Table 2. Studies comparing inertial sensors with a video-based optoelectronic motion analysis system.

Study	Description of study	Body area	Type of sensor/portability = size	Accuracy of the sensor				Gold standard	Validity				Participants
					Yaw	Pitch	Roll			Yaw	Pitch	Roll
Plamondon et al. (2007)	The purpose of this study was to evaluate a hybrid system for the 3D measurement of trunk posture in motion.	T (TT, P)	Microstrain 3DM-G, Burlington weight 40 g. 64×64×25 mm	Global angles: P	2.0±0.5	0.5±0.2°	0.7±0.2°	Optoelectronic system (Optotrak 3020, Northern Digital Inc., Waterloo, Ont,, Canada)	Global angles: P (CMC)	0.998	0.974	0.975	n = 6 (6 male)Age (32±12 years)
				Global angles: TT	1.9±0.6	0.8±0.2°	0.7±0.1°		Global angles: TT (CMC)	0.988	0.993	0.971
				Relative angles: P/TT	2.2±0.4	1.1±0.4°	1.6±0.8°		Relative angles: P/TT (CMC)	0.657	0.987	0.953
Jasiewicz et al. (2007)6	The aim of this study was to determine the accuracy of new generation sensorsof wireless orientation.	T (CT)	Inertial Cube 3 sensor (Intersense, Bedford, MA, USA)/26.2×39.2×14.8 mm	Head mounted sensors	2.3±0.9	2.1±1.1°	2.5±0.9°	The 3-Space Fastrak (Polhemus, Colchester, VT, USA)	Head mounted sensors (cross-correlation)	0.97	0.98	0.97	n = 10 (mean age 33.4±9.9 SD, range 20–51 years)
				C7/Trunk mounted sensors	0.9±0.5	1.2±0.5°	0.7±0.7°		C7/Trunk mounted sensors (cross-correlation)	0.98	0.98	0.99
Bourke et al. (2008)33	This study investigates distinguishing falls from normal activities of daily living by thresholding of the vertical velocity of the trunk.	T	ADXRS300 (Gyro) and ADXL210E (accel)/12×12×5 mm	RMS (M±SD): STSI = 0.09±0.05; Kneeling = 0.102±0.04; Object picking = 0.95±0.03; Lying on floor = 0.15±0.05; W = 0.08±0.03; Coughing = 0.06±0.02; Forward fall/knee FLX = 0.13±0.03; Side-fall right/Knee FLX = 0.15±0.09; Backward fall = 0.11±0.05				Optical motion capture system (6 cameras)	CMC (M±SD): STSI = 0.98±0.02; Kneeling = 0.96±0.03; Object kicking = 0.96±0.02; Lying on floor = 0.96±0.03; W = 0.89±0.07; Coughing = 0.73±0.29; Forward fall/knee FX = 0.98±0.01; Side-fall right/Knee FX = 0.98±0.02; Backward fall = 0.98±0.98				n = 5 (5 male)Age (25.6±1.9 years)
O’Donovan et al. (2007)36	The technique presented in this paper is concerned with ankle joint angles measurement.	LL (ankle)	ADXL210E (accel) ADXRS150 (Gyro) HMC2003 (mag) 60×40×24 mm	Angular errors in the measurement	3.33°	0.49°	–	Optoelectronic system (Evart 3D)					n = 2 (2 males)Age (25 and 23 years)
Picerno et al. (2008)11	This paper describes an anatomical calibration technique for three wearable inertial and magnetic sensing modules using palpable anatomical landmarks.	LL (hip, knee, ankle)	MTx (Xsens Technologies, The Netherlands)/weights 30 g. 38×53×21 mm	Hip absolute value (M±SD).	6.7±6.1	1.8±0.7°	3±2.2°	Optoelectronic system (Vicon Mx cameras, Oxford Metrics, UK)	The correlation coefficient for the FLX/EXT was equal to 1 for all the joints whereas the ÄRoM was less than 0.5°. The lowest R was the knee IER, and it was equal to 0.942				n = 1
				Knee absolute value (M±SD)	6.3±3.9	1.9±0.7°	4.6±1.1°
				Ankle absolute value (M±SD)	8.3±1.6	1.3±0.9°	5.7±1.5°
Martin-Schepers et al. (2010)9	This study proposes and evaluates an alternative algorithm for relative position and orientation. A complementary Kalman filter structure was presented.	TT, UL, LL	MTx (Xsens Technologies, The Netherlands)/weights 30 g. 38×53×21 mm	Orientation error: TT	4.3±0.3	4.5±0.7°	–	Optoelectronic system(Vicon, Oxford Metrics, UK)					n = 5
				Orientation error: UL	–	–	2.8±0.7°
				Orientation error: LL	–	3.6±0.9°	–
Wong and Wong (2008)8	The aim of this study was to introduce accelerometers and gyroscopes to detect posture in the sagittal and coronal planes.	TT (TT, LT, P)	KXM52-Tri-axis Kionix (Aceel) and Epson gyroscopes (Gyros)/22×9.20×9.12 mm, Weights 6 g	Peak value TT (degrees±SD)	–	22.8±11.1	3.8±1.5	Optoelectronic system (Vicon 370, Oxford Metrics, UK)	Correlation coefficient TT±SD	–	0.983±0.014	0.829±0.308	n = 5 (4 female and 5 male, age: 25.2±4.8 years, weight: 50.5±7.2 kg, height: 1.7±0.09 m)
				Peak value LT (degrees±SD)	–	24.7±7.0	6.2±2.2
									Correlation coefficient LT±SD	–	0.981±0.014	0.984±0.015
				RMS angular velocity (deg s−1±SD)	–	6.3±3.0	4.5±1.3
Zhou et al. (2008)32	This paper presents a new human motion tracking system that is placed near the wrist and elbow joints.	Upper limb (shoulder, elbow, wrist)	MT9B (Xsens Technologies, The Netherlands)/weights 38 g. 39×54×28 mm	RMS elbow angles (degrees)	4.83	2.41	–	Optical motion tracker (CODA, Charnwood, UK)	Correlation coefficients in elbow	0.94	0.98	–	n = 4 (age range: 20–40 years)
Zhou and Huosheng (2007)13	A novel motion tracking prototype will be developed on the basis of the previously designed motion detector.	Upper limb (shoulder, elbow, wrist)	MTx (Xsens Technologies, The Netherlands)/weights 30 g. 38×53×21 mm	Arm position RMS (m)	–	0.004	0.005	Optical motion tracker (CODA, Charnwood, UK)	Correlation coefficients in arm	–	0.97	0.97	n = 4 (age range: 27–40 years)
Musić et al. (2008)12	Model validation was performed on simulated data and on measurements acquired with the Optotrak optical motion analysis system.	T, LL		Average self-selected STSI speed (Shank)	–	3.6°	–	Optotrak 3010 optical motion capture system (Northern Digital Inc., Waterloo, Ont., Canada),					n = 1
				Average self-selected STSI speed (Thigh)	–	5.2°	–
				Average self-selected STSI speed (HAT)	–	5.8°	–
Roetenberg et al. (2007)7	The objective of this study is to design and evaluate a new system for ambulatory measurements of position and orientation on the body.	T (TT), UL	MTx (Xsens Technologies, The Netherlands)/weights 30 g. 38×53×21 mm	Orientation error: TT	2.6±0.5	2.4±0.5°	2.6±0.5°	Optoelectronic system (Vicon 460, Oxford Metrics, UK)					n = 1
				Position error (mm): TT	4.9±1.0	4.8±1.1	5.0±0.9°
				Orientation error: UL	–	2.4±0.5°	2.3±0.5°
Goodvin et al. (2006)5	They propose a new method for accurately measuring the real-time orientation and position of the spine in a portable, non-invasive, and clinically meaningful manner.	T (CT, TT, LT)	MT9B (Xsens Technologies, The Netherlands)/weights 38 g. 39×54×28 mm	Cervical average deviation	0.2°	0.42°	0.1°	Optoelectronic system (Vicon 460, Oxford Metrics, UK)					n = 5
				Torso average deviation	0.23°	0.06°	0.03°
				Hip average deviation	1.35°	0.33°	3.1°
Zhou and Hu (2010)10	This paper presents the effects of changes in error reduction by using Kalman filtering.	Upper limb (shoulder, elbow, wrist)	MTx (Xsens sTechnologies, The Netherlands)/weights 30 g. 38×53×21 mm	Statistical error before Kalman filter	–	14.62°	14.02°	Optical motion tracker (CODA, Charnwood, UK)					n = 4
				Statistical error after Kalman filter	–	2.13°	2.01°
Lee et al. (2010)38	In this study they present sensor nodes (accel) with a goniometer probe.	UL	Freescale MMA7261QT (accel)/6×6×1.45 mm	A linear increasing trend from 0±2.5° at a mean angular speed of 10° s−1 to 3.5±7° at 80° s−1.				Goniometer probe (PS-2137 from PASCO)					n = 1

Note: ADL, activities of daily live; CMC, coefficient of multiple correlation; M, mean; SD, standard deviation; STSI, sit-to-stand; FX, flexion-extension; FLX, flexion; EXT, extension; ABD, abduction; ADD, adduction; IER, internal–external rotation; PT, protraction; RT, retraction; MLR, medio-lateral rotation; APT, anterior–posterior tilting; RMS, root mean square; ACRL, angular coefficient of the regression line; IQR, inter-quartile ranges; P, pelvis; LB, lateral bending; R, rotation; TT, thoracic trunk; UL, upper limb; LL, lower limb; G, gait; CT, cervical trunk; LT, lumbar trunk; T, trunk.