P1486Semi-automated volume-strain loops: a new tool in TTE to assess diastolic dysfunction

Aim This work aims to evaluate a novel semi-automatic tool for the assessment of volume-strain loops by transthoracic echocardiography (TTE). The proposed method was evaluated on a typical model of left ventricular (LV) diastolic dysfunction: the cardiac amyloidosis. Method 18 patients with proved cardiac amyloidosis were compared to 19 controls, from a local database. All TTE were performed using Vivid E9 or E95 ultrasound system. The complete method includes several steps: 1) extraction of LV strain full traces from apical 4 and 2 cavities views, 2) estimation of LV volume from these two traces by spline interpolations, 3) resampling of LV strain curves, determined for the same cardiac beat, (in apical 4-, 2- and 3- cavities views) as a function of pre-defined percentage increments of LV-volume and 4) calculation of the LV volume-strain loop area. (Figure 1, panel B) Results (Table 1): LVEF was similar between both groups whereas global longitudinal strain was significantly lower in amyloidosis group (−14.4 vs −20.5%; p<0.001). Amyloidosis group had a worse diastolic function with a greater left atrial volume index (51 vs 22ml/m2), a faster tricuspid regurgitation (2.7 vs 2.0 m/s), a greater E/e' ratio (17.3 vs 5.9) with a p<0.001 for all these indices. Simultaneously, the global area of volume-strain loop was significantly lower in amyloidosis group (36.5 vs 120.0%.mL). This area was better correlated with mean e' with r=0.734 (p<0.001) than all other indices (Figure 1, panel A). Table 1 Amyloidosis (N=18) Controls (N=19) p Global strain-volume loop area (%.mL) 36.5±21.3 120.0±54.2 <0.001 Global longitudinal strain (%) −14.4±3.8 −20.5±1.8 <0.001 Left ventricular ejection fraction (%) 62±7 65±5 0.08 Left atrial volume index (ml/m2) 51±22 22±5 <0.001 E/A 1.72±0.97 2.07±0.45 0.17 Mean e' 5.5±1.3 14.4±2.8 <0.001 Mean E/e' 17.3±5.4 5.9±1.4 <0.001 Tricuspid regurgitation velocity (m/s) 2.7±3.8 2.0±0.3 <0.001 Figure 1 Conclusion LV volume-strain loop area appears a very promising new tool to assess semi-automatically diastolic function. Future applications will concern the integration of LV volume-strain loop area as novel feature in machine-learning approach.