The present invention relates to a method to perform an accurate correlation between a first set of data (1), collected from a first source (A1), and a second set of data (2), collected from a second source (A2), in the presence of noise in one or both of the first and second sets of data (1, 2). The present invention specifically teaches that a third set of data (3) is collected from the first source (A1) under the same conditions as the conditions of the collection of the first set of data (1), and that a fourth set of data (4) is collected from the second source (A2) under the same conditions as the conditions of the collection of the second set of data (2). A measured correlation value (rm) between data collected from the first source (A1) and the second source (A2) is taken as one of the correlation measurements out of the possible combinations of the first and second set of data (r12), the first and fourth set of data (r14), the third and second set of data (r32) and the third and fourth set of data (r34). A correction value (C12) is calculated based on a correlation measurement between the first and third set of data (r13) and a correlation measurement between the second and fourth set of data (r24) according to the expression: (Formula (I)). This correction value (C12) is applied to the measured correlation value (rm) resulting in a corrected correlation value rc=rm×C12. C ⁢ ⁢ 12 = 1 r ⁢ ⁢ 13 × r ⁢ ⁢ 24 ( I )