A method of data signal processing is provided, wherein the method comprises filtering a pre-filtered quadrature data signal by using an m×n filtering matrix, wherein m≠n. In particular, the filtering may be performed by using a rectangular matrix, i.e. a non-quadratic matrix. That is, the filtering may be performed by a matrix multiply in an image cancellation filtering unit. In general m relates to the 5 number of rows of the matrix while n relates to the number of columns of the matrix. In particular, the processing may form or may be a part of an image cancellation process.