Is there any of training a neural network to recognise abnormal images in a sequence?

Hi, I am using MATLAB R2020a on a MacOS. I am currently analysing ECG signals to identify abnormal cycles in real-time on a cycle by cycle basis. This involves plotting the trajectories of each cycle in the phase-space on a 2D plane. I was wondering if there was any way of feeding some cycles from an individual's trace into a deep learning algorithm to train it and generate a template for 'normal' cycles which is then compared to each subsequent cycle to find and flag 'abnormal' cycle trajectories? Is there also a way of feeding newly found normal cycles into the template as they are found to update it and then using this template for future comparisons?

Any suggestions would be greatly appreciated. Thanks in advance.