Methods and systems for analysis of wafer scan data

A computerized method for detecting defects or potential defects on a sample, the method comprising operations of:

• receiving scan data obtained in a scan of a sample, the scan data corresponding to a tested pixel on the sample;
• computing a difference vector d based on the scan data and corresponding reference data;
• computing a parameter D, which depends on a vector t=Γd−(Glinear/∥Γs∥2)Γs, wherein ΓTΓ=K−1 with K being a noise covariance matrix corresponding to the tested pixel, s is a predetermined kernel characterizing a defect signal, and Glinear=s·(K−1d) is a gaussian approximation of a likelihood ratio test expression for distinguishing the defect signal from noise, and wherein D substantially monotonically increases with ∥t∥, at least over a range of values of t∥ characteristic of sample analysis;
• and
• computing a score q(g, D) indicative of a probability of the tested pixel being defective, wherein g is a parameter indicative of a value of Glinear, and wherein q(g, D) is a substantially monotonically increasing function of g and a substantially monotonically decreasing function of D, at least for ranges of values of g and D characteristic of sample analysis, thereby generalizing the gaussian approximation-based test to take into account the parameter D and consequently improving defect detection rates.