Atrium, Left

The dataset consists of short-axis and long-axis cine CMR images of 5,008 subjects (61.2 ±7.2 years, 52.5% female), acquired from the UK Biobank. The baseline characteristics of the UK Biobank cohort can be viewed in the data showcase at [16 ]. For short-axis images, the in-plane image resolution is 1.8 ×1.8 mm² with slice thickness of 8.0 mm and slice gap of 2 mm. A short-axis image stack typically consists of 10 image slices. For long-axis images, the in-plane image resolution is 1.8 ×1.8 mm² and only 1 image slice is acquired. Each cardiac cycle consists of 50 time frames. For both short-axis and long-axis views, the balanced steady-state free precession (bSSFP) magnitude images were used for analysis. Details of the image acquisition protocol can be found in [17 (link)].
Manual image annotation was undertaken by a team of eight observers under the guidance of three principal investigators and following a standard operating procedure [18 (link)]. For short-axis images, the LV endocardial and epicardial borders and the RV endocardial borders were manually traced at ED and ES time frames using the cvi⁴² software (version 5.1.1, Circle Cardiovascular Imaging Inc., Calgary, Alberta, Canada). For long-axis 2-chamber view (2Ch) images, the left atrium (LA) endocardial border was traced. For long-axis 4-chamber view (4Ch) images, the LA and the right atrium (RA) endocardial borders were traced.
In pre-processing, the CMR DICOM images were converted into NIfTI format. The manual annotations from the cvi⁴² software were exported as XML files and also converted into NIfTI format. The images and annotations were quality controlled to ensure that annotations cover both ED and ES frames and without missing slices or missing anatomical structures. For short-axis images, 4,875 subjects (with 93,500 annotated image slices) were available after quality control, which were randomly split into three sets of 3,975/300/600 for training/validation/test, i.e. 3,975 subjects for training the neural network, 300 validation subjects for tuning model parameters, and finally 600 test subjects for evaluating performance. For long-axis 2Ch images, 4,723 subjects were available after quality control, which were split into 3,823/300/600. For long-axis 4Ch images, 4,682 subjects were available, which were split into 3,782/300/600.

LV end-diastolic diameter (LVEDD) was measured on the short-axis cine image of the LV at the level of the papillary muscles. LV end-diastolic volume index (LVEDVi), LV end-systolic volume index (LVESVi), LV ejection fraction (LVEF), LV cardiac output (CO), LV cardiac index (CI) and LV mass (LVM) were measured using a post-processing workstation (Philips Intellispace Portal 7.0 and Advantage Workstation 4.6). LV endocardial and epicardial contours were drawn on LV short-axis cine images (papillary muscles were excluded).
LA anteroposterior (AP) diameters were measured on transversal dark blood images. LA volume and function were analyzed using commercial post-processing software (QStrain, Medis Suite 3.1, Leiden, the Netherlands). The LA endocardial border was manually delineated using a point-and-click approach when the atrium was at its maximum and minimum volumes in both the 2-and 4-chamber cine images (pulmonary veins and LA appendage were excluded) (Fig. 1A1-B2). Then the contours were automatically propagated in all frames throughout the entire cardiac cycle (25 frames/cardiac cycle). CMR-FT was visually reviewed to ensure accurate tracking. In cases of inadequate tracking, the endocardial border was manually readjusted and then the propagation algorithm was reapplied. LA global strain and SR were calculated as the average of the two and four chamber views [20 (link)]. Tracking was blindly repeated three times in both the 2-and 4-chamber views, and the results of the LA volume, strain and SR from the three tracking repetitions were averaged in both views. Three aspects of LA strain were analyzed as previously described [19 (link)–21 (link)] (Fig. 1C): total strain (ε_s, corresponding to LA reservoir function), active strain (ε_a, corresponding to LA booster pump function) and passive strain (ε_e, corresponding to LA conduit function, the difference between ε_s and ε_a). Accordingly, three SR parameters were evaluated (Fig. 1D): peak positive strain rate (SRs, corresponding to LA reservoir function), peak early negative strain rate (SRe, corresponding to LA conduit function) and peak late negative strain rate (SRa, corresponding to LA booster pump function).
Fig. 1

This figure shows a representative example of left atrial (LA) tracking on both the 2-and 4- chamber cines in a normal control subject. A1 and A2 left ventricular (LV) end-diastole and end-systole respectively on the 2-chamber view, B1and B2 LV end-diastole and end-systole respectively on the 4-chamber view. C and D The LA strain and strain rate curves. The total strain (ε_s), Passive strain (ε_e) and active strain (ε_a) were identified from the strain curves. The strain rates during LV systole (SRs), LV early diastole (SRe), and atrial contraction (SRa) were also determined from the strain rate curve. E LA volume curve. The LA maximum volume (Vmax), the pre-contraction volume (Vpre-a), and the minimum volume (Vmin) are shown here

LA volume (LAV) was assessed at LV end-systole (LAV_max), at LV diastole before LA contraction (LAV_pre-a), and at late LV diastole after LA contraction (LAV_min) (Fig. 1E). The parameters of the LAV were obtained from a volume curve generated using Simpson’s method. From the LAV, the LA emptying fractions (LAEF) were calculated as follows: (1) LA total EF = (LAV_max-LAV_min) × 100%/LAV_max, (2) LA passive EF = (LAV_max-LAV_pre-a) × 100%/LAV_max, (3) LA active EF = (LAV_pre-a-LAV_min) × 100%/LAV_pre-a [21 (link)].
For estimating the LA segmental function, the software automatically divided the LA wall into 6 segments on both the 2- and 4-chamber views and generated strain curves and SR for each segment. As has been previously described [10 (link)], the LA segments were described as anterior, antero-roof, inferior, septal, septal-roof and lateral walls (Fig. 2A1-B4). The values of the LA segmental strain and SR were obtained from the average of the three repeated measurements. In the case of insufficient tracking quality, the corresponding segments were excluded from the final analysis. Patients with inadequate tracking quality in more than three segments were excluded from the study.
Fig. 2

LA segmentation in representative cases of a healthy subject (A1–4) and a NOHCM patient (B1–4). The LA wall is automatically divided into 6 segments by the software [segment1(S1): anterior, segment2(S2): antero-roof, segment3(S3): inferior, segment4(S4): septal, segment5(S5): septal-roof, segment6(S6): lateral]. Comparison of LA global strain (C) and strain rate (D), segmental strain (C1–6) and strain rate (D1–6) between the non-obstructive hypertrophic cardiomyopathy (NOHCM) (yellow line) and the control (white line), the LA global strain and strain rate in the NOHCM were similar to the control, while segmental strain (inferior) and strain rate (antero-roof, inferior, septal and septal-roof) were lower in the NOHCM than the control. The yellow X axis represented the cardiac cycle length of a patient with NOHCM, and the white X axis represented the cardiac cycle length of a healthy control. ε_s = total strain, ε_e = passive strain, ε ₌, active strain, SRs = peak positive strain rat, SRe = peak early negative strain rate, SRa = peak late negative strain rate. Time dependent curves of the strain parameters were plotted offline using raw values provided by software

Mice were treated with 50ul of 1000U/ml s.c. heparin, and then euthanized with Isoflurane. The chest cavity was opened, the left atrium nicked, and the lungs perfused with 10ml of PBS through the right atrium. The trachea was then exposed, a small incision was made at its top, and an 18 gauge angiocath was inserted and secured. Lung was inflated with digestion solution containing 1.5mg/ml of Collagenase A (Roche) and 0.4mg/ml DNaseI (Roche) in HBSS plus 5% fetal bovine serum and 10mM HEPES. Trachea was tied off with 2.0 sutures. The heart and mediastinal tissues were carefully removed and the lung parenchyma placed in 5ml of digestion solution and incubated at 37°C for 30 minutes with gently vortexing every 8–10 minutes. Upon completion of digestion, 25ml of PBS was added; and the samples were vortexed at maximal speed for 30 seconds. The resulting cell suspensions were strained through a 70um cell strainer and treated with ACK RBC lysis solution.