Pythonic way to compute sensitivity and specificity

Focusing on compactness through NumPy supported ufunc-vectorized operations, broadcasting and array-slicing, here's an approach -

C = (((mask==255)*2 + (truth==255)).reshape(-1,1) == range(4)).sum(0)
sensitivity, specificity = C[3]/C[1::2].sum(), C[0]/C[::2].sum()

Alternatively, going a bit NumPythonic, we could have counts C with np.bincount -

C = np.bincount(((mask==255)*2 + (truth==255)).ravel())

To make sure we are getting floating pt numbers as the ratios, at the start, we need to use : from __future__ import division.

test for same shape:

a = np.random.rand(4,4)
b = np.random.rand(4,4)
print(a.shape == b.shape) #prints true

test for truth values:

#assuming you have scaled a and b to only include 1 or 0 (divide by 255)
true_positive = np.sum(mask * truth)

true_negative = len(mask.flat) - np.count_nonzero(mask + truth)

false_positive = np.count_nonzero(mask - truth == 1)

false_negative = np.count_nonzero(truth - mask == 1)

The four arrays can be find and organized like that:

categories=dstack((mask&truth>0,mask>truth,mask<truth,mask|truth==0))

then the scores :

tp,fp,fn,tn = categories.sum((0,1))

finally the results :

sensitivity,specificity = tp/(tp+fn),tn/(tf+fp)

My idea is use collections.Counter from standard library.

# building pair list (can be shortened to one-liner list comprehension, if you want)
pair_list = []
for y in range(mask.shape[0]):
    for x in range(mask.shape[1]):
        pair_list.append((mask[y, x], truth[y, x]))

# getting Counter object
counter = collections.Counter(pair_list)
true_positive = counter.get((255, 255))
false_positive = counter.get((255, 0))
false_negative = counter.get((0, 255))
true_negative = counter.get((0, 0))