Skip to main content

Table 2 Statistical comparison of the values of reference manual parameter measurements (GT) and those obtained automatically by the prediction LanDet model (PR) for the combined dataset

From: Landet: an efficient physics-informed deep learning approach for automatic detection of anatomical landmarks and measurement of spinopelvic alignment

 

MAE (±SD)

Mean PR (±SD)

Mean GT (±SD)

Correlation Analysis

Acc (%)

    

R

p

 

SS\(^\circ \)

3.7 (2.7)

39.8 (9.4)

37.9 (10.3)

0.89

<0.0001

93.1

PT\(^\circ \)

1.3 (1.1)

19.2 (13.3)

20.2 (12.9)

0.98

 

94.6

PI\(^\circ \)

4.2 (3.1)

53.7 (12.4)

52.8 (12.3)

0.93

 

93.4

LL\(^\circ \)

5.1 (6.4)

38.3 (19.5)

40.3 (16.7)

0.83

 

91.2

SVA\(_{mm}\)

2.1 (1.9)

3.5 (3.1)

3.3 (2.9)

0.96

 

94.5

  1. The mean average error (MAE) are presented along with standard deviations (SD) of each parameter evaluated in the two datasets (DS1 & DS2). The table also include the correlation analysis using R Pearson correlation coefficients, and relative accuracy of prediction (Acc)