A novel algorithm to detect the baseline value of a time signal in Dynamic Contrast Enhanced-Computed Tomography

Reproducibility of Computed Tomography perfusion parameters in hepatic multicentre study in patients with colorectal cancer

“…For both vessels, one mean TCC (C A (t) and C P (t), respectively) is achieved from the whole ROIs while, from the tissue, single voxel-based TCCs (C T (t)) are extracted and kept, after excluding voxels undergoing dynamic artefacts [18] and performing a 3D edge preserving filtering. Finally, after an accurate baseline removal [28], a preliminary non-parametric fitting of the real TCCs is performed to up-sample the signal from 30 s on, to the end of the acquisition, so to have a uniform sampling frequency of 1 Hz.…”

The effects of baseline length in Computed Tomography perfusion of liver

“…The bounds of fitting parameters were adjusted to speed-up the overall processing time and to avoid non-physiological solutions. Accordingly, E min was set to the baseline value found with the algorithm described in [47]. As regards E max , to set the lower bound we have not to forget that the Hill equation is a monotonic sigmoidal model conceived to represent only the ascent part of the TCC.…”

“…In fact, while lower values clearly include many samples of the descent phase, choosing a higher values selects too a low number of samples in the peak's neighbourhood (or even isolated peaks due to noise), that makes them loosing representativeness. As for EC50, it was left free within its whole physiological range, that is between the end of baseline (t B ) [47] and the TCC's peak time. Finally, as far as α is concerned, negative values yield decreasing ĉT (t), while 0 < α ≤ 1 yields curves without baseline, soon increasing at the first time instant, this being infeasible with our CTp protocol (Sect.…”

“…In particular, the set of delayed CTp examinations E τ was obtained from E 0 ones by progressively disregarding the first i CT scans. The maximum number of samples disregarded (i.e., fifteen) was selected to assure the presence of at least one baseline data-point in each TCC, after an automatic analysis of the patients' baseline lengths carried out through the algorithm described in [47]. Being the sampling time equal to 1 s for the first thirty seconds, the number i of CT scans discarded coincides with τ .…”