T. Musha, H. Matsuzaki, Y. Kobayashi, Y. Okamoto, M. Tanaka, and T. Asada
Volume:60, Issue: 8, Page:2332 – 2338 (Open Access)
The mathematical background of a new diagnosis-assisting tool “NAT” is described. This is based on the 21-ch EEG analyses and is featured by 1) an inexpensive terminal unit for EEG recording for 5 minutes (2.5 MB), 2) quick data analyses in a few sec by a remote server via the internet, 3) free from exposure to radiations, 4) high sensitivity to local impairments of the brain activities, and 5) easy operation. A pair of EEG markers sNAT and vNAT is introduced. Each of them is composed of 210 sub-markers (21 signal channels´10 frequency bins), and the brain activity state is represented by a point in each of the 21-dimensional NAT spaces. The control members of a specified brain disease form a cluster in each NAT space, and its center is called the template state of that disease. The proximity of the state of a test subject to the template state in each NAT space is defined as the likelihood to that disease. The cutoff likelihood value separates patients in Alzheimer’s disease (AD) from the normal controls (NLc) with a false rate of 15%. Part of the false judgments is caused by dynamic fluctuations of the brain activity state which cannot be avoided in the living state, but it is dramatically reduced by repeating the NAT exams two or three times within several days. The follow-up of the NAT state of a test subject on the likelihood diagram spanned by a pair of the likelihoods of AD predicts a proper timing for application of the brain activation exercises for slowing down the aproach toward AD.
The case sNAT < sNATNLc denotes that the local brain activity is less excited than that of NLc, and the case vNAT > vNATNLc denotes that the local brain activity is less coherent or organized than that of NLc, in which biological signal transmission is somewhat troubled. Such a situation is shown in the figure compairing with the SPECT images of the same group of 28 AD patients.