Methods and systems for detecting ECG anomalies
Methods, apparatus and systems for robust and accurate detection of anomalies in medical images and electrocardiograms are disclosed. One example system for training a neural network engine includes a processor that is configured to receive a set of training electrocardiogram signals. At least one electrocardiogram signal in the set of training electrocardiogram signals is associated with metadata identifying a region of interest that includes a heart anomaly. The processor is configured to input the set of training electrocardiogram signals into the neural network engine. The neural network engine is trained using an objective function having a first regularization parameter and a second regularization parameter. The processor is also configured to operate the neural network engine to identify the heart anomaly by classifying the set of training electrocardiogram signals and adjust the neural network engine based on the identified heart anomaly and the metadata.
1 . A method for performing classification of image data, the method comprising:
receiving an input image data having a feature of interest, wherein the input image data is associated with a mask identifying a region that includes the feature of interest;
inputting the input image data into a neural network engine that is trained using an objective function having a first regularization parameter and a second regularization parameter different from the first regularization parameter, wherein the first regularization parameter indicates a first degree of sensitivity associated with samples located within the mask, wherein the second regularization parameter indicates a second degree of sensitivity associated with samples located outside of the mask, and wherein the first degree of sensitivity is different from the second degree of sensitivity; and
identifying the feature of interest by classifying the input image data using the neural network engine.
2 . The method of claim 1 , further comprising:
receiving feedback information in response to an identified feature of interest, wherein the feedback information is used to validate the identified feature of interest or correct the identified feature of interest, and wherein the feedback information is used to adjust the neural network engine.
3 . The method of claim 2 , wherein the neural network engine is adjusted by online training or offline re-training of the neural network engine.
4 . The method of claim 1 , wherein the first regularization parameter and the second regularization parameter are greater than zero.
5 . The method of claim 1 , wherein the objective function is based on gradient-based regularization using the first regularization parameter and the second regularization parameter.