Delete whole 3d array if contains any nan

As already answered by Michael Szczesny, more Pythonic way would be:

filtered_images=a[~np.isnan(a).any(axis=(2,1))]

If that piece of code is hard to understand, then consider extracting each image with a for loop as follows:

filtered_images=list()
for value in a:
  if(np.isnan(value).any()!=True):
    filtered_images.append(value)

Both approaches should give you similar output!