I think Hellion has the right approach for a puzzle like this. This instance is certainly simple enough to do more or less by inspection. However, since the OP was asking for an approach, I thought a more analytical one might be worth looking at. And as I worked through it, it became clear that there are some interesting characteristics here that may be entertaining.
First, let's set four variables, $a$, $b$, $c$, and $d$, which is enough to establish the whole matrix.
$$\left(\begin{array}{ccc}
a & b & 10-a-b \\
c & d & 19-c-d \\
19-a-c & 24-b-d & a+b+c+d-22
\end{array}\right)
$$
The bottom right cell can be calculated two ways, and obviously these should yield the same result. This is the equivalent of saying that the sum of the row-sums should be the same as the sum of the column-sums (which I checked before I did anything else).
Now, we can set up a bunch of inequalities, since we know that all these numbers are between 1 and 9. The thing that turns out to be interesting (spoiler alert) is that the inequalities can be gradually refined by adding them together and stuff.
Start with all the cells being between 1 and 9, and rearrange into a more readable format.
$$\begin{array}{cccccccc}
10 & \leq & +a & & +c & & \leq & 18 \\
15 & \leq & & +b & & +d & \leq & 23 \\
1 & \leq & +a & +b & & & \leq & 9 \\
10 & \leq & & & +c & +d & \leq & 18 \\
23 & \leq & +a & +b & +c & +d & \leq & 31 \\
\end{array}
$$
There are now refinements we are able to make. Firstly, the lack of repeated numbers in a row or column means that we can say that the maximum sum of two elements should be 17, rather than 18, and the minimum sum with similar logic is 2. Hence
$$\begin{array}{cccccccc}
10 & \leq & +a & & +c & & \leq & 17 \\
15 & \leq & & +b & & +d & \leq & 23 \\
2 & \leq & +a & +b & & & \leq & 9 \\
10 & \leq & & & +c & +d & \leq & 17 \\
23 & \leq & +a & +b & +c & +d & \leq & 31 \\
\end{array}
$$
Also, we can do an operation where we mix equations in various ways and then keep the tightest option. I'll spell out the first few operations.
Add the first and second inequalities, and add the third and fourth inequalities. These, along with the existing fifth inequality, give various expressions for $a+b+c+d$, and we can keep the tightest of these.
$$\begin{array}{rccccccc}
25 & \leq & +a & +b & +c & +d & \leq & 40 \\
12 & \leq & +a & +b & +c & +d & \leq & 26 \\
23 & \leq & +a & +b & +c & +d & \leq & 31 \\
\implies 25 & \leq & +a & +b & +c & +d & \leq & 26 \\
\end{array}
$$
Already, these bounds are tightening up nicely, and the process has been largely mechanical. Here's another example. I'll take the fourth inequality in $c+d$ and multiply it through by $-1$. This I will add to the inequality for $a+b+c+d$, and use the result to refine the inequality for $a+b$.
$$\begin{array}{rccccccr}
-17 & \leq & & & -c & -d & \leq & -10 \\
25 & \leq & +a & +b & +c & +d & \leq & 26 \\
8 & \leq & +a & +b & & & \leq & 16 \\
\implies 8 & \leq & +a & +b & & & \leq & 9 \\
\end{array}
$$
One can go on in this fashion and deduce that $16\leq c+d\leq 17$ and so on. Eventually there are just a few cases to consider and the answers Hellion worked out can be found. Obviously this method is more interesting when the question is more complex, but hopefully some will find this approach entertaining. With pen and paper, the refinements can be made relatively easily on a single matrix.