def largestCC(mask):

    """Finds the largest connected component.

    Takes a binary numpy array.

    Returns the mask with only the largest component

    """

    import numpy as np

    from skimage.measure import label

    # Creates a labelmap of the mask, where connected components have the same label

    labels = label(mask)

    # labels[np.nonzero(labels)] is a list of all nonzero values in the labelmap. np.argmax(np.bincount()) returns the most common number in that list. largest is set to a mask showing where the labelmap is equal to that value.

    largest = labels == np.argmax(np.bincount(labels[np.nonzero(labels)]))

    if np.count_nonzero(largest) == 0:

        # Throw an exception if the output mask is empty. This could happen if the input mask is also empty

        raise Exception("Largest connected component not found")

    return largest