Integrated Assessment of Gait and Spinal Kinematics Using Optoelectronic Motion Analysis Systems: Validation and Usability Assessment of a Novel Protocol

Abstract

Background: Gait assessment is a complex task involving locomotion and balance control across all body segments, requiring a global analysis in the event of motor disorders. Among these are spinal disorders, where an understanding of spinal kinematics during walking is important to improve treatment decisions and outcomes. The technique of stereophotogrammetric motion analysis is currently the gold standard in this context. A new integrated protocol for whole-body kinematic gait analysis is proposed in this study, which takes into account the movements of the spine. Methods: A new protocol with 30 passive markers was developed to analyze gait. Of these markers, 22 implemented the Davis protocol for gait measurement, while the other 8 were placed onto the spine to record spinal movements. The protocol’s accuracy was assessed through comparisons of the constructive angles of a manikin replicating the human body and the angles measured with the optoelectronic system. An assessment of intra- and inter-operator repeatability and protocol usability was carried out by recruiting and applying the protocol in a population composed of ten subjects (mean age 17.36—SD 10.12) without any history of spine pathology. Results: The protocol was validated successfully. The validation accuracy was more than satisfactory: the measured RMSE was 1.2 ± 1° for the data collected with the optoelectronic system with respect to the manikin. The intra-operator repeatability was also good in the sagittal and frontal planes (average ICC > 0.867), and the inter-operator repeatability was moderate or good in all planes (average ICC > 0.77). The usability score obtained using the System Usability Scale was satisfactory (mean 74.75, SD 5.88). Conclusions: This study proposes a new protocol to assess total body kinematics, including the spine in its three main segments, during gait. The successful validation of this protocol in terms of reliability and usability allows for its subsequent clinical application.

1. Introduction

In recent decades, several technologies for the analysis of human movement have been developed in the clinical field. At present, marker-based optoelectronic systems represent the most significant technique, particularly for clinical/rehabilitation applications, providing the best metrological results in terms of accuracy, repeatability, and recording quality (i.e., frequency of acquisition of motor tasks) [1,2,3]. Stereophotogrammetric systems measure anatomical angles during the execution of activities of daily living (ADLs) or specific tasks in a laboratory through a non-invasive methodology [4,5,6,7]. These systems use optoelectronic cameras with infrared (IR) LED panels to illuminate markers on specific body landmarks, which reflect IR light and are recognized and tracked by the system. Then, a biomechanical model applied to their 3D trajectories allows for the computation of the anatomical angles representing human movement.

The first application of video and stereophotogrammetric technologies for the study of human movements was to gait [2]. Despite its apparent simplicity, locomotion is a complex task requiring the synergic functioning of the brain, musculoskeletal system, and sensorial system to correctly perform the motor task.

A functional assessment of locomotion should include both the forward movement and balance control acting in an integrated approach, with a simultaneous analysis of the contributions of lower limbs (for motion) and the spine and upper limbs (for balance and its control). From this perspective, most of the developed protocols for gait analysis only focus on lower limbs or on a limited vision of the whole body. The most frequently adopted stereophotogrammetric models [7,8] assess gait kinematics by considering the lower quarter (pelvis, hip, knee, ankle), and in few studies, the trunk is also included in the biomechanical model, but it is considered as a single rigid segment [9,10]. Indeed, the spine is the central supporting structure of the body, allowing for trunk movements in most ADLs [11,12], protecting the spinal cord, and supporting the head and upper limbs for sensory and motor functions. Therefore, evaluating the spine’s contributions to total body movements such as gait, both in its global and segmental contributions or the effects of spinal pathologies (e.g., scoliosis in the developmental stage), represents an important research and clinical perspective. Studying the synergic motion of the spine and other body segments is relevant for several clinical applications, such as the diagnosis of complex motor disorders, endoprosthesis design, and in rehabilitation for the evaluation of treatment outcomes.

A review of the literature showed that some studies have analyzed multi-segmental trunk kinematics in upright standing [13,14] or during elementary exercises [15,16,17], but not during gait, and with the important limitation of introducing a large number of skin markers on the body. This is relevant for the complexity of the procedures, their acceptance by patients and operators, and the reliability of the protocol in terms of repeatability. The same literature search provided some models that have been proposed for the study of spinal kinematics during gait, but with the relevant limitation of considering only a limited portion of the spine [18,19,20,21,22,23]; for example, considering only the thoracic column or the lumbar column. In addition, these studies often considered spinal movements in only one or two anatomical planes [24,25,26]. Only a few studies have considered the three main sub-segments of the spine (upper thoracic, lower thoracic, and lumbar) and their movements in the three anatomical planes [27,28]. These studies offer interesting biomechanical models for assessing trunk kinematics, but they do not provide information about the reliability of these protocols. Indeed, to our knowledge, no studies in the literature have systematically assessed the intra- and inter-operator reliability of spinal kinematics during walking.

This study proposes a new protocol to assess total body kinematics, including the spine in its three main segments (upper thoracic, lower thoracic, and lumbar spine), during walking. The protocol consists of a specific marker set implementing a more detailed biomechanical model of the human body for global kinematic analysis during gait. Starting from the Davis marker set [10], which is widely used in clinical practice for gait analysis, we add a limited number of additional skin markers on the trunk to better characterize the functionality of the three main segments of the spine in the sagittal, frontal, and transversal planes, allowing for the study of integrated motor strategies (balance control, compensation, global movements, etc.).

The aim of this paper is to describe how this protocol was developed and validated for kinematic measurements of the spine, with respect to its accuracy and repeatability, in order to facilitate its introduction into clinical practice.

2. Materials and Methods

2.1. Subjects and Sample Size Determination

The validation of a new protocol is mandatory before its application in clinical practice. Similar motion capture validation studies, particularly for spinal mobility assessments, typically evaluate 10 subjects [16,17,29], which is considered sufficient for accuracy, test–retest repeatability, and acceptance. This aligns with usability study guidelines that 10 users can identify over 80% of issues, as demonstrated by Faulkner and Nielsen [30]. Thus, we adopted this sample size. In this validation phase, 10 healthy subjects were recruited [16]: the study involved 7 males and 3 females without any history of low back pain or spine-related pathology (

Table 1. Mean, standard deviation (SD), and range of baseline features of healthy subjects.

Baseline Features
Age [y] (mean—SD; range)	17.36—SD 10.2; 5.4–33.1
Weight [kg] (mean—SD; range)	43.10—SD 19.3; 17.00–66.00
Height [cm] (mean—SD; range)	149.30—SD 27.15; 105.0–183.0
BMI (mean—SD; range)	18.11—SD 3.64; 13.55–25.39

). All subjects participated voluntarily after giving their written informed consent (for children, the consent form was signed by their parents). The Ethics Committee of the Institute approved the study (Study ID: 0053, date of approval 18 May 2023).

2.2. Data Collection

All measurements were obtained using an optoelectronic multicamera system for human motion analysis (SMART DX, BTS SpA., Milan, Italy) comprising eight high-resolution cameras with infrared light and a sampling frequency of 100 Hz, which recorded the position of passive retroreflective markers placed onto the skin. The experimental protocol required the placement of 30 markers (plastic spheres covered with reflective film, 10 mm in diameter). Twenty-two markers were placed according to the Davis protocol marker set [10], which is widely used in clinical practice for gait analysis, and the other eight reflective markers were placed on the back in correspondence with the following landmarks (Figure 1):

• Two onto the paravertebral points with reference to the left and right transverse processes of the third dorsal vertebra (r 3rd dv, l 3rd dv);
• One in correspondence to the spinous process of the sixth dorsal vertebra (6th dv);
• Two onto the paravertebral points with reference to the left and right transverse processes of the ninth dorsal vertebra (r 9th dv, l 9th dv);
• One in correspondence to the spinous process of the twelfth dorsal vertebra (12th dv);
• Two onto the paravertebral points with reference to the left and right transverse processes of the third lumbar vertebra (r 3rd lv, l 3rd lv).

The positions of these markers were chosen from the models proposed by Müller [27] and Arauz [28]; in fact, these two models allow a small number of markers to describe the three main segments of the vertebral column. In addition, their marker configuration is compatible with the Davis protocol, allowing us to obtain an integrated protocol. In relation to the pediatric application of the protocol, simplicity and quick subject preparation are essential. This aspect was also considered in choosing between the positioning of a large number of markers onto the entire spine or the selection of the most relevant spinal tracts to be assessed; indeed, this element of the protocol is not only important from the technical point of view, but also specifically analyzed in terms of usability and acceptance. In our experience and reasoning, the proposed marker set represents a good and balanced solution because it can be set up quickly and maintain a good clinical description of the parameters of interest.

Markers were placed by clinical operators (physiotherapists with training in the optoelectronic system for human motion analysis) with training and experience in recognizing the position of the spine and transversal processes, by means of manual identification. During the data acquisition procedure, the subject was asked to perform five walking trials barefoot along a 6 m distance at a self-selected normal-pace speed.

2.3. Data Processing

Raw data were processed with Smart Analyzer software 1.10.0470 (BTS Bioengineering, Milano, Italy). First, the 3D data were filtered and eventually interpolated in cases of short periods of missing data. The proposed protocol allows for the computation of spatial–temporal parameters (cycle duration, cadence, gait speed, stance phase, swing phase, double-support phase, stride length, and step width) and conventional kinematic parameters of the lower limbs, which are typically computed for Davis marker-set protocols [10]. In this study, we also computed the kinematic parameters of the spine and assessed their reliability. In particular, the following axes and angles were calculated:

• Sagittal vertical axis (SVA), defined as the horizontal distance between the S1 marker and a line dropped from the C7 marker, in the sagittal plane.
• Coronal vertical axis (CVA), defined as the horizontal distance between the S1 marker and a line dropped from the C7 marker, in the coronal plane.
• The angle of obliquity, tilt, and rotation of each considered segment, referring to the adjacent segment or to a global reference system, through a specific model based on the computation of Euler angles in the order of the sagittal plane (flexion or extension), frontal plane (lateral bending), and transversal plane (axial rotation), which is consistent with International Society of Biomechanics recommendations [31].

The considered segments were the shoulder–walk reference system, shoulder–pelvis, upper dorsal–lower dorsal, lower dorsal–lumbar, lumbar–pelvis, and pelvis–laboratory reference system (Figure 2). All trials were normalized with respect to time from 0 to 100 for the duration of the gait cycle. For each anatomical angle, we computed the range of motion (RoM).

2.4. Technical Validation

The technical validation of the protocol considered its repeatability and accuracy in a static setting. Accuracy tests were performed for the spinal tract using a static and geometric manikin by comparing the constructive and given angles of the manikin, simulating the human body as a reference, with the same angles measured with the optoelectronic system. The manikin is a physical mock-up of the spine, divided into three segments (upper dorsal spine, lower dorsal spine, and lumbar spine, as shown in Figure 3): it was built with fixed angles corresponding to the average healthy dorsal kyphosis and lumbar lordosis for young adults [32,33]. Fourteen retroreflective markers were placed in correspondence with the positions of the same anatomical landmarks used in the proposed protocol to model the spine, the shoulder, and the pelvis: the right and left acromion, spinous process of the 7th cervical vertebra, left and right transverse processes of the 3rd dorsal vertebrae, spinous process of the 6th dorsal vertebra, left and right transverse processes of the 9th dorsal vertebra, spinous process of the 12th dorsal vertebra, left and right transverse processes of the 3rd lumbar vertebra, spinous process of the 1st sacral vertebra, and right and left antero-superior iliac spine. Four static measurements were carried out for the manikin in different positions, the anatomical angles were calculated, and the errors were computed.

To verify the repeatability, we considered variability from two possible sources of errors or differences in the protocol [16]: the same observer in different sessions, or different observers. To analyze the variability in marking the subject, the following tests were performed:

-

Intra-operator remarking: repositioning of the markers on the subject by the same operator in different sessions;

-

Inter-operator remarking: repositioning of the markers on the subject by 2 different operators.

To recapitulate, the experimental protocol for the intra- and inter-operator validation consisted of the following steps:

• Preparation of the subject by the first operator, positioning the markers (30) on the selected body landmarks and taking anthropometric measurements.
• The acquisition of 5 walking trials; during each trial, the subject walks at his/her preferred speed for 6 m.
• Detachment of the markers from the subject.
• Preparation of the subject by the same operator, positioning the markers (30) on the selected body landmarks.
• The acquisition of 5 walking trials; during each trial, the subject walks at his/her preferred speed for 6 m.
• Detachment of the markers from the subject.

For the analysis of the inter-operator repeatability, steps 4 to 6 were carried out by a different therapist.

In total, each subject attended 2 sessions with each of 2 different operators, so we analyzed 20 walking trials for each subject.

The repositioning of markers by both the same operator and the second operator was carried out in consecutive sessions, with about a five-minute break between each one. The total test time for this repeatability validation was about 1 h and 30 min per session.

2.5. Usability Assessment

A usability evaluation is indispensable in the process of medical product development and is one of the key factors for successful implementation in the clinic [34]. The International Standardization Organization 9241–11 [35] international standard defines usability as the effectiveness, efficiency, and user-subjective satisfaction of a product when it is used for a specific purpose by a specific user in a specific use environment. Therefore, after each acquisition session, the participants were given the System Usability Scale (SUS) [36] and an assessment to evaluate how challenging the different phases of the test were using a 5-point Likert-scale methodology (ranging from 1 to 5). The considered operations were as follows:

• Phase 1: instrument preparation and marker preparation (for the test subjects, the part of instrument preparation was excluded from the questionnaire because the subjects were not involved in this operation);
• Phase 2: subject preparation with patient undressing, anthropometric measurements, and marker positioning;
• Phase 3: recording the movements with the optoelectronic system;
• Phase 4: removing the markers.

Each phase was timed to assess the duration required.

2.6. Statistical Analysis

For the accuracy analysis, the difference between the reference measures and the measures obtained with the optoelectronic system for each segment of the spine was assessed through the root mean square error (RMSE). In the intra-/inter-operator repeatability analysis, we calculated the interclass correlation coefficient (ICC) (two-way mixed effects model, absolute agreement, average measurements) of the ROMs of the considered spine tracts. We referred to standard interpretation criteria: ICC values between 0.5 and 0.75 indicate moderate reliability, values between 0.75 and 0.9 indicate good reliability, and values greater than 0.90 indicate excellent reliability [37]. In the usability analysis, for the survey, we considered the average value and the standard deviation for the rated usability/difficulty in the different operations and the average score for the overall SUS [38]. SPSS 21 software (IBM, Armonk, NY, USA) was used to perform the statistical analysis.

3. Results

The presented protocol aims to quantify gait spine 3D angles normalized to the gait cycle; with the lower limbs part of this protocol having already been extensively validated, we focused on the validation of spinal kinematics. Figure 3 presents an example of the data and of the clinical report that is automatically generated by the system.

3.1. Accuracy Analysis

Regarding the accuracy assessment for the kinematic measures of the spine,

Table 2. RMSE values for the comparison with reference measures (CVA: coronal vertical axis; SVA: sagittal vertical axis).

		RMSE
Shoulder–Ref System (°)	Obliquity	0.60
	Rotation	0.88
	Tilt	0.95
Shoulder–Pelvis (°)	Obliquity	0.70
	Rotation	0.48
	Tilt	1.58
Upper Dorsal–Lower Dorsal (°)	Obliquity	1.18
	Rotation	1.55
	Tilt	2.20
Lower Dorsal–Lumbar (°)	Obliquity	4.40
	Rotation	1.65
	Tilt	1.67
Lumbar–Pelvis (°)	Obliquity	0.40
	Rotation	0.09
	Tilt	1.40
Pelvis–Ref System (°)	Obliquity	0.10
	Rotation	0.40
	Tilt	0.73
CVA (mm)		0.87
SVA (mm)		3.08

illustrates the RMSE values computed for four different acquisitions of the constructive angles of the spine static model in relation to the angles measured with the optoelectronic system.

The total average error (±SD) is 1.2° ± 1.0° with a minimum value for lumbar–pelvis rotation (0.09°) and maximum error (4.4°) for the obliquity of the upper dorsal tract with respect to the lower dorsal segment.

3.2. Intra-Operator Repeatability Analysis

Table 3. ICC values for intra-operator reliability and their levels (+++ = excellent, ++ = good, + = moderate; CVA: coronal vertical axis; SVA: sagittal vertical axis) [37].

Metric		ICC
		Right Cycle	ICC Level	Left Cycle	ICC Level
Shoulder– Ref System	Obliquity	0.934	+++	0.924	+++
	Rotation	0.852	++	0.834	++
	Tilt	0.808	++	0.853	++
Shoulder– Pelvis	Obliquity	0.875	++	0.865	++
	Rotation	0.943	+++	0.957	+++
	Tilt	0.954	+++	0.862	++
Upper Dorsal– Lower Dorsal	Obliquity	0.923	+++	0.927	+++
	Rotation	0.761	++	0.839	++
	Tilt	0.739	+	0.839	++
Lower Dorsal– Lumbar	Obliquity	0.891	++	0.856	++
	Rotation	0.810	++	0.813	++
	Tilt	0.898	++	0.951	+++
Lumbar–Pelvis	Obliquity	0.851	++	0.840	++
	Rotation	0.736	+	0.719	+
	Tilt	0.861	++	0.933	+++
Pelvis– Ref System	Obliquity	0.893	++	0.872	++
	Rotation	0.962	+++	0.958	+++
	Tilt	0.867	++	0.839	++
Range CVA		0.926	+++	0.949	+++
Range SVA		0.914	+++	0.768	++

present the results obtained in the repeatability analysis for intra-operator remarking.

The data showed excellent or very good results in most cases. The average ICC was 0.888 (sd 0.032, range 0.840–0.934) for the obliquity angles; similarly, the average ICC was 0.867 (sd 0.061, range 0.739–0.954) for the angles on the sagittal plane. Regarding the rotation angles, the average ICC value was 0.849 (sd 0.088, range 0.719–0.962), slightly lower but still over the “good” threshold. Only the ICC for lumbar–pelvis rotation in both the right and left steps (ICC = 0.736 and 0.719, respectively) and the upper dorsal–lower dorsal tilt angle in the right cycle (ICC = 0.739) were moderate, though the values for the three cases were very close to the 0.75 threshold for a “good” assessment.

Motion capture data usually include outliers [39] arising from skin detachment, mislabeled markers, or intentional comfort-related sources such as postural adjustments [40]. It is often useful to consider the presence of outliers that could alter the final results [41]. Similarly to what happened in the accuracy validation with the static manikin, most of the low values were caused by 3D reconstruction errors for some markers. During the walking task, some back markers overlapped in the views of the lateral camera markers. This effect produced mislabeling that the filtering process could not completely smooth, which reduced the reconstruction accuracy of the 3D positions [42]. Furthermore, the precise placement of markers at anatomical points is not easy to achieve, especially in regions like the upper dorsal column, due to the presence of paravertebral muscles, particularly at the scapula, which limits the accurate identification of bony prominences [43]. In the case of pelvic–lumbar rotation, the slightly lower value is probably due to partial or total occlusion of the markers placed on the ASIS, due to the commuting of the arms during gait. This could worsen the reconstruction of the marker trajectories by decreasing the reliability [44].

For these reasons, in the analysis, we also proceeded with detecting the presence of and excluding possible outliers. By plotting the data of the intra-operator repeatability error for the upper dorsal–lower dorsal tilt angle (Figure 4), we identified an out-of-range value affecting the analysis; when this value was removed, the evaluation became good (ICC = 0.787 without the outlier).

With regard to the CVA, the intra-operator reliability ICC was excellent, while for the SVA, it was excellent for the right step, but only moderate for the left gait cycle (ICC = 0.73). In this case, we identified another outlier (Figure 5); when this was removed from the observations, the ICC was good (ICC = 0.858 without the outlier value).

3.3. Inter-Operator Repeatability Analysis

Table 4. ICC values for inter-operator reliability and their levels (+++ = excellent, ++ = good, + = moderate; CVA: coronal vertical axis; SVA: sagittal vertical axis) [37].

Metric		ICC
		Right Cycle	ICC Level	Left Cycle	ICC Level
Shoulder– Ref System	Obliquity	0.943	+++	0.938	+++
	Rotation	0.886	++	0.944	++
	Tilt	0.827	++	0.898	++
Shoulder– Pelvis	Obliquity	0.511	+	0.645	+
	Rotation	0.932	+++	0.943	+++
	Tilt	0.847	++	0.877	++
Upper Dorsal– Lower Dorsal	Obliquity	0.893	++	0.892	++
	Rotation	0.596	+	0.563	+
	Tilt	0.677	+	0.655	+
Lower Dorsal– Lumbar	Obliquity	0.839	++	0.851	++
	Rotation	0.692	++	0.745	+
	Tilt	0.912	++	0.911	+++
Lumbar–Pelvis	Obliquity	0.655	+	0.742	+
	Rotation	0.511	+	0.641	+
	Tilt	0.899	++	0.947	+++
Pelvis– Ref System	Obliquity	0.606	+	0.731	+
	Rotation	0.955	+++	0.954	+++
	Tilt	0.721	+	0.783	++
Range CVA		0.927	+++	0.905	+++
Range SVA		0.817	++	0.586	+

presents the ROMs of the different spinal tracts that were considered in the inter-operator repeatability analysis.

As expected, in general, the reliability with different operators was higher and the repeatability results were worse than in the intra-operator case, but they were still positive. In particular, the obtained data show an average ICC of 0.829 (sd 0.099, range 0.655–0.947) for the angles on the sagittal plane, an average ICC of 0.770 (sd 0.149, range 0.511–0.943) for the angles on the frontal plane, and an average ICC of 0.780 (sd 0.173, range 0.511–0.955) for the angles on the transversal plane.

The CVA and SVA inter-operator reliability analysis showed similar results to the intra-operator validation: again, the ICC was excellent (ICC > 0.9), and the SVA in the left gait cycle still presented outliers affecting the global outcome (Figure 6), given that the dataset the same for one of the operators. In this case, when the outlier was removed, the ICC was good (ICC = 0.751 without the outlier value).

3.4. Usability Assessment Results

To assess the usability of the proposed protocol,

Table 5. Assessment of the difficulty of application of the protocol (1: very difficult; 5 very easy). Phase 1 corresponds to instrument preparation, phase 2 is subject preparation, phase 3 is the recording of movements with the optoelectronic system, and phase 4 is the removal of the markers from the subject.

Operators					Patient
Phase	1	2	3	4	2	3	4
Mean	4.45	4.45	4.40	4.37	4.45	4.40	4.38
SD	0.93	0.60	0.78	0.67	0.58	0.78	0.67
MAX	5.00	5.00	5.00	5.00	5.00	5.00	5.00
min	4.00	3.00	3.00	3.00	3.00	2.00	3.00

presents data on the difficulty perceived by both the patient and the operator during the different steps of acquisition.

The SUS questionnaires were collected at the end of each acquisition, and the average score was calculated: mean 74.75, SD 5.88, MAX 85, and min 60. The SUS score obtained herein corresponds to good usability. Finally, in

Table 6. Time (in seconds) required for the entire acquisition process and for each phase.

Time to Complete the Protocol (Also Divided into Four Phases)
Phase No.	1	2	3	4	TOTAL
Mean [min]	1.74	13.05	5.35	1.32	21.46
SD	0.54	1.35	0.91	0.19	2.06
MAX [min]	3.38	16.40	8.83	2.02	28.55
Min [min]	1.13	11.18	3.65	1.10	18.93

, the durations of each step and the entire acquisition process are reported. The time required for each phase and globally was appropriate and acceptable, including for children.

4. Discussion

The method presented herein is a non-invasive and integrated approach to evaluate spinal mobility during gait. Its application requires a marker-based system for human motion analysis, and its relative simplicity (studied in close connection with clinicians) makes it suitable for daily use in clinical practice. The proper balance between the need for a detailed motion description of the spine (requiring a high number of markers) and limited complexity (i.e., minimizing the number of markers to be prepared and placed onto the subject) produced good usability and acceptance in both operators and subjects. The time needed for the experimental examination is less than 30 min in total. The average durations of the single phases are more than acceptable: as expected, subject preparation (13 min) and data recording (5+ min) are the most time-consuming steps, but they are necessary operations. To the contrary, setup preparation (<2 min) and the removal of the markers from the subject (<1.5 min) are very short procedures (Table 6). Moreover, the overall difficulty of application for this protocol was perceived as easy by both the operator and the patient. The usability of this approach is good, with an average SUS score of 74.75 [45] (Table 5).

The validation of the protocol was satisfactory concerning the accuracy of kinematic measures of the spine and repeatability tests. The accuracy of this method was assessed in a static setting using a manikin that reproduces the three evaluated segments of the spine. This evaluation demonstrated excellent results: the total average error was 1.2°, with a minimum value for lumbar–pelvis rotation (0.09°) and maximum error (4.4°) for the obliquity of the upper dorsal tract with respect to the lower dorsal segment (Table 2). The main source of this error is the very close field of view of the lateral cameras in the standard laboratory setup for these markers. In fact, markers that are very close to each other could overlap in the field of vision; this effect produces tracking errors that may decrease the reconstruction accuracy of their 3D positions [42]. However, the global error mode was 0.4°, demonstrating that the error outliers were very limited.

In the intra-operator repeatability analysis, the flexion and obliquity angles generally showed good reliability; when outlier values were excluded, the mean ICC increased in the sagittal plane to 0.867 and in the frontal plane to 0.888. The rotation angles also showed good reliability: the mean ICC was 0.849 in the transversal plane (Table 3). The inferior value and high variability of this measure are probably due to muscle contractions that change the morphology of the back during movement, especially for lateral markers that are close to the paravertebral muscles. These values are in line with those found in previous studies on the intra- and inter-operator reliability of gait analysis [26,46,47]. The CVA and SVA measurements, even considering their outliers, also showed good reliability (for CVA, mean ICC = 0.937; for SVA, mean ICC = 0.841).

In the inter-operator repeatability analysis, the results were generally good for all the angles, especially after the outliers were removed: the mean ICC was 0.829 (i.e., good reliability) in the sagittal plane, while the data showed moderate reliability in the frontal and transversal planes (respectively, mean ICC = 0.770 and mean ICC = 0.780) (Table 4). The most critical tracts were the upper dorsal–lower dorsal and lumbar–pelvic sectors, confirming the results of previous studies [29,46]. This is explained by the paravertebral muscles underneath the markers: during movement, their contractions modify the morphology of the back, thus adding an intrinsic 3D displacement to the corresponding markers and producing unavoidable, unremovable artefacts. Regarding the CVA and SVA measures, the analyses of inter-operator reliability showed at least moderate reliability (CVA mean ICC = 0.916 and SVA mean ICC = 0.702). In general, the results confirmed that the reproducibility of each joint motion was good for all anatomical planes (sagittal, frontal, and transverse planes) and for testing–retesting by the same or different operators. These findings may be improved by enhancing some methodological factors of this study, such as the number of subjects participating in the analysis or the familiarization of the operator with marker placement. Intrinsic errors like skin motion artefacts were confirmed to be irrelevant, in accordance with other studies [48].

There are some limitations related to the present study, including the small sample size and the heterogeneity of the recruited group with respect to age, weight, and height. The last factor is a limit but also a positive strength of the protocol, as the obtained results were stable in all subjects. For this reason, the preliminary normalcy dataset utilized herein could be used for comparison with a pathological population. Nevertheless, future studies on a larger population are recommended in order to verify the inter-subject variability in relation to different features (age, weight, height, etc.) of the healthy population.

5. Conclusions

This study introduced a new protocol for the integrated assessment of gait and spinal movements during gait, in order to produce a global analysis of spinal and multi-segmental body kinematics. Information related to the spinal movement patterns during gait is a fundamental outcome of this methodology; thus, it could be very useful for integrated functional assessments and clinical diagnoses in several pathologies.

This protocol was analyzed and validated in terms of its accuracy, repeatability, and usability in a population of 10 healthy subjects. The reliability assessment results were good and satisfactory: the results confirmed that the proposed protocol is accurate and repeatable. Additionally, the usability evaluations by operators and patients were more than good. These positive results regarding intra-/inter-operative repeatability encourage the introduction of this protocol in clinical practice to support diagnosis or the evaluation of rehabilitation treatments for neurological and orthopedic disorders impacting the spinal morphology and/or movements and gait.

This first dataset on healthy young subjects can also be considered a preliminary database of normalcy values. Additional experimental activities will be needed to populate the database with normal reference values and to create datasets related to specific pathologies such as idiopathic scoliosis in children. The latter case will be the first one to be investigated with this protocol. This application would also support an evaluation of the clinical impact of this protocol in neurorehabilitation.