Model-free functional MRI analysis using Kohonen clustering neural networkand fuzzy c-means

Conventional model-based or statistical analysis methods for functional MRI (fMRI) suffer from the limitation of the assumed paradigm and biased resul ts. Temporal clustering methods, such as fuzzy clustering, can eliminate th ese problems but are difficult to find activation occupying a small area, s ensitive to noise and initial values, and computationally demanding. To ove rcome these adversities, a cascade clustering method combining a Kohonen cl ustering network and fuzzy c means is developed. Receiver operating charact eristic (ROC) analysis is used to compare this method with correlation coef ficient analysis and t test on a series of testing phantoms, Results show t hat this method can efficiently and stably identify the actual functional r esponse with typical signal change to noise ratio, from a small activation area occupying only 0.2% of head size, with phase delay, and from other noi se sources such as head motion, With the ability of finding activities of s mall sizes stably, this method can not only identify the functional respons es and the active regions more precisely, but also discriminate responses f rom different signal sources, such as large venous vessels or different typ es of activation patterns in human studies involving motor cortex activatio n. Even when the experimental paradigm is unknown in a blind test such that model-based methods are inapplicable, this method can identify the activat ion patterns and regions correctly.