Keywords

1 Introduction

The central venous access (CVA) is an invasive medical procedure that allows accessing the big circulatory vessels (e.g., jugular, subclavian or femoral vein) to provide drugs, anesthesia, fluids or treatments to the patient in a critical condition [12]. It requires skilled specialists to successfully perform the technique and avoid any complications that may compromise the patients life [17, 20]. In neonates, the procedure is even more challenging because of the size, gaps between organs, tissues and their mechanical properties. It has been shown that skills related to this procedure are improved with the use of training simulators [4, 16, 18, 21], where virtual or real manikins are used to represent patient anatomy and vital functions in a controlled environment [8, 15, 27]. There are several CVA simulators for 5+ years old patients (CVC Insertion Simulator II [1], Gen II Ultrasound Central Line Training Model [7], Internal Jugular Central Line Ultrasound Manikin [5], VascularAccessChild Training System [22]) designed for jugular, femoral and subclavian vein access, while CVA simulators for neonates are only focused on umbilicus, femoral and peripheral venous access and intraosseous access (Simbaby [13], SimNewB [14], BabySIM [6], Newborn PEDI Simulator [9]). However, neonatal CVA simulators for jugular and subclavian veins are scarce, lack to be remedied since the cannulation of these vessels is common in medical practice [25, 26] and its realization requires greater skills obtained through practice [28].

An interactive simulation system for practicing the needle insertion in the access of the internal jugular vein in neonates is proposed. The system has a physical syringe coupled with inertial sensors and a virtual model of the neonatal patient displayed on a touch screen of a tablet device. During the simulation, the system generates a visual feedback for successful access to the internal jugular vein, or a failed attempt when the carotid artery is accessed. Both are closely spaced vessels. The response is based on a linear estimation of the intersection between the needle tip and the vessel, considering the insertion point on the patient skin, the angle chosen by the user and the position, orientation, and diameter of the target vessels. This paper is divided as follows: in Sect. 2 the medical procedure and the design of the proposed simulation system are described. In Sect. 3 the results of the user experience are presented. Finally, conclusions and future work are presented in Sect. 4.

2 3DUI Syringe Development

In real practice, an important and challenging step during CVA is the needle insertion using a syringe [17]. Anatomically, the access region for the internal jugular vein is bounded by the clavicle and the sternocleidomastoid muscle, as seen in the Fig. 1. A point within this region should be selected to insert the needle at an angle of 45 degrees relative to the coronal plane and in the direction of the ipsilateral nipple. Typically the vein is accessed less than 1.3 cm. A successful entry into the vessel is visually identified when venous blood returns into the syringe [11].

Fig. 1.
figure 1

Needle insertion

Full size image

2.1 System Architecture

Using the needle insertion procedure in jugular CVA as a reference, we define the 3DUI inputs and outputs so our solution represents life-like interactions with a real syringe. For the proposed system a physical model of the syringe was built, coupled with inertial sensors to estimate its orientation, interacting with a virtual model of a neonate displayed on a touchscreen of a mobile device (tablet). During the procedure, the user can freely handle the syringe to select an access point on the touchscreen where the neonate is displayed. When the tip of the syringe touches the screen, the system estimates if the needle is accessing the internal jugular vein or the carotid artery, performing an extension of the needle into the neonate, gradually, to a defined length, continuously evaluating the possible collision of the needle tip with the vessels represented by cylindrical geometric shapes. A description of the proposed system is presented in Fig. 2.

Over the skin of the neonate, a region access on the virtual model has been previously defined, as it is presented in Fig. 3. The spherical coordinates that compound this region are transformed into pixel coordinates to compare their position with the position chosen by the user with the syringe tip over the touchscreen. This transformation is done by using the visual model matrix, defined by the position and orientation of the camera, the features of the perspective projection and the display region on the screen [2]. When the system detects a collision between the tip of the syringe with any of the projected spheres, it can be determined the chosen position \(P_0\) on the skin of the neonate doing an inverse transformation. If a collision with multiple spheres exists, the sphere closer to the camera is chosen.

Fig. 2.
figure 2

System diagram

Full size image
Fig. 3.
figure 3

Region access over the skin of neonate

Full size image

With the information taken from the accelerometer (G) a magnetometer (B) coupled to the body of the syringe, an orthogonal basis estimation is stated \(\left\langle i_s, j_s,k_s \right\rangle \), describing the orientation of the inertial sensors regarding to a general orthogonal reference system \(\left\langle U, V, W \right\rangle \) defined by the magnetic and gravitational fields [19]. For this, taken as reference the normalized measure of the accelerometer \(W=G/|G|\), without the effect of external accelerations besides gravity, we have

$$\begin{aligned} V=W x \left( \frac{B}{|B|} \right) \end{aligned}$$
(1)

and

$$\begin{aligned} U=V x W \end{aligned}$$
(2)

then, the orthogonal basis of the inertial sensors is defined by

$$\begin{aligned} i_s&= \left( U_x,V_x,W_x \right) \end{aligned}$$
(3)
$$\begin{aligned} j_s&= \left( U_y,V_y,W_y \right) \end{aligned}$$
(4)
$$\begin{aligned} k_s&= \left( U_z,V_z,W_z \right) \end{aligned}$$
(5)

Therefore, the direction of the syringe is determined by the alignment of its central axis with \(i_s\), \(j_s\), \(k_s\). With the position vectors \(P_0\) and orientation vectors R of the tip of the syringe, the P vector representing the extension of the needle is defined, such as

$$\begin{aligned} P=P_0+K R \end{aligned}$$
(6)

where the scalar \(K>0\) defines the magnitude of the vector. The value of K increases gradually in each iteration until a maximum value, a process where the collision between P and blood vessels is evaluated using the method described by Sunkel [24]. The method consists in projecting P to the reference vessel system, evaluating the resulting coordinates within the confines of the cylindrical body of the vessel. This projection is done by \(P=P-C_0\), being \(C_0\) the origin position of the vessel reference system and using the scalar product, such as

$$\begin{aligned} P'=(P \cdot i_v,P \cdot j_v,P\cdot k_v) \end{aligned}$$
(7)

where \(i_v\), \(j_v\) y \(k_v\) are the unit vectors that define the reference system of the vessel. If we assumed that central axis of the cylinder is \(i_v\), the component x of \(P' (P_x')\) is evaluated between the length of the cylinder (L), while the components \(P_y'\) and \(P_z'\) are compared with the vessel radius (radius), as

$$\begin{aligned} \text{ if } \left( min(L) \le P_x' \le max(L) \text{ and } \sqrt{P_y'^{2}+P_z'^{2} }\le radius \right) \text{ then } \text{ collision } \end{aligned}$$
(8)

If the collision exists, a graphical response is produced indicating the access to the jugular vein or the carotid artery. In Fig. 4 a graphical description of the actors in the collision detection is presented.

Fig. 4.
figure 4

Collision detection process between needle and jugular vein

Full size image

The system requires an initial calibration of the measures taken by the magnetometer and also to align the framework of sensors with the reference frame of the virtual model. The calibration of the magnetometer consists in centering the data regarding the origin, finding maximum and minimum values of each component and normalize them, as proposed by [29]. On the other hand, to align the reference systems of the syringe and mobile device, they must be placed on a flat surface by making that the z-axis of the two coordinate systems points to the normal direction of the surface. Then, the central axis of the syringe is aligned with the x-axis of the virtual model that is parallel to one side of the mobile device.

3 Results

The virtual environment is composed by a 3D model considering the basic anatomical features described in anatomy books of a term neonate. This model is oriented in the Trendelenburg position with \(15^o\) of inclination which is the recommended for the jugular central venous access procedure. The software was built with Android Studio [10] using libgdx [30]. For the implementation of the syringe, it was used an Arduino Mini Pro [3], a 9-DOF inertial measurement unit MPU-9150 [23] which encapsulates an accelerometer, a gyroscope and a magnetometer, a Bluetooth interface for wireless transmission between sensors and the mobile device. This application was implemented on a Samsung Galaxy Note Tablet 8.0 as presented in Fig. 5.

The proposed system includes a user interface to manage different options in the virtual environment. The user has the possibility to adjust the camera in order to change the perspective view of the model with some hand gestures that enable to rotate the model, to approach or to take a distance from it. During the procedure, the user can freely manipulate the syringe to select an access point from the images of the neonate. There is a function that blocks the movement of the camera to help to find the point for performing the procedure.

To evaluate the usefulness of the system, user experience tests were performed. The test consisted of manipulating the syringe to access the internal jugular vein. The test was performed with 12 users. Each of the participants freely practiced with the system for 5 min, recognizing the components and method of use. At the end of this practice, users responded to a set of questions, with skill levels, regarding their perception of interaction and visual feedback during the procedure.

Fig. 5.
figure 5

Implemented prototype

Full size image

3.1 User Experience

After the tests users stated:

  • The visual representation of the neonate is realistic, however, it is important a further enhancement of anatomical landmarks such as the collarbone and nipple, which are key benchmarks for the procedure.

  • The syringe allows adequate control of the position and orientation at the time of the puncture.

  • Although the system allows the viewer to manipulate the virtual environment by changing the position of observation of the neonate, it is sufficient that the system allows a small lateral movements around it, as well as zoom in and out of the region of interest.

Within additional observations made by the users are: allowing a three-dimensional view of the virtual environment, for example by defining different views of the same scene on the screen to better identify the position of the observer with respect to the neonate; and to make known the bevel of the needle tip, since its orientation is important for access.

4 Conclusions and Discussion

An interactive simulation system for training puncture in the access of the internal jugular vein was developed. The system is an innovative tool that can complement traditional learning based on the information presented in books and videos, allowing the user to interact with the virtual anatomy and performing the puncture in different manners.

The system can be extended to simulate vital signs, breathing movements and sounds in the environment, under normal conditions and possible complications.

Force feedback is the next step. Using a vibrotactile actuator inside the syringe would be ideal for such applications to represent the rupture of the skin and vessels. This system can be complemented to perform the complete procedure step by step.