Ana R Fouto1, Rita G Nunes1, Amparo Ruiz-Tagle1, Marc Golub1, Inês Esteves1, Athanasios Vourvopoulos1, Raquel Gil-Gouveia2, Andreia C Freitas1, Nuno A Silva3, and Patrícia Figueiredo1
1ISR-Lisboa/LARSyS and Department of Bioengineering, Instituto Superior Técnico, Universidade de Lisboa, Lisbon, Portugal, 2Neurology Department, Hospital da Luz, Lisbon, Portugal, 3Learning Health, Hospital da Luz, Lisbon, Portugal
1ISR-Lisboa/LARSyS and Department of Bioengineering, Instituto Superior Técnico, Universidade de Lisboa, Lisbon, Portugal, 2Neurology Department, Hospital da Luz, Lisbon, Portugal, 3Learning Health, Hospital da Luz, Lisbon, Portugal
To shorten exams, subsampling dMRI acquisitions may be desirable. Several histogram metrics of DTI and DKI maps showed a significant effect of subsampling, producing biased estimates. Hence, this should be carefully considered depending on the expected effect size on a patient’s population.
Figure 3 - Boxplots showing the distributions of: Top) histogram metrics extracted from the skeletonised DKI parameter maps (MK, AK, RK), in fully sampled (full) and respective subsets ; and Bottom) corresponding relative difference (in percentage) from the ground truth (full). Significant pairwise differences are indicated with ***p≤0.001; **p≤0.01;*0.01<p≤0.05.
Figure 2 - Boxplots showing the distributions of: Top) histogram metrics extracted from the skeletonised DTI parameter maps (FA,MD,AD,RD), in fully sampled (full) and respective subsets ; and Bottom) corresponding relative difference (in percentage) from the ground truth (full). Significant pairwise differences are indicated with ***p≤0.001; **p≤0.01;*0.01<p≤0.05. MD, AD and RD are expressed in ×10−3 mm2/s.