Detection of Low Back Physiotherapy Exercises With Inertial Sensors and Machine Learning: Algorithm Development and Validation.

BACKGROUND: Physiotherapy is a critical element in the successful conservative management of low back pain (LBP). A gold standard for quantitatively measuring physiotherapy participation is crucial to understanding physiotherapy adherence in managing recovery from LBP. OBJECTIVE: This study aimed to develop and evaluate a system with wearable inertial sensors to objectively detect the performance of unsupervised exercises for LBP comprising movement in multiple planes and sitting postures. METHODS: A quantitative classification design was used within a machine learning framework to detect exercise performance and posture in a cohort of healthy participants. A set of 8 inertial sensors were placed on the participants, and data were acquired as they performed 7 McKenzie low back exercises and 3 sitting posture positions. Engineered time series features were extracted from the data and used to train 9 models by using a 6-fold cross-validation approach, from which the best 2 models were selected for further study. In addition, a convolutional neural network was trained directly on the time series data. A feature importance analysis was performed to identify sensor locations and channels that contributed the most to the models. Finally, a subset of sensor locations and channels was included in a hyperparameter grid search to identify the optimal sensor configuration and best performing algorithms for exercise and posture classification. The final models were evaluated using the F1 score in a 10-fold cross-validation approach. RESULTS: In total, 19 healthy adults with no history of LBP each completed at least one full session of exercises and postures. Random forest and XGBoost (extreme gradient boosting) models performed the best out of the initial set of 9 engineered feature models. The optimal hardware configuration was identified as a 3-sensor setup-lower back, left thigh, and right ankle sensors with acceleration, gyroscope, and magnetometer channels. The XGBoost model achieved the highest exercise (F1 score: mean 0.94, SD 0.03) and posture (F1 score: mean 0.90, SD 0.11) classification scores. The convolutional neural network achieved similar results with the same sensor locations, using only the accelerometer and gyroscope channels for exercise classification (F1 score: mean 0.94, SD 0.02) and the accelerometer channel alone for posture classification (F1 score: mean 0.88, SD 0.07). CONCLUSIONS: This study demonstrates the potential of a 3-sensor lower body wearable solution (eg, smart pants) that can identify exercises in multiple planes and proper sitting postures, which is suitable for the treatment of LBP. This technology has the potential to improve the effectiveness of LBP rehabilitation by facilitating quantitative feedback, early problem diagnosis, and possible remote monitoring.

Personalized Activity Recognition with Deep Triplet Embeddings.

A significant challenge for a supervised learning approach to inertial human activity recognition is the heterogeneity of data generated by individual users, resulting in very poor performance for some subjects. We present an approach to personalized activity recognition based on deep feature representation derived from a convolutional neural network (CNN). We experiment with both categorical cross-entropy loss and triplet loss for training, and describe a novel loss function based on subject triplets. We evaluate these methods on three publicly available inertial human activity recognition datasets (MHEALTH, WISDM, and SPAR) comparing classification accuracy, out-of-distribution activity detection, and generalization to new activity classes. The proposed triplet algorithm achieved an average 96.7% classification accuracy across tested datasets versus the 87.5% achieved by the baseline CNN algorithm. We demonstrate that personalized algorithms, and, in particular, the proposed novel triplet loss algorithms, are more robust to inter-subject variability and thus exhibit better performance on classification and out-of-distribution detection tasks.

Physiotherapy Exercise Classification with Single-Camera Pose Detection and Machine Learning.

Access to healthcare, including physiotherapy, is increasingly occurring through virtual formats. At-home adherence to physical therapy programs is often poor and few tools exist to objectively measure participation. The aim of this study was to develop and evaluate the potential for performing automatic, unsupervised video-based monitoring of at-home low-back and shoulder physiotherapy exercises using a mobile phone camera. Joint locations were extracted from the videos of healthy subjects performing low-back and shoulder physiotherapy exercises using an open source pose detection framework. A convolutional neural network was trained to classify physiotherapy exercises based on the segments of keypoint time series data. The model's performance as a function of input keypoint combinations was studied in addition to its robustness to variation in the camera angle. The CNN model achieved optimal performance using a total of 12 pose estimation landmarks from the upper and lower body (low-back exercise classification: 0.995 ± 0.009; shoulder exercise classification: 0.963 ± 0.020). Training the CNN on a variety of angles was found to be effective in making the model robust to variations in video filming angle. This study demonstrates the feasibility of using a smartphone camera and a supervised machine learning model to effectively classify at-home physiotherapy participation and could provide a low-cost, scalable method for tracking adherence to physical therapy exercise programs in a variety of settings.

Automated detection of dental artifacts for large-scale radiomic analysis in radiation oncology.

BACKGROUND AND PURPOSE: Computed tomography (CT) is one of the most common medical imaging modalities in radiation oncology and radiomics research, the computational voxel-level analysis of medical images. Radiomics is vulnerable to the effects of dental artifacts (DA) caused by metal implants or fillings and can hamper future reproducibility on new datasets. In this study we seek to better understand the robustness of quantitative radiomic features to DAs. Furthermore, we propose a novel method of detecting DAs in order to safeguard radiomic studies and improve reproducibility. MATERIALS AND METHODS: We analyzed the correlations between radiomic features and the location of dental artifacts in a new dataset containing 3D CT scans from 3211 patients. We then combined conventional image processing techniques with a pre-trained convolutional neural network to create a three-class patient-level DA classifier and slice-level DA locator. Finally, we demonstrated its utility in reducing the correlations between the location of DAs and certain radiomic features. RESULTS: We found that when strong DAs were present, the proximity of the tumour to the mouth was highly correlated with 36 radiomic features. We predicted the correct DA magnitude yielding a Matthews correlation coefficient of 0.73 and location of DAs achieving the same level of agreement as human labellers. CONCLUSIONS: Removing radiomic features or CT slices containing DAs could reduce the unwanted correlations between the location of DAs and radiomic features. Automated DA detection can be used to improve the reproducibility of radiomic studies; an important step towards creating effective radiomic models for use in clinical radiation oncology.

Prediction of Human Papillomavirus (HPV) Association of Oropharyngeal Cancer (OPC) Using Radiomics: The Impact of the Variation of CT Scanner.

Studies have shown that radiomic features are sensitive to the variability of imaging parameters (e.g., scanner models), and one of the major challenges in these studies lies in improving the robustness of quantitative features against the variations in imaging datasets from multi-center studies. Here, we assess the impact of scanner choice on computed tomography (CT)-derived radiomic features to predict the association of oropharyngeal squamous cell carcinoma with human papillomavirus (HPV). This experiment was performed on CT image datasets acquired from two different scanner manufacturers. We demonstrate strong scanner dependency by developing a machine learning model to classify HPV status from radiological images. These experiments reveal the effect of scanner manufacturer on the robustness of radiomic features, and the extent of this dependency is reflected in the performance of HPV prediction models. The results of this study highlight the importance of implementing an appropriate approach to reducing the impact of imaging parameters on radiomic features and consequently on the machine learning models, without removing features which are deemed non-robust but may contain learning information.

Investigation of the Spatial Distribution of Methane Sources in the Greater Toronto Area Using Mobile Gas Monitoring Systems.

For methane emission reduction strategies in urban areas to be effective, large emitters must be identified. Recent studies in U.S. cities have highlighted the contribution of methane emissions from natural gas distribution networks and end use. We present a methane emission source identification and quantification method for the Greater Toronto Area (GTA), the largest metropolitan area in Canada, using mobile gas monitoring systems. From May 2018 to August 2019, we collected 77 surveys of methane mixing ratios, covering a distance of about 6400 km, and sampled emission plumes from sources such as closed landfills, natural gas compressor stations, and waterways. Our results indicate that inactive landfills emit less than inventory estimates. Despite this discrepancy, we confirm that the waste sector is the largest methane emitter in the GTA. We also report that the frequency of methane leaks from the local distribution system ranges between 4 and 22 leaks per 100 km of roadway in downtown Toronto, which is comparable to the range observed in U.S. cities, which have invested in modern natural gas distribution infrastructure. Last, we find that engineered waterways, whose emissions are currently not reported in inventories, may be a significant source of methane.

Prioritizing Urban Habitats for Connectivity Conservation: Integrating Centrality and Ecological Metrics.

Connectivity among fragmented areas of habitat has long been acknowledged as important for the viability of biological conservation, especially within highly modified landscapes. Identifying important habitat patches in ecological connectivity is a priority for many conservation strategies, and the application of 'graph theory' has been shown to provide useful information on connectivity. Despite the large number of metrics for connectivity derived from graph theory, only a small number have been compared in terms of the importance they assign to nodes in a network. This paper presents a study that aims to define a new set of metrics and compares these with traditional graph-based metrics, used in the prioritization of habitat patches for ecological connectivity. The metrics measured consist of "topological" metrics, "ecological metrics," and "integrated metrics," Integrated metrics are a combination of topological and ecological metrics. Eight metrics were applied to the habitat network for the fat-tailed dunnart within Greater Melbourne, Australia. A non-directional network was developed in which nodes were linked to adjacent nodes. These links were then weighted by the effective distance between patches. By applying each of the eight metrics for the study network, nodes were ranked according to their contribution to the overall network connectivity. The structured comparison revealed the similarity and differences in the way the habitat for the fat-tailed dunnart was ranked based on different classes of metrics. Due to the differences in the way the metrics operate, a suitable metric should be chosen that best meets the objectives established by the decision maker.