Evaluation of the Validity, Reliability, and Kinematic Characteristics of Multi-Segment Foot Models in Motion Capture

Abstract

This study aimed to evaluate the validity and reliability of our new multi-segment foot model by measuring a dummy foot, and examine the kinematic characteristics of our new multi-segment foot model by measuring the living body. Using our new model and the Rizzoli model, we conducted two experiments with a dummy foot that was moved within a range from -90 to 90 degrees in all planes; for the living body, 24 participants performed calf raises, gait, and drop jumps. Most three-dimensional (3D) rotation angles calculated according to our new models were strongly positively correlated with true values (r > 0.8, p < 0.01). Most 3D rotation angles had fixed biases; however, most of them were in the range of the limits of agreement. Temporal patterns of foot motion, such as those in the Rizzoli model, were observed in our new model during all dynamic tasks. We concluded that our new multi-segment foot model was valid for motion analysis and was useful for analyzing the foot motion using 3D motion capture during dynamic tasks.

Keywords: bias; foot; kinematics; optical motion capture; reproducibility of results.

Figure 1

The process flow chart.

Figure 2

Marker location, segment reference planes (dash triangles), and the primary- and tertiary-axis (red solid arrows) on these planes are shown. Rear-foot segment references in the planes of two conditions; transverse plane (A) and sagittal plane (B).

Figure 3

Three infrared-reflecting markers (point o, point x, and point z) were attached, respectively, to the three plates that were placed on each dummy foot to calculate the true value. The X and Z axes (red solid arrows) on these plates were shown. Angles for the plates were defined as the true values for the corresponding joint angles; the angle of plate (A) relative to plate (B) corresponded the Hallux relative to forefoot (Met_Hal) angle and the angle of plate (B) relative to plate (C) corresponded the forefoot relative to rearfoot (Cal_Met) angle. Measurement positions were from −90 to 90 degrees in all planes.

Figure 4

The right foot which attached the infrared-reflection markers, shown from the front (A), from the back (B), and from the outside (C).

Figure 5

Univariate analysis using Pearson’s correlation analysis. Left to right, the Met_Hal angle in the frontal, transverse, and sagittal planes are shown. Correlations were assessed between the true values and the calculated values according to the two models; our new model (first row) and the Rizzoli model (second row).

Figure 6

Univariate analysis using Pearson’s correlation analysis. Left to right, the Cal_Met angle in the frontal, transverse, and sagittal planes are shown. Correlations were assessed between the true values and the calculated values according to the three models, respectively; our new model (first row), our new model_2 (second row), and the Rizzoli model (third row).

Figure 7

Bland-Altman plots showing the differences between the true values of Met_Hal angle and Met_Hal angle calculated according to the two models: (A) our new model and (B) the Rizzoli model, against their means. When they do not have either a fixed or proportional bias, the mean is shown (one blue solid line). When they have a fixed or proportional bias, the mean (the middle one of three blue solid lines) and limits of agreement are shown (the outer two of three blue solid lines).

Figure 8

Bland–Altman plots showing the difference between the true values of the Cal_Met angle and the Cal_Met angle calculated according to the three models; (A) our new model, (B) our new model_2, and (C) the Rizzoli model, against their means. When they do not have either a fixed or proportional bias, the mean is shown (one blue solid line). When they have fixed or proportional biases, the mean (the middle one of three blue solid lines) and limits of agreement are shown (the outer two of three blue solid lines). When they have a proportional bias, the difference (Y-axis) calculates the relative value to the mean (X-axis).

Figure 9

Left to right, the temporal patterns of rotation in the frontal, transverse, and sagittal planes during the calf raise. New model (red solid line), New model_2 (blue solid line), and the Rizzoli model (pink dot line) are shown. First to third row in order, the three joint angles (Met_Hal, Cal_Met, and Sha_Cal) are shown.

Figure 10

Left to right, the temporal patterns of rotation in the frontal, transverse, and sagittal planes during gait. New model (red solid line), New model_2 (blue solid line), and the Rizzoli model (pink dot line) are shown. First to third row in order, the three joint angles (Met_Hal, Cal_Met, and Sha_Cal) are shown.

Figure 11

Left to right, the temporal patterns of rotation in the frontal, transverse, and sagittal planes during the drop jump. New model (red solid line), New model_2 (blue solid line), and the Rizzoli model (pink dot line) are shown. First to third row in order, the three joint angles (Met_Hal, Cal_Met, and Sha_Cal) are shown.