Table 1.
Summary of studies addressing the segmentation of magnetic resonance imaging and computed tomography images.
Fig 1.
Brain mapping: (a) specific brain regions and (b) different brain structures.
Fig 2.
Brain exams: (a) computed tomography e (b) magnetic resonance.
Fig 3.
Flowchart of the proposed methodology.
Fig 4.
Flowchart of the proposed architecture.
Table 2.
and σ along the twelve simulations for each NR and for each class c1, c2 and c4.
Table 3.
Hyperparameters optimized based on the predefined quantity of labels.
Fig 5.
Windowing filter 40 × 80 [HU].
Fig 6.
Segmentation of exams: (a) to (f) CT042 and (g) to (l) CT195.
Fig 7.
Segmentation of exams: (a) to (f) CT200, (g) to (l) CT299, and (m) to (r) CT418.
Table 4.
Comparison of Dc between the proposed method versus Kim, Kanezaki, and Tanaka’s method [38].
Fig 8.
Segmentation overlap of the CT195 exam: (a) c1, (b) c2, (c) c3, (d) c4, (e) c5 e (f) c6.
Table 5.
Comparison of cranial volumetry values between the proposed method versus CTSeg.
Fig 9.
Cranial volumetry for class c4 in the CT042 exam.
Fig 10.
Segmentation using the proposed method in exams: (a) CT042 NR = 7, (b) CT042 NR = 8, (c) CT195 NR = 7, (d) CT195 NR = 8, (e) CT200 NR = 7, (f) CT200 NR = 8, (g) CT299 NR = 7, (h) CT299 NR = 8, (i) CT418 NR = 7, and (j) CT418 NR = 8.