Normally, you shoudn't have any amplification in your negative (no template) control (No Ct value given by the software of the machine). In you case, it seems that your plate is contaminated by your DNA template and you get a quite high Ct value.
Try to rerun a new experiment with fresh aliquots of primers, SYBR green or probes etc. If you work with human DNA, you should fill your plate under a hood to avoid contamination. If not, working on a very clean bench and avoiding to go over your plate with your hands/arms is enough.
If you are using SYBR Green the two major causes for NTC amplification are contamination (as Winka mentioned) or primer dimers. Have you checked your melting curve for NTC? This can help you clarify primer dimer formation.
However this kind of amplification happens at higher Cts (above 35th cycle). Since you have NTC amplification around the 18th cycle you probably have a major source of contamination. I agree with Winka that you should rerun your experiment using fresh aliquots and pay a lot of attention with GLPs.
Sometimes poorly designed primers result in the very high burden of unspecific amplification. In that case (if you use SYBR Green or other DNA binding dye) the Cts in NTC may reach even 18-20. To check this, after your amplification compare melting profiles for NTCs and positive control runs - if they appear clearly different, that means that you have problems with your design rather than problems with contamination
Thanks people for your valuable comment, I have checked melting profiles for NTC which is sometimes nothing visible and some times very small peak with low temperature.
The issue is now i am checking DNA contamination from isolated total RNA, so I used this as template,using gene specific primer it gives amplification, Why ?
Do your primers span an exon-exon junction? This approach helps avoid gDNA amplification. If your primers are located within the same exon than you should increase your DNAse concentration and treatment time.
For NTC you should have no amplification. If you do your data is compromised, you won't know in your test sample how much is real amplification and how much is unspecific amplification.
Since the melting curve for NTC shows a small peak with low temperature you probably have a primer dimer. What's you qPCR efficiency? I believe you will have to redesign your primers.