Clinical Gait Evaluation of Patients with Lumbar Spine Stenosis

Abstract

Objective

The third generation Intelligent Device for Energy Expenditure and Activity (IDEEA3, MiniSun, CA) has been developed for clinical gait evaluation, and this study was designed to evaluate the accuracy and reliability of IDEEA3 for the gait measurement of lumbar spinal stenosis (LSS) patients.

Methods

Twelve healthy volunteers were recruited to compare gait cycle, cadence, step length, velocity, and number of steps between a motion analysis system and a high‐speed video camera. Twenty hospitalized LSS patients were recruited for the comparison of the five parameters between the IDEEA3 and GoPro camera. Paired t‐test, intraclass correlation coefficient, concordance correlation coefficient, and Bland–Altman plots were used for the data analysis.

Results

The ratios of GoPro camera results to motion analysis system results, and the ratios of IDEEA3 results to GoPro camera results were all around 1.00. All P‐values of paired t‐tests for gait cycle, cadence, step length, and velocity were greater than 0.05, while all the ICC and CCC results were above 0.950 with P < 0.001.

Conclusions

The measurements for gait cycle, cadence, step length, velocity, and number of steps with the GoPro camera are highly consistent with the measurements with the motion analysis system. The measurements for IDEEA3 are consistent with those for the GoPro camera. IDEEA3 can be effectively used in the gait measurement of LSS patients.

Introduction

Lumbar spinal stenosis (LSS) has an annual incidence of approximately 1 per 20 000 population and is the most common reason for spinal surgery1. LSS is the main cause for lower back and lower limb pain, and can lead to neurological symptoms. The predominant symptom is intermittent claudication2. Doctors usually make the final diagnosis and decide the treatment based on the physical examination, CT and MR images, as well as the history and presentation of neurogenic claudication that can be evaluated by gait analysis2, so the quantification of the gait is significantly important.

Gait analysis has been regarded as an effective and concurrent diagnostic tool for assessing surgical outcome3 and prescribing appropriate rehabilitation treatments in patients with LSS4, 5, 6. Papadakis et al. report that LSS patients show higher gait variability compared with healthy subjects7. However, gait variability is reduced after surgery in LSS patient8. Cadence, stride length, and speed are general gait parameters that give a general idea of how well a patient can walk9. Desrosiers et al. (2014) report that cadence, stride length, and speed in spinal cord injury patients are lower than those in healthy people at natural speeds10. Conrad et al. (2013) conclude that the stride length and gait velocity are strongly correlated with the Oswestry Disability Index but weakly correlated with pain score6. The pain score is related to individually subjective factors, while gait analysis evaluates the walking function objectively. Therefore, accurate measurements of gait cycle, cadence, step length, velocity, and number of steps are important for monitoring the walking functions of LSS patients and progress of treatment based on the previous studies2, 9, 11, 12, 13.

Traditional gait measurement methods, such as video tape analysis, optical motion tracking and analysis systems, force plates, and a variety of electromyography methods are used to carry out quantitative gait analysis and to determine the pathological impact on gait characteristics14, 15, 16, 17. In clinical practice, gait is commonly estimated visually or evaluated using questionnaires, which produce subjective and mainly pain‐related functional results9. Gait laboratory measurements present difficulties in clinical application, owing to the high cost, need for a specific space, technology requirements, tedious data analysis work, and special transportation of patients. Perhaps most importantly, analyzing a single stride or limited data collected in the gait laboratory may not catch enough statistically and clinically valuable information. Clinicians and scientists have been calling for a portable, reliable, low‐cost, and time‐saving gait analysis system for decades. In recent years, wearable sensors like accelerometers and gyroscopes have enabled us to record spatial temporal, kinematic and kinetic data, and to analyze the gait outside conventional gait laboratories14, 18, 19, 20, 21, including the quantification of gait and movement disorders22, 23.

The Intelligent Device for Energy Expenditure and Activity (IDEEA, MiniSun, LLC, Fresno, CA, USA) is a portable device that uses accelerometers and gyroscopes to measure the type, duration, intensity, and frequency of physical activities and gait parameters. The IDEEA has been widely used to measure the spatial temporal gait variables in healthy people and orthopaedic patients15, 16, 17, 24, 25, 26, 27, 28, 29. The reliability of IDEEA in the quantification of step parameters during walking among healthy people has been demonstrated24, 25, 26. However, the reliability and validity of IDEEA2 in clinical orthopaedic patients is not consistent15, 16, 17, 27, 28, 29. The third generation of IDEEA (IDEEA3) has been developed with the improved gait measurement of orthopaedic patients.

This study was designed to evaluate accuracy and reliability of IDEEA3 measurements of gait cycle, cadence, step length, velocity, and number of steps in LSS patients with a portable high‐speed video camera. As the motion analysis system has long been regarded as a reliable gait measurement method for limited distance30, the high‐speed video camera was first validated by a motion analysis system to insure the accuracy. Data analysis using a high‐speed video camera was tedious and time‐consuming, so we used the camera to validate the accuracy of IDEEA3, which was more convenient than the video camera measurement.

Materials and Methods

Participants

Twelve healthy volunteers (age: 28.4 ± 4.2 years, height: 170.0 ± 5.2 cm, weight: 66.6 ± 10.5 kg) and 20 LSS patients (age: 58.0 ± 9.1 years, height: 167.2 ± 8.0 cm, weight: 70.7 ± 12.1 kg) with intermittent claudication symptom caused by back pain or leg pain were involved in this study. Diagnosed by a senior spine specialist according to the patients’ CT and MR images, all patients suffered only from the mentioned LSS diseases without other apparent problems. Their Japanese Orthopaedic Association Scores were between 2 and 6. There were no indications for surgical treatment, and both the straight leg raising test (SLR) and the strengthened straight leg raising test were negative. Subjects were excluded if they could not walk 20 m without assistance.

The healthy volunteers were recruited for the measurement comparison between a high‐speed video camera (GoPro, San Mateo, CA, USA) and a motion analysis system (Motion analysis, USA) to confirm that a GoPro camera can be used reliably for gait measurement. Hospitalized LSS patients were recruited for the measurement comparison between IDEEA3 and a GoPro camera to evaluate whether IDEEA3 can be effectively used for gait measurement in LSS patients. Informed consent was obtained before the experiment.

Devices

The 3‐D digital optical motion capture system (Motion analysis, USA) can non‐invasively measure and record movement with extreme accuracy in real time. It has a sampling rate of 120 frames per second. Fifteen markers were attached to the volunteers’ skin and six motion capture cameras were set to record their motion. Two markers were fixed on the right and left anterior superior iliac spines (ASIS). One other marker was placed on the superior aspect at the L_5–sacral interface. The other 12 markers were placed at the following locations: the medial/lateral femoral epicondylus lateralis and medial/lateral fibular malleolus lateralis for both legs; the space between the second and third metatarsal heads of both feet; the right and the left heels; and 4 markers were placed on the mid‐thigh and the mid‐shank for both legs. These 15 markers were used for measuring the lower extremity kinematics (Fig. 1A).

Figure 1.

Gait analysis experiment. (A) Experiment of healthy participant using the high‐speed video camera (GoPro) and motion capture system in university gait laboratory. (B) Marked carpet along the hospital corridor. (C) Experiments of lumbar spinal stenosis (LSS) patients using the IDEEA3 system and the high‐speed video camera (GoPro).

IDEEA3 is a precise time‐based system with multiple sensors, which worked relying on a mathematical model and algorithms. IDEEA3 is composed of sensors and recorders with large data storage capacity. Seven tiny 3D accelerometers/inclinometers (1.4 cm × 1.1 cm × 0.3 cm in size) were fixed to the skin by medical tape. One sensor was taped to the chest (4 cm below the jugular notch, vertical alignment); two sensors to each thigh (mid‐way between the patella and the anterior–superior iliac spine, vertical alignment); two sensors taped proximal to the ankles, laterally; and the other two sensors to the bottom of the feet (2 cm proximal to the head of the fourth metatarsal; horizontal alignment). Three sensors on the chest and thighs were directly connected to the recorder by thin (1.1 mm in diameter) and flexible cables. For consistency, the placement of the sensors was performed by the same individual for all trials and participants. Each participant sat in an upright position with the thighs parallel to the floor and hips, knees, and ankles close to 90°, while the system performed a baseline data collection. Afterwards, the IDEEA3 started recording continuously throughout the test. All acquired data were downloaded to a computer for data analysis (Fig. 1B).

The GoPro Hero3 high‐speed camera (GoPro, San Mateo, CA, USA) has a sampling rate of 60 frames per second. The GoPro Hero camera, a portable shockproof camera, is useful for personal instructional videos31. It was attached to the participants using a custom‐designed light frame behind the waist. Two miniature light‐emitting‐diodes (1.0 mm in diameter) were attached to the heel‐ground edges of soles. When the subjects walked on a carpet, the GoPro Hero camera traced the trajectory of heel motion and the moment of initial contact of each foot, while gait information was recorded simultaneously by the motion capture system or the IDEEA3 system (Fig. 1C).

Experimental Protocols

Experiment 1: Comparison of Measurements from GoPro and the Motion Capture System

Although video tape analysis has been widely used for gait analysis for decades, to ensure our measurement by GoPro camera was of high accuracy, an experiment was designed to compare the GoPro camera with a 3D motion capture system that is considered the gold standard for the gait measurement. The GoPro camera was fixed to a rigid light stick, which was attached to the pelvis through a large soft patch (0.15 m × 0.2 m). To validate the accuracy of the GoPro camera measurement method, 12 healthy volunteers walked in at slow, normal and fast speed using a GoPro camera and the motion capture system, barefoot and in flat shoes, respectively. The special‐made carpet, which uses high strength materials with low thermal expansion, was marked with precision grids (34 ± 1 mm/grid, total errors<5 mm for the entire 20 m length by three different rulers). The variation trends of data were checked to guarantee the synchronization. Because we only need to know when and where the heel contacts the floor for the five parameters (gait cycle, cadence, step length, velocity and number of steps), step length was recorded by the relative displacement of 2 feet according to the mark of the carpet in the frames. The accuracy is approximately 2 pixels on the frame, with less than 0.002 m error in displacement (Fig. 1).

Experiment 2: Comparison of Measurements from GoPro and IDEEA3

All patients wore IDEEA3 with the GoPro camera tied behind their pelvis, which recorded the walking gait. Each participant walked on a 20m × 1.5m long marked carpet in the hospital corridor in a normal and relaxed way (Fig. 1). After the experiments, data from IDEEA3 and the videos from the GoPro camera were loaded to the computer for analysis. To obtain the synchronized kinematics parameters data through IDEEA3 and camera measurements, we started the two systems simultaneously.

The experiments were conducted in a bright environment with low noise and no visual distraction to ensure participants’ safety, and participants wore comfortable clothing and flat shoes, thus following the protocol recommendations by Kressig et al. 32, that measurements should be performed in a reproducible, well‐lit, quiet environment, and the clothing of participants should be comfortable with appropriate types of footwear with heel height not exceeding 3 cm.

The initial contact moment of each participant’s walking was analyzed by video tape replay methods using the software GoPro studio (GoPro, San Mateo, CA, USA). In addition, the step length in the video was determined by the exact locations of the heel (the bright light‐emitting‐diode marker) on the carpet grid during foot–ground contact (accuracy to 3 mm based on the number of 34mm grids between steps and the locations within the grid on the video). Gait cycle from video was determined by the precise time period between initial contact of the same foot; cadence was determined by the inverse of the time period between adjacent initial contact; and velocity was obtained from dividing the gait cycle by the stride length.

To ensure participants walked naturally at their usual pace, data from the first and last 2 m of the testing session were not used, according to conventions set by Perry et al. 33, leaving 16 m of walking for gait analysis. The mean and standard deviation of the five variable results were calculated.

Statistical Analysis

The means of the 12 healthy participants’ data from GoPro camera measurement and the motion capture system were compared using paired t‐tests to confirm whether the two measurements were consistent. The intraclass correlation coefficient (ICC) concordance and correlation coefficient (CCC) were used to estimate the consistency of the two measurements. The ICC model is a two‐way mixed effects model where people effects are random and measurement effects are fixed34. The paired t‐test was used to compare the means between the results of IDEEA3 measurement and the GoPro camera measurement in 20 LSS participants. The ICC, CCC, Bland–Altman plot, and scatter plot were used to estimate the consistency of the two measurements.

The average errors were calculated to describe the deviations of two measurements. The error for each parameter of each subject is defined as:

$$\text{Error}\left(\%\right)=\frac{\sum \text{Motion}\text{results}-\text{GoPro results}}{\sum \text{Motion\hspace{0.17em}\hspace{0.17em}results}}\times 100\%$$

$$\text{Error}\left(\%\right)=\frac{\sum \text{GoPro}\text{results}-\text{IDEEA results}}{\sum \text{GoPro\hspace{0.17em}results}}\times 100\%.$$

The N is the total gait cycles of a patient. The average error of each parameter in this study is obtained by the average of errors derived from the above equation for all patients during the 16‐m tests.

SPSS Version 21.0 (SPSS, Inc., Chicago, IL, USA) statistical package was used for statistical analysis. Statistical significance is defined as P < 0.05.

Results

Comparison of Measurements from the GoPro and the Motion Capture System

The results for the GoPro camera and the motion capture system in 12 healthy subjects under normal speed with bare feet were 1.077 ± 0.699 s and 1.082 ± 0.637 s for gait cycle, 112.134 ± 7.260 steps/min and 111.343 ± 6.751 steps/min for cadence, 0.625 ± 0.454 m and 0.625 ± 0.513 m for step length, and 1.170 ± 0.127 m/s and 1.158 ± 0.109 m/s for velocity, respectively. The results of the GoPro camera and the motion capture system in 12 healthy subjects under normal speed with flat shoes were 1.119 ± 0.066 s and 1.121 ± 0.071 s for gait cycle, 107.573 ± 6.524 steps/min and 107.626 ± 7.102 steps/min for cadence, 0.646 ± 0.053 m and 0.646 ± 0.053 m for step length, and 1.161 ± 0.100 m/s and 1.160 ± 0.106 m/s for velocity, respectively. The counts of steps for the two measurements were both 11.160 ± 1.200. All of the ratios were above 0.998 (Table 1).

Table 1.

Paired t‐test results/CCC/ICC results of high‐speed camera (GoPro) and motion capture system in healthy volunteers

Speed	Parameters	High‐speed camera results (a)	Motion capture system results (b)	Ratio (a/b)	t‐test (P value)	CCC (r value)	ICC (r value)
Barefoot normal speed	Gait cycle	1.077 ± 0.699	1.082 ± 0.637	0.995	0.50	0.934*	0.965*
	Cadence	112.134 ± 7.260	111.343 ± 6.751	1.007	0.35	0.923*	0.959*
	Step‐length	0.625 ± 0.454	0.625 ± 0.513	1.000	0.93	0.974*	0.985*
	Velocity	1.170 ± 0.127	1.158 ± 0.109	1.010	0.33	0.957*	0.972*
Barefoot fast speed	Gait cycle	0.888 ± 0.620	0.892 ± 0.585	0.996	0.32	0.972*	0.985*
	Cadence	135.725 ± 9.491	135.161 ± 9.075	1.004	0.37	0.975*	0.987*
	Step‐length	0.767 ± 0.070	0.768 ± 0.719	0.999	0.63	0.998*	0.999*
	Velocity	1.750 ± 0.240	1.747 ± 0.238	1.002	0.71	0.993*	0.995*
Barefoot slow speed	Gait cycle	1.231 ± 0.077	1.232 ± 0.075	0.999	0.76	0.977*	0.995*
	Cadence	97.938 ± 6.166	97.720 ± 6.107	1.005	0.43	0.980*	0.994*
	Step‐length	0.546 ± 0.066	0.546 ± 0.065	1.000	0.23	0.999*	0.999*
	Velocity	0.892 ± 0.121	0.890 ± 0.122	1.002	0.21	0.998*	0.999*
Shoes normal speed	Gait cycle	1.119 ± 0.066	1.121 ± 0.071	0.998	0.40	0.996*	0.997*
	Cadence	107.573 ± 6.524	107.626 ± 7.102	0.999	0.84	0.996*	0.996*
	Step‐length	0.646 ± 0.053	0.646 ± 0.053	1.000	0.63	0.997*	0.998*
	Velocity	1.161 ± 0.100	1.160 ± 0.106	1.001	0.50	0.996*	0.997*
Shoes fast speed	Gait cycle	0.913 ± 0.061	0.916 ± 0.065	0.997	0.46	0.977*	0.988*
	Cadence	131.991 ± 9.243	131.625 ± 9.748	1.003	0.53	0.980*	0.990*
	Step‐length	0.805 ± 0.086	0.806 ± 0.085	0.999	0.57	0.999*	0.999*
	Velocity	1.783 ± 0.274	1.776 ± 0.275	1.004	0.29	0.996*	0.996*
Shoes slow speed	Gait cycle	1.306 ± 0.131	1.304 ± 0.133	1.002	0.50	0.998*	0.998*
	Cadence	92.727 ± 9.002	92.865 ± 9.233	0.999	0.54	0.998*	0.998*
	Step‐length	0.574 ± 0.087	0.572 ± 0.082	1.003	0.29	0.999*	0.999*
	Velocity	0.887 ± 0.148	0.886 ± 0.148	1.001	0.75	0.999*	0.999*

Units in Table 1: Seconds are used for cycle, steps/min for cadence, meter for step length, and m/s for velocity

CCC, concordance correlation coefficient; ICC, intraclass correlation coefficient (ICC)

^* P = 0.001.

All the P‐values of the paired t‐test for gait cycle, cadence, step length, and velocity between results of the GoPro measurement and the motion capture system were above 0.05 in healthy participants with bare feet and flat shoes for normal speed, fast speed, and slow speed, while all the ICC results were above 0.950 for normal speed, fast speed and slow speed. The CCC results were above 0.920 for normal speed, fast speed, and slow speed (Table 1).

Comparison of Measurements from GoPro and IDEEA3

The results for the IDEEA3 system and the GoPro camera in 20 LSS subjects under normal speed with flat shoes were 1.111 ± 0.109 s and 1.113 ± 0.109 s for gait cycle, 109.185 ± 10.537 steps/min and 109.219 ± 10.499 steps/min for cadence, 0.557 ± 0.085 m and 0.559 ± 0.088 m for step length, and 1.023 ± 0.215 m/s and 1.025 ± 0.222 m/s for velocity, respectively. The counts of steps for the two measurements were both 94.000 ± 36.245. All of the ratios were above 0.996 (Table 2).

Table 2.

t‐test results/CCC/ICC of results of IDEEA3 results and high‐speed camera (GoPro) results in 20 Spine patients

Parameters	IDEEA results (a)	GoPro results (b)	Ratio (a/b)	t‐test (P value)	CCC (r value)	ICC (r value)	B‐A Plot results (%)
Gait cycle (s)	1.111 ± 0.109	1.113 ± 0.109	0.998	0.10	1.000*	1.000*	95.00
Cadence (steps/min)	109.185 ± 10.537	109.219 ± 10.499	1.000	0.79	0.999*	0.999*	90.00
Step‐length (m)	0.557 ± 0.085	0.559 ± 0.088	0.996	0.66	0.985*	0.993*	95.00
Velocity (m/s)	1.023 ± 0.215	1.025 ± 0.222	0.998	0.79	0.991*	0.995*	95.00
Steps	94.000 ± 36.245	94.000 ± 36.245	1.000	‐	1.000	1.000	100.00

^* P = 0.001

CCC, concordance correlation coefficient; ICC, intraclass correlation coefficient.

All the P‐values of paired t‐tests for gait cycle, cadence, step length, and velocity between results of the GoPro camera measurement and the IDEEA3 measurement were above 0.05, except that the P‐value for the gait cycle parameter was 0.048 in 20 LSS patients, which will be discussed later. All the ICC results were above 0.990, while the CCC results were above 0.985 (Table 2). Bland–Altman plots showed that 95.00%, 90.00%, 95.00%, 95.00% and 100% of the 20 LSS patients were within the 95% confidence interval of the agreement limit for the five parameters. Scatter plots are shown in Fig. 2.

Figure 2.

Scatter plots of the motion analysis measurement results and GoPro measurement results. The dotted line represents the identity line: y = x. Figure (A) and (B) are scatters of cadence and velocity results of healthy subject results. (C) and (D) are scatters of cadence and velocity results of LSS patients. The figure of cycle results is close to the cadence results, and the figure of step length results is close to the velocity results.

The average errors (%) of IDEEA3 measurement compared with those from video camera measurement in gait cycle, cadence, step length, and velocity were 0.142, 0.034, −0.179, and −0.027, respectively (Fig. 3).

Figure 3.

Description of all 20 lumbar spinal stenosis (LSS) patients measured by IDEEA3 and GoPro camera. Each of the error bars above the data point represents the intra‐individual standard deviation of the errors between two methods for a particular patient (multiple steps). The error bars below the gait parameters (very close to parameter value) indicate the intra‐individual standard errors between the two measurements for that patient. The assumed ID from 1 to 20 represents the data of LSS patients. The figure of cycle results is close to cadence results, and the figure of step length results is close to the velocity results.

P‐values of independent t‐test results for gait cycle, cadence, step length, and velocity were 0.858, 0.629, 0.004 and 0.025 in the normal group and LSS group, respectively.

Discussion

Measurements from the GoPro camera are highly consistent with measurements from the motion analysis system. As the motion analysis system has long been regarded as a reliable measurement method30, we conclude that the measurements of gait cycle, cadence, step length, velocity, and number of steps by the GoPro camera are reliable.

The paired t‐test results inferred that we could not demonstrate significant differences between the GoPro measurements and IDEEA3 measurements in the LSS group. As the GoPro measurements are proved reliable, we conclude the IDEEA3 measurements in LSS patients are reliable.

All ICC and CCC results in the five parameters of these groups were greater than 0.95, which indicates that the two measurement methods are consistent with each other. The Bland–Altman results were all greater than 90%. The largest deviations inside the limits of agreement were −0.023 s for gait cycle, −1.251 steps/min for cadence, 0.031 m for step length, and −0.050 m/s for velocity. As such deviations are acceptable in clinical application, the two measurements were considered consistent.

Although patients’ ambulatory functions can be evaluated by referring to the range of gait parameters in healthy individuals, reliable gait parameter measurements can provide more precise and valuable information for clinical diagnosis35, 36. Bohannon et al. studied the comfortable gait speed of adults aged between 20 and 79, and reported that mean comfortable gait speed for men and women ranged between 1.33–1.46 m/s and 1.27–1.41 m/s35. In our study, the comfortable speed results for men and women were 1.20 ± 0.13 m/s and 1.09 ± 0.09 m/s for normal subjects, and 0.99 ± 0.18 m/s and 1.01 ± 0.24 m/s for LSS patients, respectively. These results show that the velocity in the LSS patients was lower than that in healthy people in this study and other studies. It has been reported that gait parameters are less favorable in spine cord injury patients than healthy participants at natural gait speeds10, so the observation supports our results. Conrad et al. report that the velocity of LSS patients was 1.01 ± 0.33 m/s for men and 0.75 ± 0.24 m/s for women6, which concurs with the results in this study.

Macleod et al. report the ICC values from normal adult spinal patients measured using Wi‐GAT, which is another portable tool and a 3‐D motion analysis system37. The ICC values were 0.973 (left foot) and 0.975 (right foot) for stride length; 0.996 (left foot) and 0.981 (right foot) for stride duration; 0.996 (left foot) and 0.985 (right foot) for cadence; and 0.977 (average of left foot and right foot) for walking speed37. The error (%) was −2.005 (left foot) and −1.805 (right foot) for stride length; 0.262 (left foot) and 0.626 (right foot) for stride duration; −0.245 (left foot) and −0.550 (right foot) for cadence; and −2.196 (mean of left and right foot) for walking speed37. In our study, the statistical values for the gait cycles are reported (mean of left and right feet). The ICC values of IDEEA3 and camera measurements for the LSS group were: 1.000 for gait cycle, 0.999 for cadence, 0.993 for step length, and 0.995 for velocity (Table 2). In addition, the error (%) results (mean of left and right feet) were 0.142 for gait cycle, 0.034 for cadence, −0.179 for step length, and −0.027 for velocity (Fig. 3). The portable gait analysis device in this study on LSS patients was more accurate than that in Macleod’s study on normal adults. The P‐values of the independent t‐test for the normal group with flat shoes and the LSS group showed that only the step length and velocity were significant different (Tables 1 and 2). The means and standard deviations of the step length and velocity for the normal group were both larger than for the LSS group. This suggests that the LSS patients walk with moderate cadence but shorter step length and slower velocity, which might be caused by pain or neural problems as a result of LSS. The shorter step length and slower velocity might help patients to keep stable.

There are some limitations in this study. The five gait parameters examined in this study are essential to clinically evaluate patients’ gait. However, more gait parameters, such as gait phases and joint angle, still need to be analyzed for better assessment of patients’ gait in the future. Although the statistic power is 0.80, it would be ideal to have a larger sample size.

Conclusion

Lumbar spinal stenosis patients walk with moderate cadence but shorter step length and slower velocity compared with healthy subjects. The measurements for gait cycle, cadence, step length, velocity, and number of steps by GoPro camera are highly consistent with the measurements of the motion analysis system, so the GoPro camera can be reliably used for the measurement of these five parameters. The measurements by IDEEA3 are consistent with that of the GoPro camera; however, using IDEEA3 measurements is much more convenient than using a GoPro camera. In conclusion, IDEEA3 can be effectively used in the gait measurement of patients with LSS.