Reading of Excel files in MATLAB_R2016 vs MATLAB_R2018

When i am trying to extract data using the command [num_1,txt_1]= xlsread('test','Sheet1'), R2016 version considers the 1st row (which is column headers) too, despite it shows NaN. At the same time R2018 version correctly extract numbers starting the seconds row. So I cannot use the same program to extract data in two MATLAB versions. What is the reason?

[Walter Roberson]

Could you attach the .xls, just in case there is something odd about it?

The num output of xlsread() was always intended to omit all leading and trailing rows and columns that have no entries that can be interpreted as numbers. Like

nn = isnan(converted_data);
nnr = ~all(nn, 2);
sr = find(nnr, 1, 'first');
er = find(nnr, 1, 'last');
nnc = ~all(nn, 1);
sc = find(nnc, 1, 'first');
ec = find(nnc, 1, 'last');
num = converted_data(sr:er, sc:ec);