Skip to main content

Table 4 Performance on Arabidopsis 3D Digital Tissue Atlas

From: Cellstitch: 3D cellular anisotropic image segmentation via optimal transport

Dataset

Method

Precision

Recall

AP

mAP

Anther

CellStitch\(^*\)

\(\mathbf {0.66 \pm 0.10}\)

\(\mathbf {0.53 \pm 0.11}\)

\(\mathbf {0.42 \pm 0.10}\)

\(\mathbf {0.41 \pm 0.08}\)

Cellpose2D

\(0.48 \pm 0.11\)

\(\mathbf {0.53 \pm 0.10}\)

\(0.33 \pm 0.08\)

\(0.34 \pm 0.07\)

Cellpose3D

\(0.41 \pm 0.17\)

\(0.38 \pm 0.13\)

\(0.24 \pm 0.10\)

\(0.23 \pm 0.08\)

PlantSeg

\(0.31 \pm 0.05\)

\(0.60 \pm 0.10\)

\(0.26 \pm 0.04\)

\(0.26 \pm 0.04\)

Filament

CellStitch\(^*\)

\(\mathbf {0.74 \pm 0.14}\)

\(0.53 \pm 0.12\)

\(\mathbf {0.46 \pm 0.13}\)

\(\mathbf {0.46 \pm 0.11}\)

Cellpose2D

\(0.51 \pm 0.15\)

\(\mathbf {0.57 \pm 0.13}\)

\(0.38 \pm 0.13\)

\(0.38 \pm 0.12\)

Cellpose3D

\(0.03 \pm 0.02\)

\(0.22 \pm 0.11\)

\(0.03 \pm 0.02\)

\(0.03 \pm 0.02\)

PlantSeg

\(0.37 \pm 0.08\)

\(0.31 \pm 0.13\)

\(0.20 \pm 0.08\)

\(0.21 \pm 0.06\)

Leaf

CellStitch\(^*\)

\(\mathbf {0.83 \pm 0.17}\)

\(\mathbf {0.75 \pm 0.19}\)

\(\mathbf {0.68 \pm 0.21}\)

\(\mathbf {0.66 \pm 0.18}\)

Cellpose2D

\(0.65 \pm 0.19\)

\(\mathbf {0.77 \pm 0.19}\)

\(0.56 \pm 0.20\)

\(0.55 \pm 0.18\)

Cellpose3D

\(0.07 \pm 0.05\)

\(0.53 \pm 0.20\)

\(0.06 \pm 0.04\)

\(0.06 \pm 0.04\)

PlantSeg

\(0.30 \pm 0.08\)

\(0.71 \pm 0.20\)

\(0.27 \pm 0.09\)

\(0.27 \pm 0.08\)

Pedicel

CellStitch\(^*\)

\(0.57 \pm 0.23\)

\(\mathbf {0.44 \pm 0.16}\)

\(\mathbf {0.35 \pm 0.16}\)

\(\mathbf {0.36 \pm 0.13}\)

Cellpose2D

\(0.36 \pm 0.19\)

\(\mathbf {0.44 \pm 0.15}\)

\(0.25 \pm 0.13\)

\(0.26 \pm 0.11\)

Cellpose3D

\(\mathbf {0.61 \pm 0.25}\)

\(0.30 \pm 0.13\)

\(0.27 \pm 0.13\)

\(0.28 \pm 0.09\)

PlantSeg

\(0.29 \pm 0.07\)

\(0.39 \pm 0.13\)

\(0.20 \pm 0.06\)

\(0.21 \pm 0.05\)

Sepal

CellStitch\(^*\)

\(\mathbf {0.51 \pm 0.12}\)

\(\mathbf {0.43 \pm 0.16}\)

\(\mathbf {0.31 \pm 0.12}\)

\(\mathbf {0.33 \pm 0.09}\)

Cellpose2D

\(0.33 \pm 0.11\)

\(\mathbf {0.44 \pm 0.17}\)

\(0.23 \pm 0.09\)

\(0.25 \pm 0.08\)

Cellpose3D

\(0.41 \pm 0.23\)

\(0.34 \pm 0.12\)

\(0.20 \pm 0.10\)

\(0.21 \pm 0.09\)

PlantSeg

\(0.34 \pm 0.07\)

\(0.43 \pm 0.14\)

\(0.23 \pm 0.06\)

\(0.25 \pm 0.05\)

Valve

CellStitch\(^*\)

\(\mathbf {0.71 \pm 0.07}\)

\(0.47 \pm 0.04\)

\(\mathbf {0.40 \pm 0.05}\)

\(\mathbf {0.41 \pm 0.11}\)

Cellpose2D

\(0.58 \pm 0.09\)

\(0.47 \pm 0.04\)

\(0.35 \pm 0.05\)

\(0.37 \pm 0.05\)

Cellpose3D\(^*\)

\(0.67 \pm 0.10\)

\(\mathbf {0.51 \pm 0.05}\)

\(\mathbf {0.41 \pm 0.06}\)

\(\mathbf {0.40 \pm 0.05}\)

PlantSeg

\(0.53 \pm 0.08\)

\(0.38 \pm 0.07\)

\(0.29 \pm 0.06\)

\(0.32 \pm 0.05\)

  1. The best performance (within 0.03) is in bold. The method that achieves the best performance under the majority of the metric is marked with \(^*\)