A CNN pretrained with generated data achieved similar accuracy to an automatic Kalman filter method for LV segmentation, with a Dice coefficient of 0.86 vs 0.87 and Hausdorff distance of 5.9 vs 7.5 mm.
Does a deep convolutional neural network pretrained with generated data achieve similar accuracy to an automatic Kalman filter method for left ventricle ultrasound image segmentation?
Deep learning models pretrained on automatically generated annotations can achieve comparable left ventricle segmentation accuracy to traditional automatic methods, potentially reducing the need for manual expert annotations.
Absolute Event Rate: 0.86% vs 0.87%
Automatic segmentation of the left ventricle (LV) can become a useful tool in echocardiography. Deep convolutional neural networks (CNNs) have shown promising results for image classification and segmentation on several domains, however CNNs seem to require a lot of training data. In this work, CNNs are investigated for LV ultrasound image segmentation. We study if the need for manual annotation can be reduced by pretraining a CNN using a previously published automatic Kalman filter (KF) based segmentation method. The results show that a CNN is able to achieve similar accuracy to that of the automatic method, by only training with generated data. The dice similarity coefficient was measured to be 0.86 ± 0.06 for the CNN versus 0.87 ± 0.06, while the Hausdorff distance was better at 5.9 ± 2.9 mm for the CNN versus 7.5 ± 5.6 mm for the KF method. In future work, this may enable CNNs to exceed state-of-the-art with a small set of expert annotations for fine-tuning.
Smistad et al. (Fri,) conducted a other in Left ventricle ultrasound image segmentation. Deep convolutional neural networks (CNNs) pretrained with generated data vs. Automatic Kalman filter (KF) based segmentation method was evaluated on Dice similarity coefficient and Hausdorff distance. A CNN pretrained with generated data achieved similar accuracy to an automatic Kalman filter method for LV segmentation, with a Dice coefficient of 0.86 vs 0.87 and Hausdorff distance of 5.9 vs 7.5 mm.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: