CNV detection using small NGS panel

Question

[thisisi3]

We are evaluating the detection of copy number variation using small NGS panel(via hybridization). Initial testing on Horizon's HD753 suggests a 1/3 off the true value (true cn 4.5 while our method gives 3). We did some literature digging and found most NGS panels that can detect CNV consist of 100+ genes. We are wondering if small panel is difficult on this task. Has anyone tried detecting CNV using small NGS panel? What are the results?

Thanks.