NaN issue when calculating for a mean value

gh = 0;
fg = 1; 
mu_fg = MU_fib/fg;  
adi = 1; 
w = 0.042;
for ix = 1:200 
    for iy = 1:200
        gh = gh + 1;
        D = phi.*exp(MU_fat).*mu_fg.*((1-exp(-mewelement.*w))/mewelement).*((fg+adi)/adi).*(1/w);
        meanD = sum(D)/60;
    end 

Above is a section of the program I am writing. This section has the goal of find the dose for each section of tissue. However, when i run it through, I get dose as NaN. I want the mean dose for the entire phantom as a single number.

Without the line meanD = sum(D)/60;, I get a matrix of 1 x 60 values which i don't want. I want one value for the dose (D).

Thank you. Any help is appreciated

[Guillaume]

Unfortunately, we don't know what MU_fib, MU_fat, phi and mwelement are, so can't test your code. You're getting NaN because one of your expression result in either

• 0 / 0
• Inf / Inf
• 0 * Inf
• Inf - Inf or similar operation

or one of your input is NaN already.

"I get a matrix of 1 x 60 values which i don't want" If you didn't expect a vector, then clearly you've got a bug. You'll get a vector if any of the above variables are a vector.

Note that as written, your loops overwrite meanD and D at each step, so you'll end up with just the value for ix = 200 and iy = 200 at the end.

Answer 1

[Star Strider]

The NaN result is due to one or more elements of the vector you are taking the mean of being (0/0) or (Inf/Inf), since they evaluate to NaN.

There are three options:

(1) Find and eliminate the source of the NaN elements

(2) Use the approach in Mean Excluding NaN, or the nanmean equivalent

(3) use D(~isnan(D)) as the argument to mean in your present code.