I found a way that meets my requirements, after carefully reading the meshgrid and griddata help. I hope it can help someone else.
First i set up 2 vectors containing x and y positions of my points of interest, xp and yp. Then use griddata to generate interpolated values (Tp) for those positions. (Thats all i needed!)
x = pos_nos(:,1); % original data
y = pos_nos(:,2);
T = pos_nos(:,3);
%
xp = [0.0001 ; 0.0008 ; 0.0014];
yp = [0.0001 ; 0.0008 ; 0.0014];
%
Tp = griddata(x,y,T,xp,yp,'natural');
After that, i can use meshgrid to generate a fine mesh where the results will be interpolated, in case i need to extract more data. Then i just use griddata and plot the image.
[xg,yg] = meshgrid(0:0.0001:0.006, 0:0.0001:0.004);
[Xr,Yr,Tr] = griddata(x,y,T,xg,yg,'natural');
figure();
colormap(jet);
[cc,hh]=contourf(Xr,Yr,Tr,'Showtext','on','LevelList',[400 600 700 800 900 1000 1100 1200 1300]);