Skip to main content

Table 4 Summary of systematic errors in automated segmentation

From: Fully-automated segmentation of muscle and inter-/intra-muscular fat from magnetic resonance images of calves and thighs: an open-source workflow in Python

Error types

Test (freq)

Retest (freq)

Follow-up (freq)

Total (freq, %)

1) IMF merging with subcutaneous fat

2

2

2

6/92 = 6.5%

2) Motion streaks mistaken as fat

1

2

1

4/92 = 4.3%

3) Noise within lean muscle segmented as fat

1

0

0

1/92 = 1.1%

Total (freq, %)

4/57 = 7.0%

4/57 = 7.0%

3/35 = 8.6%

11/92 = 12.0%

  1. Frequencies of three error types are described: 1) IMF mistakenly labeled as subcutaneous fat due to poor discernibility of fascial boundaries. 2) Motion streaks appearing as fat were mistakenly segmented as IMF. 3) Noise with higher signal intensity within an otherwise lean muscle was erroneously segmented as IMF since ITSA searches for a bimodal distribution of fat and muscle. See Fig. 6 for examples of each
Back to article page

Skeletal Muscle

ISSN: 2044-5040

Contact us