More efficient way to find word (and its letters' indexes) diagonally in Array (or list of lists)

Question

I'm looking for one fast method to find words in Array and get them letters' indexes. Words can be put horizontally, Vertically, Diagonally, in each direction (left-right, right-left, top-bottom, bottom-top, bottomleft-topright, topright-bottomleft, bottomright-topleft, topleft-bottomright). Major problem is Diagonally dimension... How can I do?

Example:

Matrix:
   [['CHYD'],
    ['IAOB'],
    ['LGTA'],
    ['EFGE'],
    ['RXBW']]

Words:
   ['CAT',
    'DOG',
    'EGO',
    'RELIC',
    'TOY']

I need to obtain a dictionary with keys=Rowindex and Value=[set of column indexes of found words' letters]. In my Example, I'll find:

• 'CAT' in (0,0), (1,1), (2,2)
• 'DOG' in (0,3), (1,2), (2,1)
• 'EGO' in (1,2), (2,1), (3,0)
• 'RELIC' in (0,0), (1,0), (2,0), (3,0), (4,0)
• 'TOY' in (0,2), (1,2), (2,2)
• 'TAC' in (0,0), (1,1), (2,2)

I would my final output is Found_Word={0:[0,2,3], 1:[0,1,2], 2:[0,1,2], 3:[0], 4:[0]}

Can you help me? I hope I was clear :)

P.s. I can use all Basic Python libraries (included in Anaconda Distribution), like numpy, colleciotions, itemtools and similar. I can't use Global Variables or classes.