I have an alternative answer to what Steel provided. If you have an equation as an image, you COULD do just what he says and manually put in an alt tag. This will give you the most control about what assistive technology actually says.
Alternatively, if you are on a network that has MathJax properly configured, you could write out your equation in latex, and MathJax would display your equation as an image AND simultaneously put the meta-data in there so that the user can choose how they want it read to them.
A pro user of assistive technology might have preferences about whether they prefer verbose or brief descriptions from the math. They might even have a preference between whether they want the Clear Speak rules or MathSpeak rules.
Think: do you want f(x)=y to say 1) “F of ex equals why” or do you want it to say 2) “f left parenthesis ex right parenthesis equals why”
My initial reaction when I asked this question was that I would prefer 1, because that is how a mathematician would read that, however, the second one gives me an idea of the notation and how it is presented. My point is that both may have value in different contexts.
So while doing the work to put in manual alt text gives the AUTHOR that power, what might be the most time efficient and put the power most in the hands of the USER of assistive technology might be using LaTeX/MathJax.
The fact that you are asking these questions at all means that you are being very thoughtful of all sorts of future users.