Analysis of knee vibration signals using linear prediction

Time-Frequency Analysis and Parameterisation of Knee Sounds for Non-Invasive Detection of Osteoarthritis

OBJECTIVE

In this work the potential of non-invasive detection of knee osteoarthritis is investigated using the sounds generated by the knee joint during walking.

METHODS

The information contained in the time-frequency domain of these signals and its compressed representations is exploited and their discriminant properties are studied. Their efficacy for the task of normal vs abnormal signal classification is evaluated using a comprehensive experimental framework. Based on this, the impact of the feature extraction parameters on the classification performance is investigated using Classification and Regression Trees, Linear Discriminant Analysis and Support Vector Machine classifiers.

RESULTS

It is shown that classification is successful with an area under the Receiver Operating Characteristic curve of 0.92.

CONCLUSION

The analysis indicates improvements in classification performance when using non-uniform frequency scaling and identifies specific frequency bands that contain discriminative features.

SIGNIFICANCE

Contrary to other studies that focus on sit-to-stand movements and knee flexion/extension, this study used knee sounds obtained during walking. The analysis of such signals leads to non-invasive detection of knee osteoarthritis with high accuracy and could potentially extend the range of available tools for the assessment of the disease as a more practical and cost effective method without requiring clinical setups.

Vibroarthrography, arthrophonography — methods for non-invasive detection of the knee cartilage damage

Phonoarthrography, vibration arthrography are non-invasive methods for assessing the condition of cartilage and the knee joint as a whole based on the sounds made by the joint movement. Acoustic sensors (accelerometers, microphones) are attached to the knee to measure the knee joint noise both in control groups (young adults and elderly subjects) and in patients with knee osteoarthropathies. Different authors propose different methods for attaching sensors, documenting and analyzing the joint sounds. The identified specific features allowed distinguishing between asymptomatic knee joints and those with osteoarthropathies. Acoustic signals were recorded and processed, and their frequency characteristics were determined and classified. The classification effectiveness correlated with the existing diagnostic tests and hence phonoarthrography and vibration arthrography can be qualified as a useful diagnostic aid.

A Glove-Based Form Factor for Collecting Joint Acoustic Emissions: Design and Validation

Sounds produced by the articulation of joints have been shown to contain information characteristic of underlying joint health, morphology, and loading. In this work, we explore the use of a novel form factor for non-invasively acquiring acoustic/vibrational signals from the knee joint: an instrumented glove with a fingertip-mounted accelerometer. We validated the glove-based approach by comparing it to conventional mounting techniques (tape and foam microphone pads) in an experimental framework previously shown to reliably alter healthy knee joint sounds (vertical leg press). Measurements from healthy subjects (N = 11) in this proof-of-concept study demonstrated a highly consistent, monotonic, and significant (p < 0.01) increase in low-frequency signal root-mean-squared (RMS) amplitude-a straightforward metric relating to joint grinding loudness-with increasing vertical load across all three techniques. This finding suggests that a glove-based approach is a suitable alternative for collecting joint sounds that eliminates the need for consumables like tape and the interface noise associated with them.

Wearable knee health system employing novel physiological biomarkers

Knee injuries and chronic disorders, such as arthritis, affect millions of Americans, leading to missed workdays and reduced quality of life. Currently, after an initial diagnosis, there are few quantitative technologies available to provide sensitive subclinical feedback to patients regarding improvements or setbacks to their knee health status; instead, most assessments are qualitative, relying on patient-reported symptoms, performance during functional tests, and physical examinations. Recent advances have been made with wearable technologies for assessing the health status of the knee (and potentially other joints) with the goal of facilitating personalized rehabilitation of injuries and care for chronic conditions. This review describes our progress in developing wearable sensing technologies that enable quantitative physiological measurements and interpretation of knee health status. Our sensing system enables longitudinal quantitative measurements of knee sounds, swelling, and activity context during clinical and field situations. Importantly, we leverage machine-learning algorithms to fuse the low-level signal and feature data of the measured time series waveforms into higher level metrics of joint health. This paper summarizes the engineering validation, baseline physiological experiments, and human subject studies-both cross-sectional and longitudinal-that demonstrate the efficacy of using such systems for robust knee joint health assessment. We envision our sensor system complementing and advancing present-day practices to reduce joint reinjury risk, to optimize rehabilitation recovery time for a quicker return to activity, and to reduce health care costs.

Synthesis of vibroarthrographic signals in knee osteoarthritis diagnosis training

Background

Vibroarthrographic (VAG) signals are used as useful indicators of knee osteoarthritis (OA) status. The objective was to build a template database of knee crepitus sounds. Internships can practice in the template database to shorten the time of training for diagnosis of OA.

Methods

A knee sound signal was obtained using an innovative stethoscope device with a goniometer. Each knee sound signal was recorded with a Kellgren–Lawrence (KL) grade. The sound signal was segmented according to the goniometer data. The signal was Fourier transformed on the correlated frequency segment. An inverse Fourier transform was performed to obtain the time-domain signal. Haar wavelet transform was then done. The median and mean of the wavelet coefficients were chosen to inverse transform the synthesized signal in each KL category. The quality of the synthesized signal was assessed by a clinician.

Results

The sample signals were evaluated using different algorithms (median and mean). The accuracy rate of the median coefficient algorithm (93 %) was better than the mean coefficient algorithm (88 %) for cross-validation by a clinician using synthesis of VAG.

Conclusions

The artificial signal we synthesized has the potential to build a learning system for medical students, internships and para-medical personnel for the diagnosis of OA. Therefore, our method provides a feasible way to evaluate crepitus sounds that may assist in the diagnosis of knee OA.

Quantifying the Consistency of Wearable Knee Acoustical Emission Measurements During Complex Motions

Knee-joint sounds could potentially be used to noninvasively probe the physical and/or physiological changes in the knee associated with rehabilitation following acute injury. In this paper, a system and methods for investigating the consistency of knee-joint sounds during complex motions in silent and loud background settings are presented. The wearable hardware component of the system consists of a microelectromechanical systems microphone and inertial rate sensors interfaced with a field programmable gate array-based real-time processor to capture knee-joint sound and angle information during three types of motion: flexion-extension (FE), sit-to-stand (SS), and walking (W) tasks. The data were post-processed to extract high-frequency and short-duration joint sounds (clicks) with particular waveform signatures. Such clicks were extracted in the presence of three different sources of interference: background, stepping, and rubbing noise. A histogram-vector Vn(→) was generated from the clicks in a motion-cycle n, where the bin range was 10°. The Euclidean distance between a vector and the arithmetic mean Vav(→) of all vectors in a recording normalized by the Vav(→) is used as a consistency metric dn. Measurements from eight healthy subjects performing FE, SS, and W show that the mean (of mean) consistency metric for all subjects during SS (μ [ μ (dn)] = 0.72 in silent, 0.85 in loud) is smaller compared with the FE (μ [ μ (dn)] = 1.02 in silent, 0.95 in loud) and W ( μ [ μ (dn)] = 0.94 in silent, 0.97 in loud) exercises, thereby implying more consistent click-generation during SS compared with the FE and W. Knee-joint sounds from one subject performing FE during five consecutive work-days (μ [ μ (dn) = 0.72) and five different times of a day (μ [ μ (dn) = 0.73) suggests high consistency of the clicks on different days and throughout a day. This work represents the first time, to the best of our knowledge, that joint sound consistency has been quantified in ambulatory subjects performing every-day activities (e.g., SS, walking). Moreover, it is demonstrated that noise inherent with joint-sound recordings during complex motions in uncontrolled settings does not prevent joint-sound-features from being detected successfully.

Knee joint vibration signal analysis with matching pursuit decomposition and dynamic weighted classifier fusion

Analysis of knee joint vibration (VAG) signals can provide quantitative indices for detection of knee joint pathology at an early stage. In addition to the statistical features developed in the related previous studies, we extracted two separable features, that is, the number of atoms derived from the wavelet matching pursuit decomposition and the number of significant signal turns detected with the fixed threshold in the time domain. To perform a better classification over the data set of 89 VAG signals, we applied a novel classifier fusion system based on the dynamic weighted fusion (DWF) method to ameliorate the classification performance. For comparison, a single leastsquares support vector machine (LS-SVM) and the Bagging ensemble were used for the classification task as well. The results in terms of overall accuracy in percentage and area under the receiver operating characteristic curve obtained with the DWF-based classifier fusion method reached 88.76% and 0.9515, respectively, which demonstrated the effectiveness and superiority of the DWF method with two distinct features for the VAG signal analysis.

An acoustical evaluation of knee sound for non-invasive screening and early detection of articular pathology

Knee sound signals generated by knee movement are sometimes associated with degeneration of the knee joint surface and such sounds may be a useful index for early disease. In this study, we detected the acoustical parameters, such as the fundamental frequency (F0), mean amplitude of the pitches, and jitter and shimmer of knee sounds, and compared them according to the pathological conditions. Six normal subjects (4 males and 2 females, age: 28.3 ± 2.3 years) and 11 patients with knee problems were enrolled. The patients were divided into the 1st patient group (5 males and 1 female, age: 30.2 ± 10.3 years) with ruptured wounds of the meniscus and 2nd patient group (2 males and 3 females, age: 42.1 ± 16.2 years) with osteoarthritis. The mean values of F0, jitter and shimmer of the 2nd patient group were larger than those of the normal group, whereas those of the 1st patient group were smaller (p < 0.05). Also, the F0 and jitter in the standing position were larger than those in the sitting position in both the 1st and 2nd patient groups (p < 0.05). These results showed good potential for the non-invasive diagnosis and early detection of articular pathologies via an auscultation.

An enhanced algorithm for knee joint sound classification using feature extraction based on time-frequency analysis

Vibroarthrographic (VAG) signals, generated by human knee movement, are non-stationary and multi-component in nature and their time-frequency distribution (TFD) provides a powerful means to analyze such signals. The objective of this paper is to improve the classification accuracy of the features, obtained from the TFD of normal and abnormal VAG signals, using segmentation by the dynamic time warping (DTW) and denoising algorithm by the singular value decomposition (SVD). VAG and knee angle signals, recorded simultaneously during one flexion and one extension of the knee, were segmented and normalized at 0.5 Hz by the DTW method. Also, the noise within the TFD of the segmented VAG signals was reduced by the SVD algorithm, and a back-propagation neural network (BPNN) was used to classify the normal and abnormal VAG signals. The characteristic parameters of VAG signals consist of the energy, energy spread, frequency and frequency spread parameter extracted by the TFD. A total of 1408 segments (normal 1031, abnormal 377) were used for training and evaluating the BPNN. As a result, the average classification accuracy was 91.4 (standard deviation +/-1.7) %. The proposed method showed good potential for the non-invasive diagnosis and monitoring of joint disorders such as osteoarthritis.

Screening of knee-joint vibroarthrographic signals using statistical parameters and radial basis functions

Externally detected vibroarthrographic (VAG) signals bear diagnostic information related to the roughness, softening, breakdown, or the state of lubrication of the articular cartilage surfaces of the knee joint. Analysis of VAG signals could provide quantitative indices for noninvasive diagnosis of articular cartilage breakdown and staging of osteoarthritis. We propose the use of statistical parameters of VAG signals, including the form factor involving the variance of the signal and its derivatives, skewness, kurtosis, and entropy, to classify VAG signals as normal or abnormal. With a database of 89 VAG signals, screening efficiency of up to 0.82 was achieved, in terms of the area under the receiver operating characteristics curve, using a neural network classifier based on radial basis functions.

Using system identification to model the transmission of vibroarthrographic signals

Vibroarthrographic signals have been proposed as a noninvasive tool for the diagnosis of joint injury. Models of VAG generation and transmission are required before application of this technique can begin. An experiment has been designed and performed to estimate sound transmission in the human knee at set joint angles. Linear frequency domain models and linear and nonlinear time domain models were estimated from the experimental data. Linear models with high accuracy were identified for knees at an angle of 90/ degrees . Models identified from angles below 90 degrees had relatively low accuracy.

Influence of estimators of spectral density on the analysis of electromyographic and vibromyographic signals

Electromyographic (EMG) and vibromyographic (VMG) signals are related to electrical and mechanical muscle activity, respectively. It is known that variations in their frequency components are related to changes in muscle activity and fatigue. The aims of this study were: (1) to analyse the resolution, variance and bias of different estimations of power spectral density function (PSD); and (2) to evaluate the influence of the spectral estimation method on three indices calculated from the PSD of EMG and VMG signals: mean (f(m)) and median (f(c)) frequencies and the ratio of high and low frequency components (H/L ratio) to select the most suitable estimator. Myographic signals were recorded from the sternomastoid muscle, an accessory respiratory muscle, during breathing. For non-parametric methods, Welch periodograms and correlograms were analysed with different windows. Autoregressive (AR) moving average (MA) and ARMA models with different orders were evaluated in the parametric methods. The reproducibility of the results was also studied. Frequency indices, particularly the H/L ratio and f(c), changed considerably when varying the following parameters of the estimators: periodogram with segment durations longer than 150 ms in EMG and with any duration in VMG signals; correlogram with window length shorter than 10% of the total number of samples; and AR models with an order lower than 10, 20 and 40 in f(c), fm and H/L ratio, respectively, in both myographic signals.

Auditory display of knee-joint vibration signals

Sounds generated due to rubbing of knee-joint surfaces may lead to a potential tool for noninvasive assessment of articular cartilage degeneration. In the work reported in the present paper, an attempt is made to perform computer-assisted auscultation of knee joints by auditory display (AD) of vibration signals (also known as vibroarthrographic or VAG signals) emitted during active movement of the leg. Two types of AD methods are considered: audification and sonification. In audification, the VAG signals are scaled in time and frequency using a time-frequency distribution to facilitate aural analysis. In sonification, the instantaneous mean frequency and envelope of the VAG signals are derived and used to synthesize sounds that are expected to facilitate more accurate diagnosis than the original signals by improving their aural quality. Auditory classification experiments were performed by two orthopedic surgeons with 37 VAG signals including 19 normal and 18 abnormal cases. Sensitivity values (correct detection of abnormality) of 31%, 44%, and 83%, and overall classification accuracies of 53%, 40%, and 57% were obtained with the direct playback, audification, and sonification methods, respectively. The corresponding d' scores were estimated to be 1.10. -0.36, and 0.55. The high sensitivity of the sonification method indicates that the technique could lead to improved detection of knee-joint abnormalities; however, additional work is required to improve its specificity and achieve better overall performance.

Vibration arthrometry in patients with knee joint disorders

Physiological patellofemoral crepitus (PPC) is the vibration signal produced by the knee joint during slow motion (less than 5 degrees per second), which can be measured by vibration arthrometry (VAM). By using the autoregressive (AR) model for the PPC signals of patients with knee osteoarthritis, the study analyzes the PPC signals to evaluate the condition of patellar-femoral joint cartilage. Accordingly, we can divide osteoarthritis into three types, type 1: the cartilage of patellar-femoral joint is intact, the osteoarthritis found in the femoral-tibial joint surface; type 2: degeneration occurs in the surface cartilage of both the femoral-tibial joint and the femoral trochlea, but not on the patellar surface; type 3: both patellar-femoral and femoral-tibial joints have osteoarthritis. For the analysis, the intraclass distance of AR coefficients and spectral power ratio of dominant poles are adopted. Based on the proposed method, two cases of type 1, six of type 2, and 28 of type 3 were found in 36 cases of knee osteoarthritis. This is in agreement with the operative findings. For comparison, the PPC signals of 10 subjects with normal knees (without pain or wound history) were also measured. The results of analysis of the 10 normal subjects were consistent and clearly differentiable from those of the osteoarthritis patients. Therefore, the proposed method is efficient for the analysis of the condition of patellar-femoral joint cartilage and VAM may become an alternative way of noninvasive diagnosis of knee osteoarthritis.

Adaptive time-frequency analysis of knee joint vibroarthrographic signals for noninvasive screening of articular cartilage pathology

Vibroarthrographic (VAG) signals emitted by human knee joints are nonstationary and multicomponent in nature; time-frequency distributions (TFD's) provide powerful means to analyze such signals. The objective of this paper is to construct adaptive TFD's of VAG signals suitable for feature extraction. An adaptive TFD was constructed by minimum cross-entropy optimization of the TFD obtained by the matching pursuit decomposition algorithm. Parameters of VAG signals such as energy, energy spread, frequency, and frequency spread were extracted from their adaptive TFD's. The parameters carry information about the combined TF dynamics of the signals. The mean and standard deviation of the parameters were computed, and each VAG signal was represented by a set of just six features. Statistical pattern classification experiments based on logistic regression analysis of the parameters showed an overall normal/abnormal screening accuracy of 68.9% with 90 VAG signals (51 normals and 39 abnormals), and a higher accuracy of 77.5% with a database of 71 signals with 51 normals and 20 abnormals of a specific type of patellofemoral disorder. The proposed method of VAG signal analysis is independent of joint angle and clinical information, and shows good potential for noninvasive diagnosis and monitoring of patellofemoral disorders such as chondromalacia patella.

Vibration arthrometry in the patients with failed total knee replacement

This is a preliminary research on the vibration arthrometry of artificial knee joint in vivo. Analyzing the vibration signals measured from the accelerometer on patella, there are two speed protocols in knee kinematics: 1) 2 degrees/s, the signal is called "physiological patellofemoral crepitus (PPC)", and 2) 67 degrees/s, the signal is called "vibration signal in rapid knee motion". The study has collected 14 patients who had revision total knee arthroplasty due to prosthetic wear or malalignment represent the failed total knee replacement (FTKR), and 12 patients who had just undergone the primary total knee arthroplasty in the past two to six months and have currently no knee pain represent the normal total knee replacement (NTKR). FTKR is clinically divided into three categories: metal wear, polyethylene wear of the patellar component, and no wear but with prosthesis malalignment. In PPC, the value of root mean square (rms) is used as a parameter; in vibration signals in rapid knee motion, autoregressive modeling is used for adaptive segmentation and extracting the dominant pole of each signal segment to calculate the spectral power ratios in f < 100 Hz and f > 500 Hz. It was found that in the case of metal wear, the rms value of PPC signal is far greater than a knee joint with polyethylene wear and without wear, i.e., PPC signal appears only in metal wear. As for vibration signals in rapid knee motion, prominent time-domain vibration signals could be found in the FTKR patients with either polyethylene or metal wear of the patellar component. We also found that for normal knee joint, the spectral power ratio of dominant poles has nearly 80% distribution in f < 100 Hz, is between 50% and 70% for knee with polyethylene wear and below 30% for metal wear, whereas in f > 500 Hz, spectral power ratio of dominant poles has over 30% distribution in metal wear but only nonsignificant distribution in polyethylene wear, no wear, and normal knee. The results show that vibration signals in rapid knee motion can be used for effectively detecting polyethylene wear of the patellar component in the early stage, while PPC signals can only be used to detect prosthetic metal wear in the late stage.

Parametric representation and screening of knee joint vibroarthrographic signals

We have been investigating analysis of knee joint vibration or vibroarthrographic (VAG) signals as a potential tool for noninvasive diagnosis and monitoring of cartilage pathology. In this paper, we present a comprehensive comparative study of different parametric representations of VAG signals. Dominant poles and cepstral coefficients were derived from autoregressive models of adaptively segmented VAG signals. Signal features and a few clinical features were used as feature vectors in pattern classification experiments based on logistic regression analysis and the leave-one-out method. The results using 51 normal and 39 abnormal signals indicated the superior performance of cepstral coefficients in VAG signal classification with an accuracy rate of 75.6%. With 51 normal and 20 abnormal signals limited to chondromalacia patella, cepstral coefficients again gave the highest accuracy rate of 85.9%.

Adaptive filtering, modelling and classification of knee joint vibroarthrographic signals for non-invasive diagnosis of articular cartilage pathology

Interpretation of vibrations or sound signals emitted from the patellofemoral joint during movement of the knee, also known as vibroarthrography (VAG), could lead to a safe, objective, and non-invasive clinical tool for early detection, localisation, and quantification of articular cartilage disorders. In this study with a reasonably large database of VAG signals of 90 human knee joints (51 normal and 39 abnormal), a new technique for adaptive segmentation based on the recursive least squares lattice (RLSL) algorithm was developed to segment the non-stationary VAG signals into locally-stationary components; the stationary components were then modelled autoregressively, using the Burg-Lattice method. Logistic classification of the primary VAG signals into normal and abnormal signals (with no restriction on the type of cartilage pathology) using only the AR coefficients as discriminant features provided an accuracy of 68.9% with the leave-one-out method. When the abnormal signals were restricted to chondromalacia patella only, the classification accuracy rate increased to 84.5%. The effects of muscle contraction interference (MCI) on VAG signals were analysed using signals from 53 subjects (32 normal and 21 abnormal), and it was found that adaptive filtering of the MCI from the primary VAG signals did not improve the classification accuracy rate. The results indicate that VAG is a potential diagnostic tool for screening for chondromalacia patella.

Screening of vibroarthrographic signals via adaptive segmentation and linear prediation modeling

This paper proposes a noninvasive method to diagnose chondromalacia patella at its early stages by recording knee vibration signals (also known as vibroarthrographic or VAG signals) over the mid-patella during normal movement. An adaptive segmentation method was developed to segment the nonstationary VAG signals. The least squares modeling method was used to reduce the number of data samples to a few model parameters. Model parameters along with a few clinical parameters and a signal variability parameter were then used as discriminant features for screening VAG signals by applying logistic and discriminant algorithms. The system was trained using ten normal and eight abnormal signals. It correctly screened a separate test set of ten normal and eight abnormal signals except for one normal signal. The proposed method should find use as an alternative technique for diagnosis of knee joint pathology or as a test before arthroscopy or major knee surgery.

Localization of knee joint cartilage pathology by multichannel vibroarthrography

This paper proposes non-invasive techniques to localize sound or vibroarthrographic (VAG) signal sources in human knee joints. VAG signals from normal subjects, patients who subsequently underwent arthroscopy, and cadavers with arthroscopically-created lesions, obtained by stimulation with a finger tap over the mid-patella and swinging movement of the leg, were analyzed for time delays using cross-correlation functions for source localization. Correct results were obtained for 13 of the 14 subjects tested by finger stimulation, and for 11 of the 12 subjects whose VAG signals during swinging movement were analyzed. The techniques could be valuable in the diagnosis and treatment of knee pathology before and after joint surgery or drug therapy.

Adaptive cancellation of muscle contraction interference in vibroarthrographic signals

Vibroarthrography (VAG) is an innovative, objective, non-invasive technique for obtaining diagnostic information concerning the articular cartilage of a joint. Knee VAG signals can be detected using a contact sensor over the skin surface of the knee joint during knee movement such as flexion and/or extension. These measured signals, however, contain significant interference caused by muscle contraction that is required for knee movement. Quality improvement of VAG signals is an important subject, and crucial in computer-aided diagnosis of cartilage pathology. While simple frequency domain high-pass (or band-pass) filtering could be used for minimizing muscle contraction interference (MCI), it could eliminate possible overlapping spectral components of the VAG signals. In this work, an adaptive MCI cancellation technique is presented as an alternative technique for filtering VAG signals. Methods of measuring the VAG and reference signals (MCI) are described, with details on MCI identification, characterization, and step size optimization for the adaptive filter. The performance of the method is evaluated by simulated signals as well as signals obtained from human subjects under isotonic contraction.