1. Introduction
Blending Virtual Reality (VR) and Augmented Reality (AR), Mixed Reality (MR) technology is gaining growing attention in education, manufacturing, and healthcare industries. While VR technology deals with creating a digital simulation of real-world environments and completely immersive experiences in the virtual environment, AR technology overlays digital content in the real-world environment. MR technology, also known as hybrid or extended reality, combines VR and AR technologies with added features. It allows the interaction between virtual objects and physical objects. With the advancement of artificially intelligent, vision, and cloud computing, MR technology is expected to continue to enhance human–computer interactions [
1].
Recently, MR technology has been receiving more and more attention in the healthcare field, albeit not yet actively applied in medical practice and applications [
2]. In the medical field, MR technology can be effectively applied to train surgeons and medical students, such as training the implementation of medical implants and teaching anatomy classes. Moreover, the advancement of haptic technology that is capable of providing realistic tactile and kinesthetic feedback to the user significantly contributes to the development of medical training simulators. Medical training simulators based on haptic technology have become an important training alternative to achieve shorter learning curves and transfer skills for real procedures and operations. Escobar-Castillejos et al. reviewed medical simulators based on haptic devices for stitching, dental palpation procedures, endoscopy, laparoscopy, and orthopedics [
3].
One of the important potential applications of MR technology includes medical palpation training. In a previous study, an augmented reality simulation method was applied for training femoral palpation [
4,
5]. The method of giving haptic feedback is important for palpation training for diagnosing the condition of a virtual patient [
6]. In practical robotic-based telesurgery, a technique has been developed that allows the surgeon to feel the fingertip contact deformation as a vibration for ensuring the stability of the teleoperator [
7]. Although vibration feedback can be a convenient haptic sensory transmission method, pneumatic-actuated haptic feedback in VR/AR-based medical simulators can deliver a sensation similar to a real human pulse [
8,
9,
10,
11]. The pneumatic-actuated haptic feedback method, which is actively controlled by a pneumatic pump, is a convenient method of delivering haptic cues but is not suitable for applications that require precise palpation feedback, such as radial artery pulse palpation (RAPP) training.
RAPP holds significant potential for such an application. In particular, diagnosis of illness based on the palpation of the radial artery is important in traditional and modern medicine. RAPP on the wrist is the primary diagnosis method for Oriental medicine [
12], and it is one of the pragmatic diagnosis methods for predicting the risks of cardiovascular diseases in Western medicine. This is mainly attributed to the benefits of RAPP, which provides hemodynamic features (such as arterial stiffness, blood pressure, and cardiac output), ease of access to the arteries, and comfortable measurement conditions [
13,
14,
15]. In particular, by using RAPP, Oriental medical experts categorize the radial artery pulse waveforms into 28 classes to diagnose the causes of diseases [
16]. To classify subtle changes in the pulse waveforms, they feel various temporal and spatial features such as pulse amplitude, width, length, depth, rate, and shape only using their fingertips. Therefore, taking a pulse is a dexterous manipulation skill; hence, repetitive training is necessary to be an expert.
In addition to the medical implications, the quality of the RAPP serves as a diagnostic test in tactical combat casualty care (TCCC), as it can be used as an important surrogate marker of hemorrhagic shock. Naylor et al. report that the tactical combat casualty care guidelines recommend the use of the radial pulse strength to guide the administration of blood products or intravenous fluids when equipment for blood pressure monitoring is not available [
17]. They performed a correlation study between the radial pulse strength and the systolic blood pressure (SBP) readings using the dataset from the Department of Defense Trauma Registry to analyze the efficacy of the radial pulses for casualty assessment. Other studies also studied the radial pulse in relationship to systolic blood pressure and trauma guidelines [
18,
19]. The studies and TCCC guidelines suggest that the quality of RAPP (such as strong, weak, and absent radial pulse) in conjunction with SBD is used to assess combat casualty and guide resuscitation protocols. Wightman points out that palpation of radial pulses is highly subjective and inconsistent between examiners [
20]. Thus, pulse simulators that can produce controlled pulses consistently and repeatedly would be useful for training medical examiners on the battlefield. They would allow a scientific study between the radial pulses and SBPs to establish statistical correlations and quantify the subjective assessments of pulse strength.
Currently, a few radial pulse simulators are available for medical training. They offer their own advantages and disadvantages in generating sophisticated blood pressure waveforms. ViVitro Labs, Inc., developed an endovascular simulator that can generate pulsatile flow and blood pressure waveforms similar to humans [
21]. Lee et al. developed a cardiovascular simulator for studying the depth, rate, shape, and strength of radial pulses [
22]. Tellyes Scientific Inc. developed a pulse-training simulator (Victor Pulse) that can adjust the characteristic parameters of the pulse wave by manipulating the opening time of the hydraulic valve and the hydraulic pressure intensity [
23]. Although these simulators create various blood pressure waveforms, they circulate fluid using pumps. Thus, they present maintenance issues that include fluid leakage. Moreover, the repeatability and accuracy of the generated pulse waveforms can be compromised by unwanted vortexes originating from the reflection of pressure waves. In general, it is challenging to produce age-dependent pulse waveforms with precise notches in the pulse pressure waveforms due to the bifurcation of the arteries. As an alternative to fluid-based simulators, Yang et al. developed simulators based on pneumatic pressure produced by a cam mechanism [
24]. Because the cam profiles are determined based on human pulse pressure waveform data obtained from a tonometric measurement method, the pneumatic simulators can precisely reproduce actual radial pulse pressure waveforms. Despite this strength, the primary limitation of the pneumatic simulator is that it only produces a single pulse waveform defined in the cam profile. Thus, a new cam needs to be constructed for a different pulse waveform, so a large number of cams would need to be fabricated and replaced to generate a range of age-specific pulse waveforms. In other words, the cam simulator can only discretely generate a pulse waveform depending on the specific cam used in the system.
In an effort to develop a new portable pulse simulator for medical palpation training, this study intends to develop a pneumatic system based on a cam mechanism that is capable of generating a wide range of age-related radial pulse waveforms. For a simple yet versatile system design, it proposes to develop a “universal” cam, which is a single cam that encompasses a range of pulse waveform profiles. Unlike the “discrete” cam system, the universal cam allows the system to produce a range of pulses continuously without replacing cams. To design the cam, this study utilizes the mean pulse pressure data obtained from actual tonometric measurements of ages 15, 35, 65, and 85 years old, and it then interpolates these mean data to determine the profile of the proposed cam that varies continuously, representing pulse waveforms for all age groups. Following the design and fabrication of the cam, a prototype simulator was built. The performance of the prototype was then experimentally evaluated, and the results were compared to those reproduced in in vivo pulse pressure waveforms.
In the next section, the design process of the proposed cam is introduced using the average pulse pressure waveform data of 15, 35, 65, and 85 years old obtained from tonometric measurement. It then describes the development process of the pulse pressure waveform simulator equipped with the manufactured cam. Finally, the comparison results of age-specific artificial pulse pressure waveforms generated by the developed simulator with human pulse pressure waveforms are described.
3. Results and Discussion
Using the prototype pulse simulator, this study performed a series of tests and produced pulse pressure waveforms over a wide range of age groups. The performance of the simulator was then evaluated by assessing the age-specific pulse waveforms generated by it. Human radial pulse pressure waveforms are formed by complex physiological mechanisms of the cardiovascular system, and aging has a significant effect on the magnitude and shapes of pulse waveforms. Thus, the performance evaluation focuses on if the pulse simulator is capable of generating age-dependent pulse waveforms that can capture the physiological characteristics of pulse waveforms associated with aging. Moreover, the accuracy of pulse pressure waveforms is quantitively analyzed using mathematically obtained age-related pulse waveforms based on in vivo data.
Figure 4 shows the pulse pressure waveforms generated by the pulse simulator from 15 to 85 years of age with 10-year increments by changing the position of the cam follower assembly along the 3D cam. Typically, the increase in arterial stiffness with aging results in the increment of the amplitude and pulse wave velocity of the reflected wave [
25]. This leads to an increase in the magnitude of pulse waveforms and a decrease in the interval between the first and second peaks of the waveforms. These trends are observed in the experimental results. As shown in the graphs in
Figure 4, the peak pressure values increase as age increases. Moreover, the distance between the first peak and other smaller peaks decreases as the age increases from 15 to 85. In fact, the results show that the magnitude of the pulse pressure, the difference between the peak pressure (systolic pressure) value and the minimum (diastolic pressure) value, is increased from about 30 mmHg to 75 mmHg with the increase in age. Note that the diastolic pressure was set to 80 mmHg in this study. Thus, the results indicate that the pulse pressure waveforms produced by the 3D cam-based pulse pressure simulator can represent realistic age-dependent pulse waveforms that are physiologically consistent with human radial artery pulse pressure (RAPP) waveforms.
To further evaluate the (artificial) pulse pressure waveforms (produced by 3D CAM), they are compared with (human) approximate pulse pressure waveforms. To quantitatively analyze the pulse waveforms, two indexes are employed. First, the L
2 error, the best method to evaluate the error between two consecutive data through least-squares fitting, between the pulse pressure waveforms generated by the developed simulator and the approximate human waveforms, is calculated. In addition to the error analysis, the radial Augmentation Index (AI) value is analyzed. The AI is clinically used as a measure of vascular stiffness with age, and it is expressed by the following equation. Since the small error between AI values means that the pulse pressure waveforms are generated under conditions of clinically similar vascular stiffness, it is used for clinical similarity analysis between pulse pressure waveforms.
Collecting pulse waveforms using human subjects of all ages is a daunting task. For the current study, it was not feasible as it requires tremendous resources beyond the scope of this study. Thus, as an alternative to full-scale clinical testing, it uses approximate pulse pressure waveforms of 15–85 years of age, obtained by a mathematical model based on in vivo data measured on about 1000 subjects [
26], to analyze the pulse waveforms produced by the simulator. The in vivo data were measured by a robotic tonometry system at the Korea Institute of Oriental Medicine in a separate study [
26].
Figure 5 shows the approximate pulse waveforms for ages 15 through 85 with an increment of 10 years.
Figure 6 compares the pulse pressure waveforms reproduced by the developed simulator and approximate pulse pressure waveforms after normalizing them by age. Overall, they match well, and, particularly, the initial slope and the first peak coincide well. The L
2 errors and the AI values of each case are calculated, and the results are summarized in
Table 1.
In
Table 1, it can be confirmed that the L2 error is less than 9.6% at all ages, and the AI error is at most 9.4% or less at all ages. These errors fall within the acceptable level of accuracy in generating actual radial pulse pressures. As a matter of fact, in vivo data (systolic pulse pressure) obtained by a tonometric method using hundreds of subjects by age show an average standard deviation of 11.7% [
26]. Thus, the error analysis results indicate that the developed simulator is capable of generating realistic human pulse waveforms. Moreover, the low AI error results indicate that the pulse pressure waveforms generated by the 3D cam-based pulse pressure simulator can be used to estimate AI values accurately when clinically judging arteriosclerosis.
4. Conclusions
In this study, a pulse simulator system with a universal or 3D cam has been developed. The cam offers a cost-effective and simple (yet versatile) way to produce a wide range of radial pulse pressure (RAPP) waveforms. Unlike conventional thin disk-type cams, the newly devised 3D cam allows the pulse simulator to generate a specific pulse in a range of pulse waveforms without having to make multiple cams and replace them. In designing the 3D cam, this study used in vivo pulse waveform data for four age groups (15, 35, 65, and 85 years old) obtained from tonometric measurements. After a coordinate transformation of the average pulse waveforms, the cam profile (a two-dimensional cam shape) was created for each of the four age groups. These four cam profiles were then used to create the profile of the 3D cam by the linear regression method. A 3D cam was fabricated using precision machining technology, and it was incorporated into the prototype pulse simulator. The prototype simulator utilizes stepping motors to control the rotational speed of the cam and the position of the cam follower assembly at a specific location of the 3D cam in order to produce the desired pulse waveform. The prototype simulator generated age-dependent pulse waveforms from 15 years old to 85 years old with an increment of 10 years. These waveforms were compared with approximate human RAPP waveforms to evaluate the performance of the prototype system. The results show that the simulator-generated pulse waveforms capture the age-related physiological characteristics of the human RAPP well. Furthermore, the augmentation index (AI) and the L2 errors between the pulse waveforms by the simulator and those of approximate human data were analyzed. The results show that the errors are less than 10% for all ages, indicating that the 3D cam-based pulse simulator is capable of generating virtually all age-specific radial pulse pressure waveforms that are equivalent to human RAPPs.
The simulator is also capable of producing pulse waveforms consistently and repeatedly and altering pulse parameters conveniently to reproduce variant pule waveforms. This is a significant benefit of the current simulator, as existing pulse simulators cannot guarantee repeatability or consistent accuracy. For example, pulse simulators that use working fluids are prone to inconsistent measurements caused by unwanted vortexes in the fluid due to the reflection of pulse waves. Moreover, commercially available pulse simulators are significantly more complex and bulky as compared to the pulse simulator developed in this study. Hence, this pulse simulator would be an excellent system for health professionals to train RAPP palpations because it can reproduce known pulses consistently. The pulse simulator could be integrated into a medical Mixed Reality (MR) system as well. Great portability due to its compact design and ease of operation are additional benefits of the system. In order to complete the medical palpation system, we are planning research on the development of an artificial wrist, including an artificial blood vessel and skin that transmits the generated pulse pressure waveform. If an artificial wrist is installed, this simulator is expected to be used for wearable pressure sensor evaluation, transmission function or relationship research between radial and central pulse pressure, and Oriental medicine pulse diagnosis training.