Bayes' approach to probability is different from the dominant frequentist approach. In Bayes' it is the prior that makes the difference. Let's say, you do the stool test without knowing anything about your situation. In that case, your 'prior" is the knowledge that among your peer population 7% have some sort of lesions (most of which aren't cancerous yet). If your test comes back positive, then your prior for the second test is the knowledge that among all those people who tested positive the percentage of true positives is substantially larger than the 7%. So, if the second test returns a positive result, you have a substantially higher probability of being a true positive. If, on the other hand, you KNOW that your tests will return a positive result for reasons other than the presence of lesions, doing the test wouldn't make sense at all.