Pranav Bhagirath

61 Evaluation of state-of-the-art segmentation algorithms for left ventricle infarct from late Gadolinium enhancement MR images Table 4. Segmentation accuracy with volume difference (δV) on patient and porcine data for submitted algorithms and fixed-models. The standard deviation of each metric is quoted in brackets. Patient data | δV | (ml) Porcine data | δV | (ml) AIT 0.77 (0.7) 0.84 (0.5) KCL 1.05 (1.0) 0.73 (0.5) MCG 1.02 (0.5) 0.54 (0.1) MV 1.70 (2.3) 0.75 (0.3) UPF 0.70 (0.3) 0.97 (0.7) 2-SD 8.55 (0.4) 4.00 (0.2) 3-SD 6.71 (0.3) 3.52 (0.8) 4-SD 5.20 (0.2) 2.92 (0.8) 5-SD 3.92 (0.3) 2.44 (0.1) 6-SD 2.96 (0.3) 2.08 (0.1) FWHM 3.10 (1.0) 2.20 (0.2) To further evaluate more objectively, the Dice overlap of the algorithms’ segmentations were compared to the consensus based on the slice position (basal, mid and apical. Short-axis slices were subdivided according to the standard guidelines (Cerqueira et al., 2002). The results are plotted in figure 7 . It is not clear what should be a good Dice overlap for datasets of this type. To address this issue, the degree of agreement between observers and the computed consensus was analyzed and plotted in figure 8 . It provided for an estimation of a reasonable target (i.e. good Dice score) for the evaluated algorithms. Pseudo infarct regions The algorithms were evaluated on hyper-enhanced regions which mimic scar. These pseudo infarct regions occur for several aforementioned reasons and illustrated in figure 2 . In each image, pseudo infarct was manually segmented by an experienced observer. These regions were either confirmed anatomically in the case of the outflow tract or by checking adjacent slices for scar continuity in the case of partial voluming. In each image, the total volume of pseudo infarct labeled by the observer was quantified. The total volume of these spurious infarct regions present in each algorithmand fixedmodel segmentation was also quantified. This was possible by comparing each segmentation to the manual labeling of pseudo infarcts. Results are represented in figure 10 . KCL and MCG had a higher proportion of manually labeled pseudo infarct regions detected on