Information Technology Reference
In-Depth Information
The example results of AA labeling process are presented in Fig 5. Each row
presents two pairs of CT images. Left figure of each pair is static CT image supe-
rimposed on moving CT image. Right figure of each pair is static CT image with
tissues labels. First row is a state before registration, next free rows are results of
affine register with Nelder-Mead Simple method, Thirion's algorithm and FFD
algorithm (initial grid 6x6, 12 iterations, 2 mesh refinement).
Fig. 5 The example results of AA labeling process
5 Classification of Detected Abnormalities
After detection of asymmetry regions (lesions) in CBF and CBV maps semantic
interpretation of extracted features is performed (a diagnosis based on the detected
symptoms [4]). The algorithm decides what kind of lesion (ischemic / hemorrhag-
ic) and in which hemisphere was detected. It is done by comparing averaged CBF
/ CBV with normal values from Table 1 (Fig 6).
In the next step algorithm analyze both perfusion maps simultaneously in order to
detect:
1. Tissues that can be salvaged (tissues are present in CBF and CBV asymmetry
map and values of rCBF did not drop beyond 0.48 - Table 2)
Search WWH ::




Custom Search