were the independent predictors for outcome. In conclusion, the parameters of LAA function may be useful predictors of the clinical outcome in patients with DCM.