EHJ-IMP Editor-in-Chief
RT @EHJIMPEiC: Hot off the press #EHJIMP🔥🔥 🤖
Automated Echocardiographic Assessment of Dilated Cardiomyopathy Using Deep Learning in a Resource-Limited Setting 🤖
📌 Key Findings:
Population: 186 patients with dilated cardiomyopathy (DCM) at Jakaya Kikwete Cardiac Institute (JKCI), Tanzania
Approach: U-Net deep learning framework for automated left ventricular (LV) segmentation and ejection fraction (EF) estimation, using models developed on CAMUS (still images) and EchoNet-Dynamic (video)
Domain shift: The CAMUS-trained still-image model produced fragmented predictions on JKCI data, unsuitable for LV volume estimation
EF estimation after fine-tuning (EchoNet-Dynamic):
– MAE: 7.21% → 4.63%
– RMSE: 9.84% → 6.66%
LV segmentation (Dice similarity coefficient):
– End-systole: 0.90
– End-diastole: 0.84
🔗 Read the full article:
@VikashJaiswalMD @EHJIMPEiC @JGrapsa @ESC_Journals
#Echocardiography #DeepLearning #DilatedCardiomyopathy #CardiovascularImaging #GlobalHealth #EHJIMP
01101.7K
















