Low-Complexity Framework for Movement Classification Using Body-Worn Sensors

Abstract

Abstract-We present a low-complexity framework for classifying elementary arm-movements (reach-retrieve, lift-cup-tomouth, rotate-arm) using wrist-worn, inertial sensors. We propose that this methodology could be used as a clinical tool to assess rehabilitation progress in neurodegenerative pathologies tracking occurrence of specific movements performed by patients with their paretic arm. Movements performed in a controlled training-phase are processed to form unique clusters in a multi-dimensional feature-space. Subsequent movements performed in an uncontrolled testing-phase are associated to the proximal cluster using a minimum distance classifier (MDC). The framework involves performing the compute-intensive clustering on the training-dataset offline (Matlab) whereas the computation of selected features on the testing-dataset and the minimum distance (Euclidean) from pre-computed cluster centroids are done in hardware with an aim of low-power execution on sensor nodes.The architecture for feature-extraction and MDC are realized using Coordinate Rotation Digital Computer based design which classifies a movement in (9n+31) clock cycles, n being number of data samples. The design synthesized in STMicroelectronics 130nm technology consumed 5.3 nW @50 HZ, besides being functionally verified upto 20 MHz, making it applicable for realtime high-speed operations. Our experimental results show that the system can recognize all three arm-movements with average accuracies of 86% and 72% for four healthy subjects using accelerometer and gyroscope data respectively, whereas for stroke survivors the average accuracies were 67% and 60%. The framework was further demonstrated as a FPGA-based real-time system, interfacing with a streaming sensor unit.

Systematic review on the application of wearable inertial sensors to quantify everyday life motor activity in people with mobility impairments

“…to the mouth Duration Stroke [ 57 , 108 – 111 ] Classification accuracy [ 57 , 108 – 111 ] Pouring sth. (pro-/supination) Duration Stroke [ 108 – 111 ] Classification accuracy [ 108 – 111 ] Specific hand and arm activities Writing and reading Duration Parkinson’s disease [ 42 ] Classification accuracy [ 42 ] Opening a door Duration Arthritis [ 87 ], stroke [ 88 ] Classification accuracy [ 87 , 88 ] Hair combing Duration Stroke [ 57 , 112 ] Classification accuracy [ 57 , 112 ] Eating Duration Parkinson’s disease [ 42 ], stroke [ 112 ], miscellaneous [ 113 ] Classification accuracy [ 42 , 112 , 113 ] Drinking Duration Stroke [ 112 ] Classification accuracy [ ...…”

“…The first category includes measures to quantify the amount [ 30 , 40 , 46 , 47 , 89 , 103 , 104 ] and diversity [ 105 ] of hand and arm use as well as the range of motion of shoulder [ 34 , 56 , 58 , 106 ], elbow [ 34 , 58 ], and hand movements [ 107 ]. The second category contains reaching [ 34 , 58 , 72 , 108 – 111 ], lifting [ 57 , 108 – 111 ], and pouring (i.e. pro- and supination) movements [ 108 – 111 ], while reaching was further analyzed in terms of reaching distance [ 34 , 58 ] and reaching direction [ 72 ].…”

“…The second category contains reaching [ 34 , 58 , 72 , 108 – 111 ], lifting [ 57 , 108 – 111 ], and pouring (i.e. pro- and supination) movements [ 108 – 111 ], while reaching was further analyzed in terms of reaching distance [ 34 , 58 ] and reaching direction [ 72 ]. And the activities of the last category were writing and reading [ 42 ], opening a door [ 87 , 88 ], hair combing [ 57 , 112 ], eating [ 42 , 112 , 113 ], and drinking [ 112 ] as well as tooth brushing, shirt buttoning, pant lifting, and food cutting [ 57 ].…”

“…A more sophisticated approach to discriminate between functional and non-functional hand and arm use is to detect particular movement primitives such as reaching an object. One research group developed an algorithm that distinguishes between reaching, lifting, and pouring movements while making a cup of tea by using a single wrist sensor [ 108 – 111 ]. Another study specifically detected lifting food towards the mouth [ 57 ], and three studies detected reaching movements [ 34 , 58 , 72 ].…”

Healthcare Monitoring Using Low-Cost Sensors to Supplement and Replace Human Sensation: Does It Have Potential to Increase Independent Living and Prevent Disease?

“…Recognition of the appropriate comparison methodology for clinical studies is important, so the production of more randomised clinical trials would benefit sensor developments, attracting the attention of clinicians to the work. Among the searched literature, only four articles investigated daily activities using low-cost sensors worn by both healthy participants and PD patients [ 32 , 33 , 34 , 48 ]. Additionally, another publication studied the biomechanical characteristics of marathon athletes using wearable IMU sensors [ 36 ].…”