GaitKeeper: A System for Measuring Canine Gait

IMU Airtime Detection in Snowboard Halfpipe: U-Net Deep Learning Approach Outperforms Traditional Threshold Algorithms

Airtime is crucial for high-rotation tricks in snowboard halfpipe performance, significantly impacting trick difficulty, the primary judging criterion. This study aims to enhance the detection of take-off and landing events using inertial measurement unit (IMU) data in conjunction with machine learning algorithms since manual video-based methods are too time-consuming. Eight elite German National Team snowboarders performed 626 halfpipe tricks, recorded by two IMUs at the lateral lower legs and a video camera. The IMU data, synchronized with video, were labeled manually and segmented for analysis. Utilizing a 1D U-Net convolutional neural network (CNN), we achieved superior performance in all of our experiments, establishing new benchmarks for this binary segmentation task. In our extensive experiments, we achieved an 80.34% lower mean Hausdorff distance for unseen runs compared with the threshold approach when placed solely on the left lower leg. Using both left and right IMUs further improved performance (83.37% lower mean Hausdorff). For data from an algorithm-unknown athlete (Zero-Shot segmentation), the U-Net outperformed the threshold algorithm by 67.58%, and fine-tuning on athlete-specific (Few-Shot segmentation) runs improved the lower mean Hausdorff to 78.68%. The fine-tuned model detected takeoffs with median deviations of 0.008 s (IQR 0.030 s), landing deviations of 0.005 s (IQR 0.020 s), and airtime deviations of 0.000 s (IQR 0.027 s). These advancements facilitate real-time feedback and detailed biomechanical analysis, enhancing performance and trick execution, particularly during critical events, such as take-off and landing, where precise time-domain localization is crucial for providing accurate feedback to coaches and athletes.

A standardised approach to quantifying activity in domestic dogs

Objective assessment of activity via accelerometry can provide valuable insights into dog health and welfare. Common activity metrics involve using acceleration cut-points to group data into intensity categories and reporting the time spent in each category. Lack of consistency and transparency in cut-point derivation makes it difficult to compare findings between studies. We present an alternative metric for use in dogs: the acceleration threshold (as a fraction of standard gravity, 1 g = 9.81 m/s2) above which the animal's X most active minutes are accumulated (MXACC) over a 24-hour period. We report M2ACC, M30ACC and M60ACC data from a colony of healthy beagles (n = 6) aged 3-13 months. To ensure that reference values are applicable across a wider dog population, we incorporated labelled data from beagles and volunteer pet dogs (n = 16) of a variety of ages and breeds. The dogs' normal activity patterns were recorded at 200 Hz for 24 hours using collar-based Axivity-AX3 accelerometers. We calculated acceleration vector magnitude and MXACC metrics. Using labelled data from both beagles and pet dogs, we characterize the range of acceleration outputs exhibited enabling meaningful interpretation of MXACC. These metrics will help standardize measurement of canine activity and serve as outcome measures for veterinary and translational research.

Objective comparison of a sit to stand test to the walk test for the identification of unilateral lameness caused by cranial cruciate ligament disease in dogs

OBJECTIVE

The purpose of this study was to evaluate a sit to stand test with the walk test for the identification of unilateral cranial cruciate ligament rupture in dogs.

MATERIALS AND METHODS

Peak vertical force and vertical impulse were measured on a pressure-sensitive walkway, during a sit to stand test and walk test, and in 10 dogs with unilateral cranial cruciate ligament rupture and 18 non-lame dogs. Data collected were used to calculate symmetry indices (SI) of ipsilateral and contralateral hindlimbs (HL), diagonal limb pairs (DLP) and ipsilateral limb pairs (ILP).

RESULTS

The symmetry indices of peak vertical force of HL during the walk test and sit to stand test were 100% and 90% sensitive for discriminating lame and non-lame dogs respectively. The symmetry indices of vertical impulse of HLs during the walk test and sit to stand test were 100% and 50% sensitive for discriminating lame and non-lame dogs respectively. Analysis of ipsilateral and diagonal limb pairs did not improve the discrimination in either test. The time taken to collect data from the sit to stand test data was shorter than for the walk test.

CLINICAL SIGNIFICANCE

Whilst the sit to stand test required a shorter time for collection of data than the walk test, it did not accurately identify all dogs with lameness associated with CCLR, and thus has relatively limited clinical utility in its tested form.

Extraction of canine gait characteristics using a mobile gait analysis system based on inertial measurement units

This study aims to investigate two simple algorithms for extracting gait features from an inertial measurement unit (IMU) based canine gait analysis system. The first algorithm was developed to determine the hip/shoulder extension/flexion range of motion. The second algorithm automatically determines the stance and swing phase per leg. To investigate the accuracy of the algorithms, two dogs were walked on a treadmill and measured simultaneously with an IMU system, an optical tracking system and two cameras. The range of motion estimation was compared to the optical tracking systems, with a total of 280 steps recorded. To test the stance and swing phase detection, a total of 63 steps were manually annotated in the video recordings and compared with the output of the algorithm. The IMU's-based estimation of the range of motion showed an average deviation of 1.4° to 5.6° from the optical reference, while the average deviation in the detection of the beginning and end of the stance and swing phases ranged from -0.01 to 0.09 s. This study shows that even simple algorithms can extract relevant information from inertial measurements that are comparable to results from more complex approaches. However, additional studies including a wider subject pool need to be conducted to investigate the significance of the presented findings.

Four-limb wireless IMU sensor system for automatic gait detection in canines

This study aims to develop a 4-limb canine gait analysis system using wireless inertial measurement units (IMUs). 3D printed sensor holders were designed to ensure quick and consistent sensor mounting. Signal analysis algorithms were developed to automatically determine the timing of swing start and end in a stride. To evaluate the accuracy of the new system, a synchronized study was conducted in which stride parameters in four dogs were measured simultaneously using the 4-limb IMU system and a pressure-sensor based walkway gait system. The results showed that stride parameters measured in both systems were highly correlated. Bland-Altman analyses revealed a nominal mean measurement bias between the two systems in both forelimbs and hindlimbs. Overall, the disagreement between the two systems was less than 10% of the mean value in over 92% of the data points acquired from forelimbs. The same performance was observed in hindlimbs except for one parameter due to small mean values. We demonstrated that this 4-limb system could successfully visualize the overall gait types and identify rapid gait changes in dogs. This method provides an effective, low-cost tool for gait studies in veterinary applications or in translational studies using dog models of neuromuscular diseases.

Could it be osteoarthritis? How dog owners and veterinary surgeons describe identifying canine osteoarthritis in a general practice setting

•

Owners describe a range of early behavioural indicators of canine osteoarthritis.

•

Their beliefs and prior knowledge may impact when and how they seek advice.

•

Vets in general practice describe a common “typical osteoarthritis” presentation.

•

History, examination and trial treatment are used to diagnose osteoarthritis.

•

Vets and owners may find osteoarthritis consultations frustrating and unrewarding.

Canine osteoarthritis is a common, painful condition that is typically managed in a general practice setting. Osteoarthritis may have significant negative impacts on the welfare of both dogs and their owners. Anticipated early clinical signs of canine osteoarthritis and the preferred route to its diagnosis are well described by veterinary subject experts in published literature. However, little is known about changes owners first recognise in a dog ultimately diagnosed with osteoarthritis, how they make decisions about when to present that dog to a general practitioner, or whether the described diagnostic pathways are followed by general practitioners. The aim of this research was to investigate how dog owners and veterinary surgeons describe identifying and diagnosing canine osteoarthritis.

Owners of osteoarthritic dogs were recruited for semi-structured interview, and veterinary surgeons working in general practice were invited to take part in practice-based focus groups. Transcripts from both datasets were thematically analysed using a contextualist epistemology with an ontology based on critical realism to construct convergent themes from latent and semantic codes. Thirty-two interviews were completed with 40 owners from 32 households who discussed 35 dogs with osteoarthritis, and 26 veterinary surgeons engaged in four practice-based focus groups.

Owners described identifying a wide range of acute and chronic, typically subtle and intermittent, behavioural and demeanour changes prior to their dogs’ osteoarthritis diagnosis. Few attributed these changes to canine osteoarthritis, and some waited many months before presenting their dog to a veterinary practice. Veterinary surgeons described a consistent ‘typical osteoarthritis’ presentation that they recognised through history taking and clinical examination. Their diagnostic work-up rarely followed that advocated by subject experts for reasons including lack of time and perceptions that it would not change the outcome. Many veterinary surgeons described frustration that some owners did not accept their recommendations to provide analgesia for affected dogs. Short consultation lengths, poor awareness of owner knowledge levels, and lack of recognition of the importance of owners’ prior knowledge, beliefs and assumptions may contribute to these consultations being perceived as challenging by some veterinary surgeons and owners.

This research demonstrates that veterinary surgeons and owners want dogs with clinical signs of osteoarthritis to be happy and comfortable, but that ineffective communication and lack of trust in the consulting room may be a barrier. Our data identifies many new avenues for future research and improved communication strategies that could facilitate earlier identification and treatment of canine osteoarthritis in general practice.

Review of kinematic analysis in dogs

Objective gait analysis techniques aid investigators in the study of motion. Kinematic gait analysis techniques that objectively quantitate motion are valuable tools used to understand normal and abnormal motion in domestic animals. Recent advances in video technology have made the study of motion more readily accessible. Available systems can document gait in two or three dimensions (2D or 3D, respectively). Knowledge of fundamental gait analysis concepts is critical to generating meaningful data. The objective of this report is to review principles of kinematic data collection and analyses, with a focus on differences between 2D and 3D systems.

Analysis of Agile Canine Gait Characteristics Using Accelerometry

The high rate of severe injuries associated with racing greyhounds poses a significant problem for both animal welfare and the racing industry. Using accelerometry to develop a better understanding of the complex gait of these agile canines may help to eliminate injury contributing factors. This study used a single Inertial Measurement Unit (IMU) equipped with a tri-axial accelerometer to characterise the galloping of thirty-one greyhounds on five different race tracks. The dorsal-ventral and anterior-posterior accelerations were analysed in both the time and frequency domains. The fast Fourier transform (FFT) and Morlet wavelet transform were applied to signals. The time-domain signals were synced with the corresponding high frame rate videos of the race. It was observed that the acceleration peaks in the dorsal-ventral accelerations correspond to the hind-leg strikes which were noted to be fifteen times the greyhound's weight. The FFT analysis showed that the stride frequencies in all tracks were around 3.5 Hz. The Morlet wavelet analysis also showed a reduction in both the frequency and magnitude of signals, which suggests a speed reduction throughout the race. Also, by detecting abrupt changes along the track, the wavelet analysis highlighted potentially hazardous locations on the track. In conclusion, the methods applied in this research contribute to animal safety and welfare by eliminating the factors leading to injuries through optimising the track design and surface type.

Foot-Mounted Inertial Measurement Units-Based Device for Ankle Rehabilitation

Ankle sprains are frequent injuries that occur among people of all ages. Ankle sprains constitute approximately 15% of all sports injuries, and are the most common traumatic emergencies. Without proper treatment and rehabilitation, a more severe sprain can weaken the ankle, making it more likely for new injures, and leading to long-term problems. In this work, we present an inertial measurement units (IMU)-based physical interface for measuring the foot attitude, and a graphical user interface that acts as a visual guide for patient rehabilitation. A foot-mounted physical interface for ankle rehabilitation was developed. The physical interface is connected to the computer by a Bluetooth link, and provides feedback to the patient while performing dorsiflexion, plantarflexion, eversion, and inversion exercises. The system allows for in-home rehabilitation at an affordable price while engaging the patient through active therapy. According to the results, more consistent rehabilitation could be achieved by providing feedback on foot angular position during therapy procedures.

A Wearable Magnet-Based System to Assess Activity and Joint Flexion in Humans and Large Animals

Functional outcomes, such as joint flexion and gait, are important indicators of efficacy in musculoskeletal research. Current technologies that objectively assess these parameters, including visual tracking systems and force plates, are challenging to deploy in long-term translational and clinical studies. To that end, we developed a wearable device that measures both physical activity and joint flexion using a single integrated sensor and magnet system, and hypothesized that it could evaluate post-operative functional recovery in an unsupervised setting. To demonstrate the feasibility of measuring joint flexion, we first compared knee motion from the wearable device to that acquired from a motion capture system to confirm that knee flexion measurements during normal human gait, predicted via changes in magnetic field strength, closely correlated with data acquired by motion capture. Using this system, we then monitored a porcine cohort after bilateral stifle arthrotomy to investigate longitudinal changes in physical activity and joint flexion. We found that unsupervised activity declined immediately after surgery, with a return to pre-operative activity occurring over a period of 2 weeks. By providing objective, individualized data on locomotion and joint function, this magnet-based system will facilitate the in vivo assessment of novel therapeutics in translational orthopaedic research.

Automatic characterization of stride parameters in canines with a single wearable inertial sensor

BACKGROUND AND OBJECTIVE

Gait analysis is valuable for studying neuromuscular and skeletal diseases. Wearable motion sensors or inertial measurement units (IMUs) have become common for human gait analysis. Canines are important large animal models for translational research of human diseases. Our objective is to develop a method for accurate and reliable determination of the timing of each stride in dogs using a wearable IMU.

METHODS

We built a wireless IMU sensor using off-the-shelf components. We also developed a MATLAB algorithm for data acquisition and stride timing determination. Stride parameters from 1,259 steps of three adult mixed breed dogs were determined across a range of six height-normalized speeds using the IMU system. The IMU results were validated by frame-by-frame manual counting of high-speed video recordings.

RESULTS

Comparing IMU derived results with video revealed that the mean error ± standard deviation for stride, stance, and swing duration was 0.001 ± 0.025, -0.001 ± 0.030, and 0.001 ± 0.019 s respectively. A mean error ± standard deviation of 0.000 ± 0.020 and -0.008 ± 0.027 s was obtained for determining toe-off and toe-touch events respectively. Only one step was missed by the algorithm in the video dataset of 1,259 steps.

CONCLUSION

We have developed and validated an IMU method for automatic canine gait analysis. Our method can be used for studying neuromuscular diseases in veterinary clinics and in translational research.

A Review on Accelerometry-Based Gait Analysis and Emerging Clinical Applications

Gait analysis continues to be an important technique for many clinical applications to diagnose and monitor certain diseases. Many mental and physical abnormalities cause measurable differences in a person's gait. Gait analysis has applications in sport, computer games, physical rehabilitation, clinical assessment, surveillance, human recognition, modeling, and many other fields. There are established methods using various sensors for gait analysis, of which accelerometers are one of the most often employed. Accelerometer sensors are generally more user friendly and less invasive. In this paper, we review research regarding accelerometer sensors used for gait analysis with particular focus on clinical applications. We provide a brief introduction to accelerometer theory followed by other popular sensing technologies. Commonly used gait phases and parameters are enumerated. The details of selecting the papers for review are provided. We also review several gait analysis software. Then we provide an extensive report of accelerometry-based gait analysis systems and applications, with additional emphasis on trunk accelerometry. We conclude this review with future research directions.

Evaluation of an open source method for calculating physical activity in dogs from harness and collar based sensors

BACKGROUND

The ability to make objective measurements of physical activity in dogs has both clinical and research applications. Accelerometers offer a non-intrusive and convenient solution. Of the commercialy available sensors, measurements are commonly given in manufacturer bespoke units and calculated with closed source approaches. Furthermore, the validation studies that exist for such devices are mounting location dependant, not transferable between brands or not suitable for handling modern raw accelerometry type data.

METHODS

This paper describes a validation study of n = 5 where 4 sensors were placed on each dog; 2 on a harness and 2 on a collar. Each position held two sensors from different manufacturers; Actigraph (which has previously been validated for use on the collar) and VetSens (which provides un-filtered accelerometry data). The aims of the study was to firstly evaluate the performance of an open-design method of converting raw accelerometry data into units that have previously been validated. Secondly, comparison was made between sensors mounted at each location for determining physical activity state.

RESULTS

Once the raw actigraphy data had been processed with the open-design method, results from a 7 day measurement revealed no significant difference in physical activity estimates via a cutpoint approach between the sensor manufacturers. A second finding was a low inter-site variability between the ventral collar and dorsal harness locations (Pearsons r2 = 0.96).

CONCLUSIONS

Using the open-design methodology, raw, un-filtered data from the VetSens sensors can be compared or pooled with data gathered from Actigraph sensors. The results also provide strong evidence that ventral collar and dorsal harness sites may be used interchangeably. This enables studies to be designed with a larger inclusion criteria (encompassing dogs that are not well suited for wearing an instrumented collar) and ensures high levels of welfare while maintaining measurement validity.