The Radon-Nikodym derivative is the rigorous way of dealing with a change of variables.
Given two sigma-finite measures μ, ν on a measurable space, we say ν is absolutely continuous with respect to μ (written ν < < μ) if ker μ ⊆ ker ν. In other words, every μ-null set is also a ν-null set.
If ν < < μ, there exists a measurable function f such that ν(A) = ∫Afdμ for all A. The function f is called the Radon-Nikodym derivative and denoted $$ f = \dfrac{d \nu}{d \mu}. $$
You can define a measure for free field theories, μ0. You define the interacting measure ν by using the RN derivative: $$ d \nu(\phi) = \dfrac{1}{Z} e^{-\lambda \int \phi^{4}} d\mu_{0}(\phi) $$
As it turns out, in 4D, ν ⊥ μ0 (they're mutually singular), and so the RN derivative doesn't exist.