- Email: [email protected]
Accelerometer based calf muscle pump activity monitoring
Medical Engineering & Physics 27 (2005) 717–722
Technical note
Accelerometer based calf muscle pump activity monitoring Karol J. O’Donovan a, ∗ , Derek T. O’Keeffe a , Pierce A. Grace b , Gerard M. Lyons a a
Biomedical Electronics Laboratory, Department of Electronic and Computer Engineering, University of Limerick, Limerick, Ireland b Vascular Imaging Laboratory, Department of Vascular Surgery, Mid-Western Regional Hospital, Limerick, Ireland Received 9 July 2003; accepted 2 February 2005
Abstract Long distance travel is associated with increased risk of deep vein thrombosis (DVT). There is an increased risk of travel related DVT in passengers with a predisposition to thrombosis. Assisting blood circulation in the lower limb will reduce the risk of DVT. Leg exercises are recommended as a DVT preventative measure while flying but this fails to account for a passenger who is distracted by in flight entertainment or who falls asleep for an extended period. A method for monitoring calf muscle pump activity using accelerometers has been developed and evaluated. The proposed technique could be used to alert the traveller that there is a need to exercise their calf muscle, thus reducing the risk of DVT. © 2005 IPEM. Published by Elsevier Ltd. All rights reserved. Keywords: Deep vein thrombosis (DVT); Calf muscle pump; Lower leg; Accelerometers
1. Introduction In the lower leg, through a process known as “venous return”, the blood circulatory system is assisted by the so called “calf muscle pump”. The process of contraction and relaxation of the calf muscles propels blood through the venous network embedded in the muscles and back towards the heart [1]. This process is aided by a series of venous valves, which prevent the backflow of blood in the veins. Impaired calf muscle activity may result from prolonged periods of immobility. Failure to exercise the calf muscle pump may result in stasis and venous hypertension due to pooling of blood in the leg. Medical pathologies associated with impaired calf muscle pump activity include deep venous thrombosis (DVT) and venous ulceration. DVT is a condition in which blood clots; known as thrombi, form in major veins in the leg or pelvis. Following a DVT the affected vein may remain permanently occluded or recanalise. The process of recanalisation results in venous valve destruction, venous hypertension and, if severe, may result in a clinical syndrome known as the ‘postphlebitic’ or ‘post-thrombotic’ limb. The clinical features of ∗
Corresponding author. Tel.: +353 61 213102; fax: +353 61 338176. E-mail address: [email protected] (K.J. O’Donovan).
post-phlebitic limb include venous eczema, pain, swelling, lipodermatosclerosis and recurrent ulcerations [2]. A further, potentially fatal consequence of DVT is pulmonary thromboembolism (PTE). PTE results when part of the thrombus breaks off (embolises) and travels to the lung circulation, where it lodges in and blocks one of the pulmonary arteries. This in turn can lead to pulmonary hypertension, heart failure and death [3]. According to a study by Bergqvist et al. [4], the economic effect of post-thrombotic complications is considerable and the use of measures to prevent thromboembolism and its long-term complications is justified on both clinical and economic grounds. Traditional DVT prevention methods focus on counter acting sluggish venous blood flow by assisting or enhancing the contraction of the calf muscles. Intermittent pneumatic compression [5,6], graduated compression stockings [7,8] and structured exercises [9] have all been shown to increase venous blood flow and have been used in DVT prevention. The association between long distance travel and DVT was first established as early as 1954 when Homans reported four suspected cases [10]. Cramped seating conditions and general lack of activity during travel can lead to poor calf muscle pump activity. Recent studies [11–13] have shown that a history of long distance travel is associated with an
1350-4533/$ – see front matter © 2005 IPEM. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.medengphy.2005.02.002
718
K.J. O’Donovan et al. / Medical Engineering & Physics 27 (2005) 717–722
increased risk of DVT. Other studies [14–16] have revealed that patients with travel-related DVT usually have additional risk factors. The LONFLIT study by Belcaro et al. revealed that DVT can occur in up to 2.8% of high risk passengers on flights with duration of between 12 and 15 h [17]. Passengers at high risk from DVT are recommended to exercise on a regular basis whilst on long distance flights [18]. The use of in-travel exercise aids such as inflatable pillows is an accepted practice in DVT prevention; however, these techniques require the traveller to make a conscious effort to exercise at regular intervals. On long distance flights, a combination of distractions such as in-flight entertainment and the tendency for passengers to sleep for extended periods can limit the capacity of the passenger to adhere to the required exercise program. If the passenger fails to complete the required exercise program they are at increased risk of DVT. This paper describes a technique that monitors calf muscle pump activity. The proposed technique could be used to alert the traveller that there is a need to exercise their calf muscle should they have forgotten to exercise or failed to exercise sufficiently. Although muscle activity can be quite easily monitored in a clinical environment using EMG sensors, it is not practical to implement EMG sensors into an every day device. Outside the laboratory environment EMG recordings can be difficult to set-up, requiring careful skin preparation and precise electrode placement. Also drying of electrodes can result in variability in the performance of the sensor over time and would be expected to occur during a long haul flight. To avoid the use of EMG, it is proposed to use miniature accelerometers as a surrogate sensor system to detect contraction of the calf muscle and hence calf muscle pump activity.
The developed technique uses two accelerometers and a control algorithm to detect calf muscle pump activity. One accelerometer is placed on the heel of the foot (signal H) and the other is placed at the lateral side of the lower leg in close proximity to the calf (signal S). Fig. 1 shows the set-up used for the technique with the location of the accelerometers indicated. The orientation of the accelerometers is also indicated by the direction of the arrows.
2. Methods 2.1. Algorithm design The proposed technique uses feedback information from one axis from each of the two ADXL202E1 accelerometers (active axis is vertical when person is standing upright) to determine if the calf muscle pump has been active or not. The accelerometer signals are sampled at 1 kHz. The algorithm first computes F(j), the mean absolute value of the heel accelerometer signal H(i), for the previous 2 s (2000 samples) using Eq. (1).
F (j) =
j−2000 1 |H(i)| 2
(1)
i=j
F(j) is an indication of the level of activity at the heel for the previous 2 s. A threshold ThF is used to distinguish between levels of high activity and low activity for F(j). Secondly, G(j), the mean absolute value of the difference between signal H(i) and the lower leg accelerometer signal S(i), for the previous
Fig. 1. Accelerometer set-up for calf muscle pump activity detection.
K.J. O’Donovan et al. / Medical Engineering & Physics 27 (2005) 717–722
719
Fig. 2. Calf muscle pump activity detection algorithm.
2 s, is calculated using Eq. (2). G(j) =
j−2000 1 |H(i) − S(t)| 2
(2)
i=j
G(j) is an indication of the level of activity at the heel with respect to the lower leg for the previous 2 s. A threshold ThG is used to distinguish between levels of high activity and levels of low activity for G(j). C(j), the calf muscle pump activity detected binary signal is then calculated using the algorithm in Fig. 2. The algorithm concludes that if at a given instant, F(j) is greater than ThF and G(j) is greater than ThG the calf muscle pump has been active (C(j) = 1), otherwise the calf muscle pump is considered to have been inactive (C(j) = 0). 2.2. Movement sensors Fig. 3. Experimental set-up.
devices1
The accelerometers used are the analog ADXL202E, dual–axial accelerometers with a full-scale range of ±2 g. The ADXL202E provides both analogue and digital outputs; the analogue outputs are used in this application. Capacitors were added at the analogue output pins implementing low-pass filtering with a cut-off frequency of 15 Hz.
using a Biomedical Monitor Ltd.2 programmable BM42 data logger. The experimental set-up is shown in Fig. 3. During recording the subject was seated and asked to carry out eight leg exercises based on those recommended by major airline companies as DVT preventive measures. Each exercise was repeated 10 times with a distinct pause between each repetition. The eight exercises were:
2.3. Evaluation An experimental trial was designed to evaluate the use of accelerometers to detect calf muscle pump activity. Ethical approval was obtained for this experiment from the University of Limerick Research Ethics Committee. One healthy male subject aged 24 was recruited for this experiment. The subject was provided with a volunteer information sheet and asked to give written conformed consent to participate in the trial. An ADXL202E accelerometer was attached to the lateral side of the lower leg in close proximity to the calf and another was attached to the heel of the foot. One axis from each accelerometer was observed and each accelerometer was positioned such that the active axis was vertical when the person was standing upright. The accelerometers outputs had a low pass filter with a cut-off frequency of 15 Hz. A single bipolar surface EMG sensor was placed on the calf muscle to monitor calf muscle activity. The EMG signal was passed through a band-pass filter with a lower cut-off frequency of 20 Hz and upper cut-off frequency of 400 Hz. Both the accelerometer and EMG sensor signals were sampled at 1 kHz 1
Analog Devices Inc., One Technology Way, Norwood, MA 02062-9106, USA.
(1) (2) (3) (4) (5) (6) (7) (8)
dorsiflexion; plantar flexion; foot abduction; foot abduction with dorsiflexion; foot adduction; foot adduction with dorsiflexion; knee and hip flexion with minimal foot movement; knee and hip flexion with plantar flexion.
The MATLAB3 computing program was used for all posttrial analysis. The raw EMG signal was band-pass filtered with a lower cut-off frequency of 20 Hz and upper cut-off frequency of 400 Hz. The filtered EMG signal was then root mean square (RMS) averaged using a moving window of length 249 ms. The accelerometer signals were band-pass filtered with a lower cut-off frequency of 0.5 Hz and upper cutoff frequency of 4 Hz. The first and last repetition of each exercise was discarded from the analysis, thus eight repetitions of each exercise were analysed.
2 Biomedical Monitoring Ltd., Wolfson Centre, 106 Rottenrow, Glasgow, G40NM, UK. 3 The MathWorks Inc., 3 Apple Hill Drive, Natick, MA 01760-2098, USA.
720
K.J. O’Donovan et al. / Medical Engineering & Physics 27 (2005) 717–722
3. Results and discussion The accelerometer signals H(i) and S(i) were used to generate the signals F(j), G(j) and C(j) based on the algorithm described in Section 2.1. Fig. 4 contains four columns of subplots, Fig. 4a–d. Each column of subplots contains the various signals for a specific exercise over a 4 s period. One repetition of each respective exercise is performed in the 4 s period. All signals have been normalised with a maximum value of 1. The first row of subplots contains signal E(i), the RMS EMG signal. A threshold ThE indicates whether the calf muscle was considered active or not. The second and third rows contain the signals H(i) and S(i), the accelerometer signals at the heel and lower leg, respectively. The fourth row contains signal F(j), the mean absolute value of H(i) for the previous 2 s period, the threshold ThF is also shown. The fifth row contains the signal G(j), the mean absolute value of the difference between H(i) and S(i) for the previous 2 s period, along with the threshold ThG . The final row contains the signal C(i), the accelerometer detected calf muscle activity calculated using signals F(j) and G(j), and their respective thresholds. In Fig. 4a (exercise 3, foot abduction), there was very little EMG activity, with a corresponding low level of heel activity and also a low level of heel activity with respect
to the lower leg. The algorithm correctly detected no calf muscle pump activity. In Fig. 4b (exercise 4, foot abduction with dorsiflexion), once again there was no significant EMG activity. The level of activity of the heel with respect to the lower leg was above the threshold ThG , however, there was insufficient activity at the heel and the algorithm correctly detected no calf muscle pump activity. Fig. 4c (exercise 7, knee and hip flexion with minimal foot movement) also has no significant level of EMG activity. There was a high level of heel activity, however, the movement of the heel with respect to the lower leg was below the threshold ThG and the algorithm detected no calf muscle pump activity. In Fig. 4d (exercise 8, knee and hip flexion with plantar flexion), there was a high level of EMG activity and as in Fig. 4c, high levels of activity at the heel. However, there was also a large degree of activity of the heel with respect to the lower leg and the algorithm correctly detected calf muscle pump activity. Fig. 5 contains four subplots, Fig. 5a–d. The plots show the signals E(i), F(j), G(j) and C(j) for the eight repetitions of the eight exercises. The periodic vertical lines divide the plots into the different exercises. In Fig. 5a, it can be seen that exercises 2 and 8 induced high levels of EMG activity in the calf muscle. In Fig. 5d it can be seen that the algorithm correctly detected calf muscle pump activity on each repetition of each exercise.
Fig. 4. Calf muscle EMG, heel and lower leg accelerometer signals and derived signals F(j), G(j) and C(j) for single repetition of exercises 3, 4, 7 and 8.
K.J. O’Donovan et al. / Medical Engineering & Physics 27 (2005) 717–722
721
Fig. 5. Calf muscle EMG and derived signals F(j), G(j) and C(j) for eight repetitions of all exercises.
4. Conclusion Travel related deep vein thrombosis has been shown to be a serious and significant problem due in part to the increased popularity of long distance travel. Current preventative techniques such as in flight exercises have inherent limitations, as they require the conscious effort of the passenger who may be distracted or asleep. A novel method for monitoring calf muscle pump activity using accelerometers has been developed and evaluated. The use of accelerometers as EMG substitute sensors to detect calf muscle activity has several advantages including ease of set up and reduced associated circuitry. The preliminary clinical findings of this study show promise for a valuable technique in detecting calf muscle activity.
Acknowledgements The authors would like to thank IRCSET for their sponsorship of this project and Analog Devices BV, Limerick, Ireland for providing the ADXL202E devices.
References [1] Arnold CC. Physiology and pathophysiology of the venous pump of the calf. In: Eklof B, Cjorcs JE, Thulesius O, Bergqvist D, editors. Controversies in the management of venous disorders. London: Butterworth; 1989. p. 6–23. [2] Hopkins NFG, Wolfe JHN. Deep venous insufficiency and occlusion. In: Wolfe JHN, editor. ABC of vascular diseases. London: BMJ; 1992. p. 47–50. [3] Hopkins NFG, Wolfe JHN. Thrombosis and pulmonary embolism. In: Wolfe JHN, editor. ABC of vascular diseases. London: BMJ; 1992. p. 44–6. [4] Bergqvist D, Jendteg S, Johansen L, Persson U, et al. Cost of longterm complications of deep venous thrombosis of the lower extremities: an analysis of a defined patient population in Sweden. Ann Intern Med 1997;126:454–7. [5] Kamm RD. Bioengineering studies of periodic external compression as prophylaxis against deep vein thrombosis. Part I: Numerical studies. J Biomech Eng 1982;104:87–95. [6] Inada K, Koike S, Shirai N, Matsumoto K, et al. Effects of intermittent pneumatic leg compression for prevention of postoperative deep venous thrombosis with special reference to fibrinolytic activity. Am J Surg 1988;155:602–5. [7] Christopoulos DG, Nicolaides AN, Szendro G, Irvine AT, et al. Air-plethysmography and the effect of elastic compression
722
[8]
[9]
[10] [11]
[12]
K.J. O’Donovan et al. / Medical Engineering & Physics 27 (2005) 717–722 on venous hemodynamics of the leg. J Vasc Surg 1987;5:148– 59. Colditz GA, Tuden RL, Oster G. Rates of venous thrombosis after general surgery: combined results of randomised clinical trials. Lancet 1986;2:143–6. Yang D, Vandongen YK, Stacey MC. Effect of exercise on calf muscle pump function in patients with chronic venous disease. Br J Surg 1999;86:338–41. Homans J. Thrombosis of the deep leg veins due to prolonged sitting. N Engl J Med 1954;250:148–9. Ferrari E, Chevallier T, Chapelier A, Baudouy M. Travel as a risk factor for venous thromboembolic disease: a case–control study. Chest 1999;115:440–4. Samama MM. An epidemiologic study of risk factors for deep vein thrombosis in medical outpatients: the Sirius study. Arch Intern Med 2000;160:3415–20.
[13] Scurr JH, Machin SJ, Bailey-King S, Mackie IJ, et al. Frequency and prevention of symptomless deep-vein thrombosis in long-haul flights: a randomised trial. Lancet 2001;357:1485–9. [14] Eklof B, Kistner RL, Masuda EM, Sonntag BV, et al. Venous thromboembolism in association with prolonged air travel. Dermatol Surg 1996;22:637–41. [15] Mercer A, Brown JD. Venous thromboembolism associated with air travel: a report of 33 patients. Aviat Space Environ Med 1998;69:154–7. [16] Rege KP, Bevan DH, Chitolie A, Shannon MS. Risk factors and thrombosis after airline flight. Thromb Haemost 1999;81:995–6. [17] Belcaro G, Geroulakos G, Nicolaides AN, Myers KA, et al. Venous thromboembolism from air travel: the LONFLIT study. Angiology 2001;52:369–74. [18] Travel by air: health considerations. In: International travel and health. World Health Organization; 2004.
Technical note
Accelerometer based calf muscle pump activity monitoring Karol J. O’Donovan a, ∗ , Derek T. O’Keeffe a , Pierce A. Grace b , Gerard M. Lyons a a
Biomedical Electronics Laboratory, Department of Electronic and Computer Engineering, University of Limerick, Limerick, Ireland b Vascular Imaging Laboratory, Department of Vascular Surgery, Mid-Western Regional Hospital, Limerick, Ireland Received 9 July 2003; accepted 2 February 2005
Abstract Long distance travel is associated with increased risk of deep vein thrombosis (DVT). There is an increased risk of travel related DVT in passengers with a predisposition to thrombosis. Assisting blood circulation in the lower limb will reduce the risk of DVT. Leg exercises are recommended as a DVT preventative measure while flying but this fails to account for a passenger who is distracted by in flight entertainment or who falls asleep for an extended period. A method for monitoring calf muscle pump activity using accelerometers has been developed and evaluated. The proposed technique could be used to alert the traveller that there is a need to exercise their calf muscle, thus reducing the risk of DVT. © 2005 IPEM. Published by Elsevier Ltd. All rights reserved. Keywords: Deep vein thrombosis (DVT); Calf muscle pump; Lower leg; Accelerometers
1. Introduction In the lower leg, through a process known as “venous return”, the blood circulatory system is assisted by the so called “calf muscle pump”. The process of contraction and relaxation of the calf muscles propels blood through the venous network embedded in the muscles and back towards the heart [1]. This process is aided by a series of venous valves, which prevent the backflow of blood in the veins. Impaired calf muscle activity may result from prolonged periods of immobility. Failure to exercise the calf muscle pump may result in stasis and venous hypertension due to pooling of blood in the leg. Medical pathologies associated with impaired calf muscle pump activity include deep venous thrombosis (DVT) and venous ulceration. DVT is a condition in which blood clots; known as thrombi, form in major veins in the leg or pelvis. Following a DVT the affected vein may remain permanently occluded or recanalise. The process of recanalisation results in venous valve destruction, venous hypertension and, if severe, may result in a clinical syndrome known as the ‘postphlebitic’ or ‘post-thrombotic’ limb. The clinical features of ∗
Corresponding author. Tel.: +353 61 213102; fax: +353 61 338176. E-mail address: [email protected] (K.J. O’Donovan).
post-phlebitic limb include venous eczema, pain, swelling, lipodermatosclerosis and recurrent ulcerations [2]. A further, potentially fatal consequence of DVT is pulmonary thromboembolism (PTE). PTE results when part of the thrombus breaks off (embolises) and travels to the lung circulation, where it lodges in and blocks one of the pulmonary arteries. This in turn can lead to pulmonary hypertension, heart failure and death [3]. According to a study by Bergqvist et al. [4], the economic effect of post-thrombotic complications is considerable and the use of measures to prevent thromboembolism and its long-term complications is justified on both clinical and economic grounds. Traditional DVT prevention methods focus on counter acting sluggish venous blood flow by assisting or enhancing the contraction of the calf muscles. Intermittent pneumatic compression [5,6], graduated compression stockings [7,8] and structured exercises [9] have all been shown to increase venous blood flow and have been used in DVT prevention. The association between long distance travel and DVT was first established as early as 1954 when Homans reported four suspected cases [10]. Cramped seating conditions and general lack of activity during travel can lead to poor calf muscle pump activity. Recent studies [11–13] have shown that a history of long distance travel is associated with an
1350-4533/$ – see front matter © 2005 IPEM. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.medengphy.2005.02.002
718
K.J. O’Donovan et al. / Medical Engineering & Physics 27 (2005) 717–722
increased risk of DVT. Other studies [14–16] have revealed that patients with travel-related DVT usually have additional risk factors. The LONFLIT study by Belcaro et al. revealed that DVT can occur in up to 2.8% of high risk passengers on flights with duration of between 12 and 15 h [17]. Passengers at high risk from DVT are recommended to exercise on a regular basis whilst on long distance flights [18]. The use of in-travel exercise aids such as inflatable pillows is an accepted practice in DVT prevention; however, these techniques require the traveller to make a conscious effort to exercise at regular intervals. On long distance flights, a combination of distractions such as in-flight entertainment and the tendency for passengers to sleep for extended periods can limit the capacity of the passenger to adhere to the required exercise program. If the passenger fails to complete the required exercise program they are at increased risk of DVT. This paper describes a technique that monitors calf muscle pump activity. The proposed technique could be used to alert the traveller that there is a need to exercise their calf muscle should they have forgotten to exercise or failed to exercise sufficiently. Although muscle activity can be quite easily monitored in a clinical environment using EMG sensors, it is not practical to implement EMG sensors into an every day device. Outside the laboratory environment EMG recordings can be difficult to set-up, requiring careful skin preparation and precise electrode placement. Also drying of electrodes can result in variability in the performance of the sensor over time and would be expected to occur during a long haul flight. To avoid the use of EMG, it is proposed to use miniature accelerometers as a surrogate sensor system to detect contraction of the calf muscle and hence calf muscle pump activity.
The developed technique uses two accelerometers and a control algorithm to detect calf muscle pump activity. One accelerometer is placed on the heel of the foot (signal H) and the other is placed at the lateral side of the lower leg in close proximity to the calf (signal S). Fig. 1 shows the set-up used for the technique with the location of the accelerometers indicated. The orientation of the accelerometers is also indicated by the direction of the arrows.
2. Methods 2.1. Algorithm design The proposed technique uses feedback information from one axis from each of the two ADXL202E1 accelerometers (active axis is vertical when person is standing upright) to determine if the calf muscle pump has been active or not. The accelerometer signals are sampled at 1 kHz. The algorithm first computes F(j), the mean absolute value of the heel accelerometer signal H(i), for the previous 2 s (2000 samples) using Eq. (1).
F (j) =
j−2000 1 |H(i)| 2
(1)
i=j
F(j) is an indication of the level of activity at the heel for the previous 2 s. A threshold ThF is used to distinguish between levels of high activity and low activity for F(j). Secondly, G(j), the mean absolute value of the difference between signal H(i) and the lower leg accelerometer signal S(i), for the previous
Fig. 1. Accelerometer set-up for calf muscle pump activity detection.
K.J. O’Donovan et al. / Medical Engineering & Physics 27 (2005) 717–722
719
Fig. 2. Calf muscle pump activity detection algorithm.
2 s, is calculated using Eq. (2). G(j) =
j−2000 1 |H(i) − S(t)| 2
(2)
i=j
G(j) is an indication of the level of activity at the heel with respect to the lower leg for the previous 2 s. A threshold ThG is used to distinguish between levels of high activity and levels of low activity for G(j). C(j), the calf muscle pump activity detected binary signal is then calculated using the algorithm in Fig. 2. The algorithm concludes that if at a given instant, F(j) is greater than ThF and G(j) is greater than ThG the calf muscle pump has been active (C(j) = 1), otherwise the calf muscle pump is considered to have been inactive (C(j) = 0). 2.2. Movement sensors Fig. 3. Experimental set-up.
devices1
The accelerometers used are the analog ADXL202E, dual–axial accelerometers with a full-scale range of ±2 g. The ADXL202E provides both analogue and digital outputs; the analogue outputs are used in this application. Capacitors were added at the analogue output pins implementing low-pass filtering with a cut-off frequency of 15 Hz.
using a Biomedical Monitor Ltd.2 programmable BM42 data logger. The experimental set-up is shown in Fig. 3. During recording the subject was seated and asked to carry out eight leg exercises based on those recommended by major airline companies as DVT preventive measures. Each exercise was repeated 10 times with a distinct pause between each repetition. The eight exercises were:
2.3. Evaluation An experimental trial was designed to evaluate the use of accelerometers to detect calf muscle pump activity. Ethical approval was obtained for this experiment from the University of Limerick Research Ethics Committee. One healthy male subject aged 24 was recruited for this experiment. The subject was provided with a volunteer information sheet and asked to give written conformed consent to participate in the trial. An ADXL202E accelerometer was attached to the lateral side of the lower leg in close proximity to the calf and another was attached to the heel of the foot. One axis from each accelerometer was observed and each accelerometer was positioned such that the active axis was vertical when the person was standing upright. The accelerometers outputs had a low pass filter with a cut-off frequency of 15 Hz. A single bipolar surface EMG sensor was placed on the calf muscle to monitor calf muscle activity. The EMG signal was passed through a band-pass filter with a lower cut-off frequency of 20 Hz and upper cut-off frequency of 400 Hz. Both the accelerometer and EMG sensor signals were sampled at 1 kHz 1
Analog Devices Inc., One Technology Way, Norwood, MA 02062-9106, USA.
(1) (2) (3) (4) (5) (6) (7) (8)
dorsiflexion; plantar flexion; foot abduction; foot abduction with dorsiflexion; foot adduction; foot adduction with dorsiflexion; knee and hip flexion with minimal foot movement; knee and hip flexion with plantar flexion.
The MATLAB3 computing program was used for all posttrial analysis. The raw EMG signal was band-pass filtered with a lower cut-off frequency of 20 Hz and upper cut-off frequency of 400 Hz. The filtered EMG signal was then root mean square (RMS) averaged using a moving window of length 249 ms. The accelerometer signals were band-pass filtered with a lower cut-off frequency of 0.5 Hz and upper cutoff frequency of 4 Hz. The first and last repetition of each exercise was discarded from the analysis, thus eight repetitions of each exercise were analysed.
2 Biomedical Monitoring Ltd., Wolfson Centre, 106 Rottenrow, Glasgow, G40NM, UK. 3 The MathWorks Inc., 3 Apple Hill Drive, Natick, MA 01760-2098, USA.
720
K.J. O’Donovan et al. / Medical Engineering & Physics 27 (2005) 717–722
3. Results and discussion The accelerometer signals H(i) and S(i) were used to generate the signals F(j), G(j) and C(j) based on the algorithm described in Section 2.1. Fig. 4 contains four columns of subplots, Fig. 4a–d. Each column of subplots contains the various signals for a specific exercise over a 4 s period. One repetition of each respective exercise is performed in the 4 s period. All signals have been normalised with a maximum value of 1. The first row of subplots contains signal E(i), the RMS EMG signal. A threshold ThE indicates whether the calf muscle was considered active or not. The second and third rows contain the signals H(i) and S(i), the accelerometer signals at the heel and lower leg, respectively. The fourth row contains signal F(j), the mean absolute value of H(i) for the previous 2 s period, the threshold ThF is also shown. The fifth row contains the signal G(j), the mean absolute value of the difference between H(i) and S(i) for the previous 2 s period, along with the threshold ThG . The final row contains the signal C(i), the accelerometer detected calf muscle activity calculated using signals F(j) and G(j), and their respective thresholds. In Fig. 4a (exercise 3, foot abduction), there was very little EMG activity, with a corresponding low level of heel activity and also a low level of heel activity with respect
to the lower leg. The algorithm correctly detected no calf muscle pump activity. In Fig. 4b (exercise 4, foot abduction with dorsiflexion), once again there was no significant EMG activity. The level of activity of the heel with respect to the lower leg was above the threshold ThG , however, there was insufficient activity at the heel and the algorithm correctly detected no calf muscle pump activity. Fig. 4c (exercise 7, knee and hip flexion with minimal foot movement) also has no significant level of EMG activity. There was a high level of heel activity, however, the movement of the heel with respect to the lower leg was below the threshold ThG and the algorithm detected no calf muscle pump activity. In Fig. 4d (exercise 8, knee and hip flexion with plantar flexion), there was a high level of EMG activity and as in Fig. 4c, high levels of activity at the heel. However, there was also a large degree of activity of the heel with respect to the lower leg and the algorithm correctly detected calf muscle pump activity. Fig. 5 contains four subplots, Fig. 5a–d. The plots show the signals E(i), F(j), G(j) and C(j) for the eight repetitions of the eight exercises. The periodic vertical lines divide the plots into the different exercises. In Fig. 5a, it can be seen that exercises 2 and 8 induced high levels of EMG activity in the calf muscle. In Fig. 5d it can be seen that the algorithm correctly detected calf muscle pump activity on each repetition of each exercise.
Fig. 4. Calf muscle EMG, heel and lower leg accelerometer signals and derived signals F(j), G(j) and C(j) for single repetition of exercises 3, 4, 7 and 8.
K.J. O’Donovan et al. / Medical Engineering & Physics 27 (2005) 717–722
721
Fig. 5. Calf muscle EMG and derived signals F(j), G(j) and C(j) for eight repetitions of all exercises.
4. Conclusion Travel related deep vein thrombosis has been shown to be a serious and significant problem due in part to the increased popularity of long distance travel. Current preventative techniques such as in flight exercises have inherent limitations, as they require the conscious effort of the passenger who may be distracted or asleep. A novel method for monitoring calf muscle pump activity using accelerometers has been developed and evaluated. The use of accelerometers as EMG substitute sensors to detect calf muscle activity has several advantages including ease of set up and reduced associated circuitry. The preliminary clinical findings of this study show promise for a valuable technique in detecting calf muscle activity.
Acknowledgements The authors would like to thank IRCSET for their sponsorship of this project and Analog Devices BV, Limerick, Ireland for providing the ADXL202E devices.
References [1] Arnold CC. Physiology and pathophysiology of the venous pump of the calf. In: Eklof B, Cjorcs JE, Thulesius O, Bergqvist D, editors. Controversies in the management of venous disorders. London: Butterworth; 1989. p. 6–23. [2] Hopkins NFG, Wolfe JHN. Deep venous insufficiency and occlusion. In: Wolfe JHN, editor. ABC of vascular diseases. London: BMJ; 1992. p. 47–50. [3] Hopkins NFG, Wolfe JHN. Thrombosis and pulmonary embolism. In: Wolfe JHN, editor. ABC of vascular diseases. London: BMJ; 1992. p. 44–6. [4] Bergqvist D, Jendteg S, Johansen L, Persson U, et al. Cost of longterm complications of deep venous thrombosis of the lower extremities: an analysis of a defined patient population in Sweden. Ann Intern Med 1997;126:454–7. [5] Kamm RD. Bioengineering studies of periodic external compression as prophylaxis against deep vein thrombosis. Part I: Numerical studies. J Biomech Eng 1982;104:87–95. [6] Inada K, Koike S, Shirai N, Matsumoto K, et al. Effects of intermittent pneumatic leg compression for prevention of postoperative deep venous thrombosis with special reference to fibrinolytic activity. Am J Surg 1988;155:602–5. [7] Christopoulos DG, Nicolaides AN, Szendro G, Irvine AT, et al. Air-plethysmography and the effect of elastic compression
722
[8]
[9]
[10] [11]
[12]
K.J. O’Donovan et al. / Medical Engineering & Physics 27 (2005) 717–722 on venous hemodynamics of the leg. J Vasc Surg 1987;5:148– 59. Colditz GA, Tuden RL, Oster G. Rates of venous thrombosis after general surgery: combined results of randomised clinical trials. Lancet 1986;2:143–6. Yang D, Vandongen YK, Stacey MC. Effect of exercise on calf muscle pump function in patients with chronic venous disease. Br J Surg 1999;86:338–41. Homans J. Thrombosis of the deep leg veins due to prolonged sitting. N Engl J Med 1954;250:148–9. Ferrari E, Chevallier T, Chapelier A, Baudouy M. Travel as a risk factor for venous thromboembolic disease: a case–control study. Chest 1999;115:440–4. Samama MM. An epidemiologic study of risk factors for deep vein thrombosis in medical outpatients: the Sirius study. Arch Intern Med 2000;160:3415–20.
[13] Scurr JH, Machin SJ, Bailey-King S, Mackie IJ, et al. Frequency and prevention of symptomless deep-vein thrombosis in long-haul flights: a randomised trial. Lancet 2001;357:1485–9. [14] Eklof B, Kistner RL, Masuda EM, Sonntag BV, et al. Venous thromboembolism in association with prolonged air travel. Dermatol Surg 1996;22:637–41. [15] Mercer A, Brown JD. Venous thromboembolism associated with air travel: a report of 33 patients. Aviat Space Environ Med 1998;69:154–7. [16] Rege KP, Bevan DH, Chitolie A, Shannon MS. Risk factors and thrombosis after airline flight. Thromb Haemost 1999;81:995–6. [17] Belcaro G, Geroulakos G, Nicolaides AN, Myers KA, et al. Venous thromboembolism from air travel: the LONFLIT study. Angiology 2001;52:369–74. [18] Travel by air: health considerations. In: International travel and health. World Health Organization; 2004.