The echocardiogram is a test that is widely used in Heart Disease Diagnoses. However, its
analysis is largely dependent on the physician's experience. In this regard, artificial …

The detection of recorded epileptic seizure activity in electroencephalogram (EEG)
segments is crucial for the classification of seizures. Manual recognition is a time …

Obtaining manual labels is time-consuming and labor-intensive on cardiac image
sequences. Few-shot segmentation can utilize limited labels to learn new tasks. However, it …

Abstract Machine learning promises to assist physicians with predictions of mortality and of
other future clinical events by learning complex patterns from historical data, such as …

Deep supervision, or known as' intermediate supervision'or'auxiliary supervision', is to add
supervision at hidden layers of a neural network. This technique has been increasingly …

Echocardiography is an essential diagnostic method to assess cardiac functions. However,
manually labelling the left ventricle region on echocardiography images is time-consuming …

Cancer prevention is mainly achieved by screening the transformation zones. Cervical pre-
cancerous stages can be seen in three different types, and all can transform into cancer …

Deep learning (DL) has been applied for automatic left ventricle (LV) ejection fraction (EF)
measurement, but the diagnostic performance was rarely evaluated for various phenotypes …

摘要近年来, 随着深度学习技术的发展, 深度神经网络在医学图像分割领域得到了广泛应用.
其中, U-Net 以其良好的分割性能, 逐渐成为图像分割领域的研究热点. 根据相关研究工作 …

Segmentation of cardiac structures is one of the fundamental steps to estimate volumetric
indices of the heart. This step is still performed semiautomatically in clinical routine and is …