RESEARCH/Original article
Clinical applicability of real-time, prehospital image transmission for FAST (Focused Assessment with Sonography for Trauma)
Journal of Telemedicine and Telecare 19(8) 450–455 ! The Author(s) 2013 Reprints and permissions: sagepub.co.uk/journalsPermissions.nav DOI: 10.1177/1357633X13512068 jtt.sagepub.com
Kyoung Jun Song1,2, Sang Do Shin1, Ki Jeong Hong1, Kyoung Woo Cheon3, Ilhyoung Shin3, Sung-Wook Song2,4 and Hee Chan Kim2,3
Summary We evaluated a real-time, prehospital ultrasound image transmission system for use in focused assessment with sonography for trauma (FAST). The wireless, real-time ultrasound image transmission system comprised an ultrasound scanner with a convex abdominal transducer and a notebook computer connected to a 3 G wireless network for video data transmission. In our simulation experiment, ultrasonography was performed by emergency medical technicians (EMTs) on a human body phantom with simulated haemoperitoneum. Transmitted ultrasound video clips were randomly rearranged and presented to emergency physicians to make a diagnosis of haemoperitoneum. A total of 21 ultrasound video clips was used and 13 emergency physicians participated in the study. The sensitivity and specificity were 90.0% (95% Confidence Interval, CI, 83.5–94.6) and 85.3% (95% CI 78.4–90.7) respectively, and the accuracy of detecting abnormal ultrasound results was 87.7% (95% CI 83.8–91.6). Diagnosis of hemoperitonuem in trauma patients by an emergency physician based on the transmitted video images of FAST performed by an EMT is feasible, and has an accuracy of about 88%. Accepted: 21 September 2013
Introduction During the last ten years there has been a substantial improvement in the overall quality of the emergency medical service (EMS) in Korea. However, prehospital EMS in Korea provides only basic life support. The concept of prehospital emergency medicine is changing from basic life support which focuses on rescue and transport, to advanced life support (ALS) which emphasizes rescue, evaluation, treatment, triage and transport. For this new role, it is necessary to have ambulances with sophisticated equipment and emergency medical technicians (EMTs) with training in ALS. In particular, prehospital triage of trauma patients by EMTs plays an important role in prehospital management and transmitting patient information to the designated hospital. Rapid diagnosis, triage and early intervention are important for the prognosis of trauma patients.1 For abdominal trauma, physical examination cannot be fully trusted, and imaging devices are needed.2,3 For example, abdominal ultrasound is effective for diagnosing intraabdominal haemorrhage and measuring haemorrhage severity.4,5 Therefore, if an imaging device can be used at the prehospital stage, it will have a positive effect on the prognosis of the patient.6,7 Several previous studies have shown that doctors dispatched with ambulances
and emergency physicians in the emergency room can be successfully educated through a short course (e.g. a 1-day course) in focused assessment with sonography for trauma (FAST).8–10 However, few studies have been published regarding the accuracy and clinical efficacy of FAST performed by EMTs11 after a short training course, and studies about the transmission of FAST images are especially rare.11,12 The use of ultrasound at the prehospital stage is not routine, because there is not enough evidence to support its reliability unless doctors are dispatched with the ambulance to the scene. This is not common in Korea or
1 Department of Emergency Medicine, College of Medicine, Seoul National University, Korea 2 Department of Biomedical Engineering, College of Medicine, Seoul National University, Korea 3 Interdisciplinary Program of Bioengineering, Graduate School, Seoul National University, Korea 4 Department of Emergency Medicine, College of Medicine, Jeju National University, Korea
Corresponding author: Dr Hee Chan Kim, Department of Biomedical Engineering, Seoul National University Hospital, 101 Daehak-ro, Jongno-gu, Seoul 110-744, Korea. Email: [email protected]
Song et al. North America. In Europe, on the other hand, the use of ultrasound at the prehospital stage is gaining attention in emergency medicine. It is used to detect fluid retention in the abdominal and thoracic cavities, to help with the diagnosis of pregnancy, and to assist in cardiac, liver, kidney, small bowel, splenic and pancreatic diseases. Ultrasound also plays an important role in differentiating pulseless electrical activity in cardiac arrest patients.13 However, these studies were based on ultrasound performed by emergency physicians who were on board in the ambulances or helicopters. The purpose of the present study was to evaluate a newly developed real-time, prehospital image transmission system for FAST performed by EMTs. The objective was to find out whether the ultrasound images obtained by an EMT and transmitted from a remote ambulance would be helpful for emergency physicians to diagnose haemoperitoneum.
Methods We conducted a simulation experiment on prehospital ultrasound performed by EMTs. We investigated the diagnosis of haemoperitoneum made by emergency physicians based on the transmitted ultrasound video images.
Image transmission system The wireless, real-time ultrasound image transmission system comprised a commercial ultrasound system (Titan, SonoSite) with a convex array abdominal transducer (C60/5-2, SonoSite) and a notebook computer connected to a 3 G telecommunication network through a mobile phone (iPhone 4, Apple) for video data transmission (see Figure 1). Ultrasonography was performed on the pre-determined body location by EMTs who had already completed training. The analogue ultrasound images were converted using a frame grabber (Grand AV USB2.0 PRO, Grandtec) into digital images of 640x480 pixel resolution, which were captured by a notebook computer (Lifebook P1610, Fujitsu). Each video frame was then encoded for compression and transmitted in JPEG image file format to a remote server where a viewer program reconstructed the video clip in AVI file format. All functions were implemented by in-house developed software written in the C# programming language. Although wireless transmission techniques have improved, it is still not easy to eliminate unexpected network connection losses and to secure enough bandwidth to transmit video data in real-time through a network. To provide good quality images under these conditions, we developed an application program which controlled data passage at the transport layer of the communications protocol (TCP/IP). This program handled connection loss problems through two activities: periodic network connectivity checks using a verification message and automatic re-connection for rapid recovery from an unexpected network disconnection. The integrity of the transmitted data was also guaranteed by packet-based
451
Figure 1. A. the real-time transmission system for ultrasound images. B. the ultrasound scanner, the notebook PC and the Internet-connected smart phone.
data checking. Examining network status via the round trip time and the usage frequency of a data buffer provided information for adjusting the image compression ratio and frame rate, respectively, to cope with the problems caused by wireless network bandwidth fluctuations. First, the image compression rate was adjusted based on the current round trip time value with reference to a peak signal-to-noise ratio (PSNR) value of 30–50 dB. The PSNR value is a commonly used measure of the quality of a reconstructed image after lossy compression. Generally, video is compressed with a PSNR value of 30–50 dB, but for wireless transmission, a PSNR value of 20–25 dB is often allowed. However, there is no established standard of PSNR values for medical images. The compressed information was stored in a data buffer before transmission through the wireless network. Finally, the frame rate was controlled according to the usage frequency of the data buffer to prevent buffer overflow.
Evaluation of diagnostic performance Ultrasonography was performed on an anatomical phantom with all organs similar to human organs. First, we prepared two kinds of phantoms (shown in Figure 2), one with normal intraperitoneal cavity and the other with simulated haemoperitoneum. An EMT boarded the ambulance with one phantom which was selected according to a random number table. This EMT had participated in a similar study and received education about ultrasonography and FAST.14 Ultrasonography was performed in the ambulance in both moving and stopped conditions at distances of 0, 1, 2, 5, 10 and 15 km from the emergency department (ED). Each ultrasound image obtained in a certain condition and location was transmitted to the server in the ED in real-time. All video clips from the transmitted ultrasonogram were given to an emergency physician for the diagnosis
452
Journal of Telemedicine and Telecare 19(8)
Figure 2. Phantoms used for FAST. The left-hand phantom is the haemoperitoneum model and the right-hand phantom is the normal model.
of haemoperitoneum. In order to eliminate potential factors such as distance of transmission, the running status of the ambulance and the learning effect of the EMT performing the ultrasonography, we rearranged the sequence of the transmitted ultrasound images according to another random table generated by a statistics program. The researchers and emergency physicians participating in the experiment were blinded to the sequence of the ultrasound images and the presence of haemoperitoneum. The sensitivity, specificity and accuracy of the emergency physician’s diagnosis were calculated based on the transmitted ultrasound images and predictive values. Statistical analysis was performed using receiver operating characteristics (ROC) analysis by a standard package (STATA/MP version 11.0, StataCorp., College Station, TX). In addition, we compared the sensitivity, specificity and predictive values of an adjusted model, which took into account the distance of transmission, the running status of the ambulance and the qualification of the emergency physician (attending/resident). We compared the area under the curve (AUC) values between the two models.
Results The characteristics of the transmitted ultrasound images from the ambulance are shown in Table 1. Two out of 23 transmissions were excluded because they were too short to make a diagnosis. Twenty-one ultrasound video clips (10 abnormal cases, 11 normal cases) were used for the final analysis. In total, 13 emergency physicians (8 attending physicians, 5 emergency medicine residents) participated in the ultrasound image interpretation. The sensitivity and specificity were 90.0% (95% Confidence Interval, CI 83.5– 94.6) and 85.3% (95% CI 78.4–90.7), respectively, and the accuracy of detecting abnormal ultrasound results was 87.7% (95% CI 83.8–91.6) (Table 2). The sensitivity and specificity according to the emergency physician’s qualification are shown in Table 3.
Table 1. Characteristics of the transmitted ultrasound images. Distance, D, Ambulance Image from ED (km) status length (s) Finding
Interpretation order
0
Clinical applicability of real-time, prehospital image transmission for FAST (Focused Assessment with Sonography for Trauma)
Journal of Telemedicine and Telecare 19(8) 450–455 ! The Author(s) 2013 Reprints and permissions: sagepub.co.uk/journalsPermissions.nav DOI: 10.1177/1357633X13512068 jtt.sagepub.com
Kyoung Jun Song1,2, Sang Do Shin1, Ki Jeong Hong1, Kyoung Woo Cheon3, Ilhyoung Shin3, Sung-Wook Song2,4 and Hee Chan Kim2,3
Summary We evaluated a real-time, prehospital ultrasound image transmission system for use in focused assessment with sonography for trauma (FAST). The wireless, real-time ultrasound image transmission system comprised an ultrasound scanner with a convex abdominal transducer and a notebook computer connected to a 3 G wireless network for video data transmission. In our simulation experiment, ultrasonography was performed by emergency medical technicians (EMTs) on a human body phantom with simulated haemoperitoneum. Transmitted ultrasound video clips were randomly rearranged and presented to emergency physicians to make a diagnosis of haemoperitoneum. A total of 21 ultrasound video clips was used and 13 emergency physicians participated in the study. The sensitivity and specificity were 90.0% (95% Confidence Interval, CI, 83.5–94.6) and 85.3% (95% CI 78.4–90.7) respectively, and the accuracy of detecting abnormal ultrasound results was 87.7% (95% CI 83.8–91.6). Diagnosis of hemoperitonuem in trauma patients by an emergency physician based on the transmitted video images of FAST performed by an EMT is feasible, and has an accuracy of about 88%. Accepted: 21 September 2013
Introduction During the last ten years there has been a substantial improvement in the overall quality of the emergency medical service (EMS) in Korea. However, prehospital EMS in Korea provides only basic life support. The concept of prehospital emergency medicine is changing from basic life support which focuses on rescue and transport, to advanced life support (ALS) which emphasizes rescue, evaluation, treatment, triage and transport. For this new role, it is necessary to have ambulances with sophisticated equipment and emergency medical technicians (EMTs) with training in ALS. In particular, prehospital triage of trauma patients by EMTs plays an important role in prehospital management and transmitting patient information to the designated hospital. Rapid diagnosis, triage and early intervention are important for the prognosis of trauma patients.1 For abdominal trauma, physical examination cannot be fully trusted, and imaging devices are needed.2,3 For example, abdominal ultrasound is effective for diagnosing intraabdominal haemorrhage and measuring haemorrhage severity.4,5 Therefore, if an imaging device can be used at the prehospital stage, it will have a positive effect on the prognosis of the patient.6,7 Several previous studies have shown that doctors dispatched with ambulances
and emergency physicians in the emergency room can be successfully educated through a short course (e.g. a 1-day course) in focused assessment with sonography for trauma (FAST).8–10 However, few studies have been published regarding the accuracy and clinical efficacy of FAST performed by EMTs11 after a short training course, and studies about the transmission of FAST images are especially rare.11,12 The use of ultrasound at the prehospital stage is not routine, because there is not enough evidence to support its reliability unless doctors are dispatched with the ambulance to the scene. This is not common in Korea or
1 Department of Emergency Medicine, College of Medicine, Seoul National University, Korea 2 Department of Biomedical Engineering, College of Medicine, Seoul National University, Korea 3 Interdisciplinary Program of Bioengineering, Graduate School, Seoul National University, Korea 4 Department of Emergency Medicine, College of Medicine, Jeju National University, Korea
Corresponding author: Dr Hee Chan Kim, Department of Biomedical Engineering, Seoul National University Hospital, 101 Daehak-ro, Jongno-gu, Seoul 110-744, Korea. Email: [email protected]
Song et al. North America. In Europe, on the other hand, the use of ultrasound at the prehospital stage is gaining attention in emergency medicine. It is used to detect fluid retention in the abdominal and thoracic cavities, to help with the diagnosis of pregnancy, and to assist in cardiac, liver, kidney, small bowel, splenic and pancreatic diseases. Ultrasound also plays an important role in differentiating pulseless electrical activity in cardiac arrest patients.13 However, these studies were based on ultrasound performed by emergency physicians who were on board in the ambulances or helicopters. The purpose of the present study was to evaluate a newly developed real-time, prehospital image transmission system for FAST performed by EMTs. The objective was to find out whether the ultrasound images obtained by an EMT and transmitted from a remote ambulance would be helpful for emergency physicians to diagnose haemoperitoneum.
Methods We conducted a simulation experiment on prehospital ultrasound performed by EMTs. We investigated the diagnosis of haemoperitoneum made by emergency physicians based on the transmitted ultrasound video images.
Image transmission system The wireless, real-time ultrasound image transmission system comprised a commercial ultrasound system (Titan, SonoSite) with a convex array abdominal transducer (C60/5-2, SonoSite) and a notebook computer connected to a 3 G telecommunication network through a mobile phone (iPhone 4, Apple) for video data transmission (see Figure 1). Ultrasonography was performed on the pre-determined body location by EMTs who had already completed training. The analogue ultrasound images were converted using a frame grabber (Grand AV USB2.0 PRO, Grandtec) into digital images of 640x480 pixel resolution, which were captured by a notebook computer (Lifebook P1610, Fujitsu). Each video frame was then encoded for compression and transmitted in JPEG image file format to a remote server where a viewer program reconstructed the video clip in AVI file format. All functions were implemented by in-house developed software written in the C# programming language. Although wireless transmission techniques have improved, it is still not easy to eliminate unexpected network connection losses and to secure enough bandwidth to transmit video data in real-time through a network. To provide good quality images under these conditions, we developed an application program which controlled data passage at the transport layer of the communications protocol (TCP/IP). This program handled connection loss problems through two activities: periodic network connectivity checks using a verification message and automatic re-connection for rapid recovery from an unexpected network disconnection. The integrity of the transmitted data was also guaranteed by packet-based
451
Figure 1. A. the real-time transmission system for ultrasound images. B. the ultrasound scanner, the notebook PC and the Internet-connected smart phone.
data checking. Examining network status via the round trip time and the usage frequency of a data buffer provided information for adjusting the image compression ratio and frame rate, respectively, to cope with the problems caused by wireless network bandwidth fluctuations. First, the image compression rate was adjusted based on the current round trip time value with reference to a peak signal-to-noise ratio (PSNR) value of 30–50 dB. The PSNR value is a commonly used measure of the quality of a reconstructed image after lossy compression. Generally, video is compressed with a PSNR value of 30–50 dB, but for wireless transmission, a PSNR value of 20–25 dB is often allowed. However, there is no established standard of PSNR values for medical images. The compressed information was stored in a data buffer before transmission through the wireless network. Finally, the frame rate was controlled according to the usage frequency of the data buffer to prevent buffer overflow.
Evaluation of diagnostic performance Ultrasonography was performed on an anatomical phantom with all organs similar to human organs. First, we prepared two kinds of phantoms (shown in Figure 2), one with normal intraperitoneal cavity and the other with simulated haemoperitoneum. An EMT boarded the ambulance with one phantom which was selected according to a random number table. This EMT had participated in a similar study and received education about ultrasonography and FAST.14 Ultrasonography was performed in the ambulance in both moving and stopped conditions at distances of 0, 1, 2, 5, 10 and 15 km from the emergency department (ED). Each ultrasound image obtained in a certain condition and location was transmitted to the server in the ED in real-time. All video clips from the transmitted ultrasonogram were given to an emergency physician for the diagnosis
452
Journal of Telemedicine and Telecare 19(8)
Figure 2. Phantoms used for FAST. The left-hand phantom is the haemoperitoneum model and the right-hand phantom is the normal model.
of haemoperitoneum. In order to eliminate potential factors such as distance of transmission, the running status of the ambulance and the learning effect of the EMT performing the ultrasonography, we rearranged the sequence of the transmitted ultrasound images according to another random table generated by a statistics program. The researchers and emergency physicians participating in the experiment were blinded to the sequence of the ultrasound images and the presence of haemoperitoneum. The sensitivity, specificity and accuracy of the emergency physician’s diagnosis were calculated based on the transmitted ultrasound images and predictive values. Statistical analysis was performed using receiver operating characteristics (ROC) analysis by a standard package (STATA/MP version 11.0, StataCorp., College Station, TX). In addition, we compared the sensitivity, specificity and predictive values of an adjusted model, which took into account the distance of transmission, the running status of the ambulance and the qualification of the emergency physician (attending/resident). We compared the area under the curve (AUC) values between the two models.
Results The characteristics of the transmitted ultrasound images from the ambulance are shown in Table 1. Two out of 23 transmissions were excluded because they were too short to make a diagnosis. Twenty-one ultrasound video clips (10 abnormal cases, 11 normal cases) were used for the final analysis. In total, 13 emergency physicians (8 attending physicians, 5 emergency medicine residents) participated in the ultrasound image interpretation. The sensitivity and specificity were 90.0% (95% Confidence Interval, CI 83.5– 94.6) and 85.3% (95% CI 78.4–90.7), respectively, and the accuracy of detecting abnormal ultrasound results was 87.7% (95% CI 83.8–91.6) (Table 2). The sensitivity and specificity according to the emergency physician’s qualification are shown in Table 3.
Table 1. Characteristics of the transmitted ultrasound images. Distance, D, Ambulance Image from ED (km) status length (s) Finding
Interpretation order
0
