Robot and ultrasound assisted needle insertion to the transverse carpal ligament

Abstract

Background:

A potential alternative treatment to surgery for carpal tunnel syndrome is to inject enzymes into the transverse carpal ligament to decrease its stiffness and alleviate pressure off the median nerve. An accurate injection is needed for delivery to achieve the effects of tissue degradation. The purposes of this study were to 1) determine injection sites using 3D reconstructed anatomy, and 2) insert the needle to the middle of the transverse carpal ligament thickness in situ.

Methods:

Six fresh-frozen cadaveric hands were used in this study. Five injection sites were determined in the sagittal plane along the center of the transverse carpal ligament thickness ulnar to the thenar muscle attachment using 3D ultrasonographic reconstruction. Each injection was delivered by rigidly fixing a 27-gauge needle to a six degrees of freedom robot arm programmed to insert the needle tip to the intended target. Ultrasound images were taken of the needle after insertion to measure accuracy and precision of the needle placement.

Findings:

The needle tip was successfully delivered to the middle region of the transverse carpal ligament thickness and visualized using ultrasound imaging. The accuracy and precision of the needle insertion were 0.83 and 0.31 mm, respectively.

Interpretation:

Methodology was established for robot-assisted needle insertion to the transverse carpal ligament using 3D ultrasonographic reconstructed anatomy. This methodology can be used in the future to deliver enzymatic injections to the transverse carpal ligament as a potential treatment for carpal tunnel syndrome.

1. Introduction

The transverse carpal ligament (TCL) is a collagenous band of tissue in the hand that forms the volar boundary of the carpal tunnel. Its biomechanical functions include stabilizing the carpal bones and serving as an anchor for the thenar and hypothenar muscles. The TCL is critically related to carpal tunnel syndrome due to its volar restriction of the median nerve. For example, increased thickness and stiffness in the radial region of the TCL have been associated with carpal tunnel syndrome (Ferrari et al., 1997; Marquardt et al., 2016). This could be due to a thicker TCL resulting in less space for the carpal tunnel contents and a stiffer TCL not being able to palmarly bow to allow adequate space for the median nerve. The most commonly performed surgical procedure for carpal tunnel syndrome is to transect the TCL to increase carpal tunnel space. However, surgery can disrupt the biomechanical functions of the TCL and result in complications such as pillar pain and hand weakness (Karl et al., 2016).

A possible alternative treatment to surgery is to decrease TCL stiffness through biochemical means. Enzymatic injection has previously been used to regulate tissue stiffness in patients with Dupuytren’s Contracture (Abdelrahman et al., 2020; Badalamente and Hurst, 2018; Smeraglia et al., 2016), a condition characterized by the thickening of the tissue causing irregular bending of the fingers. Similarly, injection to the TCL could be used to increase TCL elasticity, alleviating pressure off the median nerve. Due to the small size of the TCL, which has an average thickness ranging from 1.3 to 3.0 mm (Pacek et al., 2010), accuracy of needle placement is needed to ensure proper delivery of enzymes and prevent damage to surrounding structures, such as the median nerve. Accuracy and precision of needle placement has been shown to increase using robot-assisted injection in percutaneous therapy (Abolhassani et al., 2007), epicardial injection (Ota et al., 2008), and brain surgery (Kwoh et al., 1988).

The purposes of this study were to 1) determine injection sites on the TCL using 3D reconstructed anatomy, and 2) insert the needle to the middle of the TCL thickness in situ. To ensure the needle tip is delivered to the middle of the TCL thickness, the anatomy can be reconstructed using 3D ultrasonography to determine the site of injection. 3D ultrasonographic reconstruction allows for the visualization of anatomical structures and the quantification of morphological parameters of these structures. By reconstructing carpal tunnel anatomy, we can precisely determine injection sites on the TCL while avoiding other nearby anatomy, such as the thenar and hypothenar muscles. Furthermore, this would allow needle placement to be based on patient-specific anatomy, thus providing more individualized treatment.

2. Methods

2.1. Specimen preparation

Six freshly frozen male cadaveric hands (58 ± 11 years old) were used in this study. Each specimen was secured in a thermoplastic splint so that the wrist was positioned at 30-degrees extension and the thumb was radially abducted.

2.2. Ultrasound scanning procedures

An 18 L6 linear array transducer (SuperSonic MACH 30, Hologic Inc., Marlborough, MA, USA) was rigidly fixed to the end effector of a six-degree of freedom robot arm (Denso Corp., Kariya, Aichi, Japan) using a 3D printed probe holder (Fig. 1). A MicroScribe digitizer (Microscribe GX2, Immersion Corp., San Jose, CA) was used to digitize three points on the ultrasound probe and three points on the table that defined the probe coordinate system and table coordinate system, respectively. Robot movement was programmed with respect to the probe coordinate system using the following equation:

$$\left[{\text{T}}_{\text{RE}}\right]=\left[{\text{T}}_{\text{RT}}\right]{\left[{\text{T}}_{\text{MT}}\right]}^{-1}\left[ {\text{T}}_{\text{MP}}\right]$$ (1)

where [T_RE], [T_RT], [T_MT], and [T_MP] are the transformation matrices between the robot and the end effector, between the robot and the table, between the MicroScribe digitizer and the table, and between the MicroScribe digitizer and the ultrasound probe, respectively. [T_MT] and [T_MP] were defined from the points digitized on the table and probe by the MicroScribe digitizer.

Fig. 1.

Experimental setup showing the ultrasound probe was rigidly fixed to the end effector of the six degrees of freedom robot arm and positioned above the distal end of the carpal tunnel of a cadaveric hand.

A thin layer of ultrasound gel was applied to the volar wrist surface of the specimen for ultrasound imaging. Using the robot arm to position the ultrasound probe, the probe was approximately aligned at the distal end of the tunnel so that the ridge of trapezium and hook of hamate could be visualized. Using a custom LabVIEW program that moved the robot according to the calculated [T_RE] from Eq. (1), the ultrasound probe was translated proximally in 0.2 mm steps for 25 mm. After each step, an ultrasound image was captured, and the corresponding robot end effector position was recorded.

2.3. Anatomical coordinate system establishment

The surfaces of the hamate, pisiform, trapezium, and scaphoid were extracted from the ultrasound images using a custom LabVIEW program that involved selecting a region of interest and thresholding the image based on its grayscale values. The extracted coordinates, (u,v), in the 2D image coordinate system were transformed to the 3D coordinates, (x,y, z), in the robot coordinate system using the following transformation equation:

$${(\text{x},\text{y},\text{z},1)}^{\text{T}}=\left[{\text{T}}_{\text{RE}}\right]\left[{\text{T}}_{\text{EP}}\right]\left[{\text{T}}_{\text{PI}}\right]{(\text{u},\text{v},0,1)}^{\text{T}}$$ (2)

where [T_EP] and [T_PI] are transformation matrices from the end effector to probe coordinate system, and the probe to image coordinate system, respectively. With each bone reconstructed as a 3D point cloud in the robot coordinate system, the most volar point of each bone was determined and used to represent the four bony attachment points for the TCL. An anatomical coordinate system was established with its origin in the center of the four bony attachment points (Fig. 2). The x-axis was defined as the vector from the trapezium to scaphoid attachment points, the z-axis was determined by crossing the x-axis with the vector from the trapezium to pisiform attachment points, and the y-axis was determined by crossing the z- and x-axes.

Fig. 2.

Anatomical coordinate system using the four bony attachment points.

2.4. Injection site determination

From each ultrasound image where the attachment of the thenar muscle to the TCL was present, the ulnar-most point of the thenar muscle was manually selected and transformed from its (u,v) image coordinate to a 3D coordinate, (x,y,z), in the anatomical coordinate system using the following transformation equation:

$${(\text{x},\text{y},\text{z},1)}^{\text{T}}=\left[{\text{T}}_{\text{AR}}\right]\left[{\text{T}}_{\text{RE}}\right]\left[{\text{T}}_{\text{EP}}\right]\left[{\text{T}}_{\text{PI}}\right]{(\text{u},\text{v},0,1)}^{\text{T}}$$ (3)

where [T_AR] is the transformation matrix from the anatomical coordinate system to the robot coordinate system. The collected points were then fit to a 2D polynomial in the xy-plane of the anatomical coordinate system, and the most ulnar point of the thenar muscle was defined as the point on the polynomial with the maximum y-value.

One ultrasound image in the sagittal (xz) plane of the anatomical coordinate system was collected for determination of five injection sites in the TCL. The footprint of the ultrasound probe was positioned parallel to the x-axis of the anatomical coordinate system and its center was positioned at zero in the x-direction, 1 mm ulnar to the most ulnar point of the thenar muscle in the y-direction, and at 15 mm in the z-direction of the anatomical coordinate system. This placement was chosen to avoid interfering with the thenar muscle during needle insertion. One ultrasound image was captured in this position, and the robot recorded the probe position.

From the one ultrasound image, the volar and dorsal boundaries of the TCL were manually traced and each fit to a 2D-polynomial. The middle line of the TCL thickness was estimated as the average of the two polynomials and transformed into the anatomical coordinate system using Eq. (3). Five injection sites were determined along the this middle line at x = −6, −3, 0, +3, +6 mm in the anatomical coordinate system.

2.5. Robot-assisted needle insertion

A custom-made jig was used to rigidly fix a 27-gauge needle to the end effector of the robot. A needle coordinate system was established to allow for robot motion with respect to the tip of the needle. The x-axis was defined as the vector from the center of the base of the needle to the center of the needle tip, and the origin was placed at the needle tip. The y- and z-axes were arbitrary.

For each injection, the robot was programmed to insert the needle into the TCL until the tip reached the targeted injection site. The needle was positioned so that the needle tip was 50 mm radial and 50 mm volar (y = z = 50) to the targeted injection site and the x-axis of the needle coordinate system was at a 45-degree angle with respect to the xz-plane of the anatomical coordinate system. The robot was programmed to translate the needle along its local x-axis to the targeted injection site using the following equation:

$$\left[{\text{T}}_{\text{RE}}\right]=\left[{\text{T}}_{\text{RA}}\right]\left[{\text{T}}_{\text{AA’}}\right]\left[{\text{T}}_{\text{A’N}}\right]\left[{\text{T}}_{\text{NE}}\right]$$ (4)

where [T_RE], [T_RA], [T_AA’], [T_A’N], and [T_NE] are the transformation matrices between the robot coordinate system and the end effector coordinate system, between the robot coordinate system and the anatomical coordinate system, between the anatomical coordinate system and the needle path, between the needle path and the needle coordinate system, and between the needle coordinate system and the end effector coordinate system, respectively.

2.6. Accuracy verification

Ultrasound imaging was used to visualize the needle inserted in the specimen, and an optical tracking system (Polaris VEGA, Northern Digital Inc., Ontario, Canada) was used to record the probe position. Reflective markers able to be detected by the system were rigidly fixed to the ultrasound probe. Three points on the probe that were previously established to calculate the probe coordinate system were digitized using a digitizer able to be detected by the system and their positions in the optical coordinate system were recorded. The coordinates of the rigidly fixed reflective markers and three points on the probe were used to calculate the transformation matrix between the probe coordinate system and optical coordinate system. The probe was then positioned on the specimen such that the needle tip could be visualized, and an ultrasound image was captured (Fig. 3). With the optical tracking camera in the same position, the table was also digitized to determine the transformation matrix between the table coordinate system and the optical coordinate system. The needle tip was converted from the image coordinate system to the anatomical coordinate system using the following equation:

$${\left\{\text{x},\text{y},\text{z},1\right\}}^{\text{T}}=\left[{\text{T}}_{\text{AR}}\right]\left[{\text{T}}_{\text{RT}}\right]{\left[{\text{T}}_{\text{OT}}\right]}^{-1}\left[{\text{T}}_{\text{OP}}\right]\left[{\text{T}}_{\text{PI}}\right]{\left\{\text{u},\text{v},0,1\right\}}^{\text{T}}$$ (5)

where [T_AR], [T_RT], [T_OT], [T_OP], and [T_PI] are the transformation matrices between the anatomical coordinate system and the robot coordinate system, between the robot coordinate system and the table coordinate system, between the optical coordinate system and the table coordinate system, between the optical coordinate system and the probe coordinate system, and between the probe coordinate system and the image coordinate system, respectively.

Fig. 3.

Accuracy of needle tip using ultrasound and optical systems.

3. Results

Fig. 4 shows an example reconstruction of volar aspects of trapezium, hook of hamate, scaphoid, and pisiform, identified line on the TCL for needle insertion, and targeted injection points. The robot assisted needle insertion into the TCL was successfully delivered to the middle region of the TCL thickness as visualized in Fig. 5. Among the 30 robot assisted injections to the TCL across six specimens, the accuracies in x-, y-, and z-axis were 0.37, 0.44, and 0.47 mm, respectively, and the precisions in x-, y-, and z-axis were 0.30, 0.28, and 0.27 mm, respectively. The resultant 3D accuracy and precision of the 30 injections were 0.83 mm and 0.31 mm, respectively.

Fig. 4.

Ultrasound image of needle inserted into TCL. The volar and dorsal boundaries of the TCL are outlined in yellow. The needle tip is marked by the white triangle.

Fig. 5.

Reconstructed volar aspects of trapezium, hamate, pisiform, and scaphoid. The center of the TCL thickness was calculated 1 mm ulnar to the ulnar edge of the thenar muscle attachment and fit to a 2D curve. The five targeted injection points are shown on the line.

4. Discussion

Carpal tunnel release is the current standard surgical procedure for carpal tunnel syndrome and consists of transecting the TCL to alleviate pressure off the median nerve. However, TCL transection can interfere with the thenar muscle attachment, disrupting the TCL’s biomechanical function as an anchor for the muscle. A potential alternative treatment is to biochemically alter TCL structure through enzymatic injection. In order to inject collagenase to the TCL, accurate placement of the needle to the center of the TCL thickness is crucial for protection of surrounding structures. Improper needle placement during corticosteroid injection into the carpal tunnel has been shown to cause median nerve damage and paresthesias (Cunnington et al., 2010; Marshall et al., 2007), and even greater accuracy would be needed for injection to the TCL, which is a smaller anatomical target. Also, inserting the needle too deep could cause collagenase to enter the carpal tunnel and implicate the carpal tunnel contents. As a first step in exploring the potential of collagenase injection to the TCL, we established a methodology for accurate and precise needle insertion that can be used for delivery of collagenase to the TCL in situ using cadaveric hands.

Thirty robot-assisted needle insertions to the TCL were performed in situ. 3D ultrasonographic reconstruction was used to determine injection sites along the center of the TCL thickness, and robot-assistance allowed for the positioning of the needle tip to the intended target with sub-millimeter accuracy and precision. By using subject-specific reconstructed anatomy, this methodology allows for individualized needle insertion to the TCL, as anatomical variance can be accounted for.

For determination of the injection sites, the ulnar edge of the thenar muscle in the sagittal plane was reconstructed. The targeted injection sites were placed in this plane to avoid the needle interfering with the thenar muscle attachment on the TCL. The structural connection between the thenar muscle and ligament is critical for maintaining normal muscle contractions and hand functions. Therefore, by avoiding the muscle attachment during needle insertion, disruption to hand function can be minimized.

The accuracy of the needle tip to the target position is 0.83 mm with a precision of 0.31 mm which included the deviation of x, y, z components. In practice, the x and y components are tolerable for some error while the z component is most critical. The z component has accuracy and precision of 0.47 and 0.27 mm, respectively. To avoid the needle tip from penetrating the TCL, the target could be slightly retracted volar to the TCL as the average thickness of the TCL ranges from 1.3 to 3.0mm (Pacek et al., 2010). Furthermore, patients with carpal tunnel syndrome tend to have a thicker TCL (Ferrari et al., 1997; Marquardt et al., 2016) which gives us more room to maintain the needle tip inside the TCL.

There were several limitations to this study. First, the curve used to represent the center of the TCL thickness was not quantitatively validated. Rather, it was subjectively evaluated whether the needle was inserted into the TCL using ultrasound imaging. Also, the needle tip was manually selected from the ultrasound image by approximately selecting the center of the needle opening, which could account for some of the error. We also suspect that the needle slightly deflected when entering the wrist, accounting for some of the error. Further refinement can be done to improve the accuracy and precision by selecting a needle with greater stiffness to reduce deflection.

This study provides a methodology that serves the initial step necessary to implement collagenase injection as a treatment for carpal tunnel syndrome. As next steps, the effects of collagenase injection on TCL stiffness and carpal tunnel space should be investigated in situ in cadaveric hands to determine the optimal dose and duration. Future studies also are needed to examine the effects of collagenase injection on the TCL and its surrounding structures in vivo in animal models to verify safety before the injection is applied to humans.