Skip to main content

Table 2 Quantitative results for each segmentation algorithm

From: Myofibre segmentation in H&E stained adult skeletal muscle images using coherence-enhancing diffusion filtering

  Fibre acc. ( %) Fragmentation Congealment Mean diameter ( μ m) VC ( μ m)
Thresholding [6] 66.97 (±20.75) 0.43 (±0.30) 0.35 (±0.28) 60.53 (±21.75) 337.10 (±92.60)
Line detectors [7] 68.16 (±19.61) 0.08 (±0.11) 0.19 (±0.13) 66.01 (±22.32) 336.00 (±139.22)
Vesselness enhancement [9] 63.90 (±23.42) 0.28 (±0.19) 0.07 (±0.12) 53.40 (±19.63) 312.84 (±58.20)
Hybrid level sets [31] 76.93 (±18.12) 0.09 (±0.09) 0.18 (±0.13) 63.41 (±21.48) 327.64 (±105.00)
Chan-Vese AC [32] 72.62 (±15.93) 0.06 (±0.08) 0.20 (±0.13) 66.41 (±22.21) 325.03 (±103.71)
Proposed method 88.65 (±8.58) 0.08 (±0.07) 0.10 (±0.11) 62.83 (±18.31) 251.85 (±59.74)
Ground truth     59.21 (±15.58) 230.31 (±40.07)