Skip to main content

Table 3 Deep learning segmentation performances

From: Automated pelvic MRI measurements associated with urinary incontinence for prostate cancer patients undergoing radical prostatectomy

 

DSC

MSD [mm]

HD95 [mm]

Volume [cc]

Coronal structures

 Prostate

0.97 (0.96–0.97)

0.22 (0.16–0.32)

1.25 (0.87–1.75)

40.93 (36.30–54.02)

 Urethra

0.85 (0.76–0.92)

0.14 (0.07–0.32)

1.25 (0.60–3.01)

0.20 (0.17–0.24)

 Corpus spongiosum

0.96 (0.96–0.97)

0.10 (0.09–0.12)

0.60 (0.42–0.63)

9.73 (8.04–12.47)

 OIM

0.95 (0.92–0.96)

0.25 (0.15–0.36)

1.25 (0.63–3.00)

38.17 (33.80–43.21)

 LAM

0.90 (0.86–0.91)

0.24 (0.17–0.33)

2.99 (0.88–3.00)

5.92 (5.08–6.83)

Sagittal structures

 Bladder

0.98 (0.97–0.98)

0.15 (0.11–0.23)

0.63 (0.63–1.20)

100.64 (73.47–148.26)

 Prostate

0.96 (0.94–0.97)

0.31 (0.22–0.47)

1.25 (1.12–1.90)

37.23 (32.52–46.61)

 Urethra

0.87 (0.77–0.90)

0.10 (0.05–0.14)

0.82 (0.63–1.88)

0.21 (0.18–0.24)

 Corpus spongiosum

0.94 (0.90–0.95)

0.22 (0.16–0.31)

1.88 (0.88–3.00)

10.84 (9.57–13.04)

  1. For the LAM and IOM: left and right are combined. Data are expressed as median (interquartile range). DSC Dice similarity coefficient, MSD Mean surface distance, HD95 95% boundary Hausdorff distance, LAM Levator ani muscle, OIM Obturator internus muscle