- Email: [email protected]
Experimental approach to study arthroscopic irrigation
Available online at www.sciencedirect.com
Medical Engineering & Physics 30 (2008) 1071–1078
Experimental approach to study arthroscopic irrigation G.J.M. Tuijthof a,b,∗ , J.L. Herder b , C.N. van Dijk a a
Orthopedic Research Center Amsterdam, Department of Orthopedic Surgery, Academic Medical Centre, Meibergdreef 9, 1105 AZ Amsterdam, The Netherlands b Department of Biomechanical Engineering, Delft University of Technology, Delft, The Netherlands Received 22 October 2007; received in revised form 6 December 2007; accepted 9 January 2008
Abstract The view during arthroscopic operations is kept clear by means of irrigation. The purpose was to determine dominant parameters on irrigation performance from which design considerations were formulated for optimization of joint irrigation. An experimental approach was chosen. The set up consisted of a human joint phantom with normal operative equipment for irrigation. Disturbances of the view were simulated with blue colored ink. With this, an objective and quantitative outcome measure was defined as the time from ink injection till complete clear view (irrigation time). The irrigation times for varying parameters were evaluated: pressure and flow, configuration of in- and outflow portals, location of bleeding, two- versus three-dimensionally shaped joint space, direction and location of inflow, and presence of an instrument. Apart from the level of pressure and flow (F(5,34) = 245, p < 0.05), the configuration of in- and outflow portals had a dominant significant influence on the irrigation time (F(2,23) = 69, p < 0.05) achieving a decrease of up to 64% and 77%, respectively. The experimental approach resulted in formulation of design criteria for new sheaths: cross-sectional area as large as possible, and stimulation of a turbulent inflow. The method can be used as a standard testing protocol for new arthroscopic devices and instruments. © 2008 IPEM. Published by Elsevier Ltd. All rights reserved. Keywords: Irrigation; Experiments; Arthroscopy; Design; Comparative study; Sheath
1. Introduction Most surgeons agree that optimal view is important to perform an arthroscopic operation (minimally invasive surgical treatment in joints) safely and fast [1–4]. During arthroscopic procedures, the view is kept clear from debris and blood by irrigation of the joint with a saline solution. The irrigation is performed with an irrigation system that consists of a pump connected in series with in- and outflow tubing, a scope–sheath combination, a cannula and the joint to be operated [5]. The scope–sheath combination consists of the arthroscope which is inserted in a sheath (hollow tube) that allows irrigation fluid to flow between the inner surface of the tube and the arthroscope. In practice, maintaining a clear view ∗ Corresponding author at: Orthopedic Research Center Amsterdam, Department of Orthopedic Surgery, Academic Medical Centre, Meibergdreef 9, 1105 AZ Amsterdam, The Netherlands. Tel.: +31 205662173; fax: +31 205669117. E-mail address: [email protected] (G.J.M. Tuijthof).
is often difficult due to the condition of the joint, soft tissue blocking the view [6], occurrence of bleedings, air bubbles or debris, and inadequate use of irrigation systems [4]. In literature, little attention has been paid to an integral approach for joint irrigation optimization. Studies in this area have mainly focused on comparing the performance of arthroscopic pumps. Some of them have used measures with little distinguishing capacity such as the operation time, and number of fluid bags used [2,7]. Other studies have documented pressure and flows experimentally and clinically when using different pump systems [3,5,8–10]. However, apart from the study of the type of pump other aspects involving irrigation have not been studied, such as the contour of the joint and the choice of inflow and outflow portals. These can be equally important in achieving optimal irrigation [4]. Additionally, none of these studies included the quality of the arthroscopic view as an objective and quantitative outcome measure. From this, the hypothesis is that aspects other than pressure and flow can have an equally significant effect on the irrigation time. Thereto, a detailed analysis on irrigation performance
1350-4533/$ – see front matter © 2008 IPEM. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.medengphy.2008.01.002
1072
G.J.M. Tuijthof et al. / Medical Engineering & Physics 30 (2008) 1071–1078
has been performed, which resulted in design considerations to optimize joint irrigation. An experimental approach was chosen, because fluid mechanics depend to a large extent on empirical input in daily practice, and the operative setting could be imitated.
2. Materials and methods When treating a joint by arthroscopy, the main interest of surgeons regarding joint irrigation is its ability to obtain and maintain a certain quality of the arthroscopic view. The arthroscopic view is defined as good if no disturbances are present [11]. In practice, a disturbance cannot always be prevented. Thereto, we defined an optimal performance for a certain irrigation configuration if this permits the shortest time to clear a disturbed arthroscopic view. This definition was translated into the primary outcome measure by creating one preset type of disturbed view. This existed of the injection of 2 ml of blue colored ink in the inflow tubing. From this, the primary outcome measure was specified as the time from colored ink injection till complete clarity (irrigation time). For a fair comparison of different irrigation settings, a physical phantom was preferred over the use of human cadaver material. The use of a phantom excludes any time constraints, and enhances reproducibility. In addition, it enables the visualization of disturbed view, and therefore the study of the irrigation time in a controlled setting. The joint phantom should resemble reality on one hand, but consist of a simple configuration on the other hand. This latter requirement is important to attribute differences between conditions to the parameter that is tested. To meet both requirements,
a joint phantom was constructed of two circular glass plates (diameter 95 mm) that were positioned at a slight angle relative to one another with a distance ranging from about 6 mm to a maximum of 12 mm (Fig. 1). The glass plates were surrounded by a compliant rubber sleeve in which three portals were placed that could be closed off if required. The height of the inner phantom joint space and its volume (64 ml) were in the same order of magnitude as those of the larger joints such as the knee, the shoulder and possibly also the ankle joint. The phantom was deliberately fabricated of transparent material to visualize the distribution of the blue colored ink in the entire phantom (flow pattern), and not solely at the location just in front of the arthroscope (Fig. 1). Although the rubber sleeve allowed increased volumes when the phantom was pressurized, the actual capacity of the phantom was not important, because the experiments were performed under steady state conditions. Additionally, leakage along the portals was prevented during the experiments. Thereto, the fluid resistance of the phantom was solely determined by the inner diameter of the outflow portal. The experimental set up was completed by connecting the physical phantom to an irrigation system consisting of an automated pump (FMS Duo+© , FMSGroup, Nice, France) with its complementary tubing, and outflow cannula (inner diameter ∅ 4.5 mm). A ∅ 4 mm arthroscope (7200BW, Karl Storz Endoscopes, Tuttlingen, Germany) and matching sheath (inner diameter ∅ 4.5 mm) with two stopcocks were used. The pressure and flow were kept constant during the experiments. The flow patterns in the entire phantom were recorded by an analogue CCD camera (Fig. 1). Different items were tested for their effect on the irrigation time: pressure and flow (PF), configuration of in- and outflow
Fig. 1. Experimental set up. Left: side view of the joint phantom showing the portal locations and the compliant rubber sleeve that is connected with hose clamps. Middle: frontal view of the joint model implemented in the complete experimental set up. A video camera is positioned in front of the transparent glass plates. Inflow and the blue colored ink injection take place just before one of the upper portals. Outflow takes place via the lower portal. Right: automated pump that permits irrigation.
Table 1 The conditions and results of the first set of items that were tested: pressure and flow (PF), inflow/outflow combination (IO), location of bleeding (BL) and two-dimensionally vs. three-dimensionally shaped joint space (2D vs. 3D) Condition
2
3
4
5
6
Results
15.0 (kPa) 1.5 (ml/s)
18 (kPa) 1.5 (ml/s)
24 (kPa) 1.5 (ml/s)
15.0 (kPa) 3.3 (ml/s)
15.0 (kPa) 4.2 (ml/s)
F(5,34) = 245, p < 0.05 Posthoc: condition 1 and 2 p = 0.081 condition 5 and 6 p = 0.330
79.9(s) ± 3.3
74.5(s) ± 4.6
64.1(s) ± 2.7
47.4(s) ± 2.2
33.9(s) ± 2.5
28.7(s) ± 1.2
F(2.23)=69, p < 0.05 Posthoc: condition 1 and 3 p=0.069
In and outflow combination (IO) 74.5(s) ± 4.6
17.4(s) ± 4.1
64.3(s) ± 13.5
F(2,14) = 0.243, p = 0.788
Location bleeding (BL) 83.1(s) ± 3.8
Two-dimensionally vs. three-dimensionally shape (2D vs. 3D)
83.2(s) ± 5.4
81.4(s) ± 4.5
DIR: p < 0.05 2D vs. 3D; p = 0.89 DIR and 2D vs. 3D combined: p < 0.05
G.J.M. Tuijthof et al. / Medical Engineering & Physics 30 (2008) 1071–1078
Pressure and flow (PF)
1 9.0 (kPa) 1.5 (ml/s)
See photographs in table. Below each condition, the mean irrigation time and the standard deviation are given in seconds. In the last column, the results of the ANOVA-tests are shown with the additional outcome of the Posthoc Bonferroni-tests (see ‘Posthoc:’). The last item where the two-dimensionally shaped joint space is compared with a three-dimensionally shaped joint space was analyzed with a two-factor univariate test.
1073
1074
G.J.M. Tuijthof et al. / Medical Engineering & Physics 30 (2008) 1071–1078
portals (IO), location of bleeding (BL), two-dimensionally versus three-dimensionally shaped joint space (2D vs. 3D) (Table 1). Unless indicated otherwise, the configuration of the in- and outflow stream for all conditions (both Tables 1 and 2) was equal to IO: condition 1 (Table 1) where the advised preset pressure and flow levels were used (pressure of 15 kPa (113 mmHg), and flow of 1.5 ml/s (90 ml/min)). For IO: condition 1, inflow takes place through the scope–sheath combination, and outflow through a separate cannula. The flow and pressure were taken as one condition, because in steady state the flow is physically dependent on the pressure difference. Their testing range was similar to the values used in Dolk and Augustini [9]. The three flow levels corresponded with the predefined minimum, medial and maximum flow levels on the automated pump. The other two conditions for IO were inflow through the separate cannula and outflow through the scope–sheath combination (IO: condition 2), inflow through the scope–sheath combination and outflow through active suction by an oscillating shaver (IO: condition 3). The shaver is a cutting device that consists of two concentrically placed tubes having both an opening at the tip with sharp edges. Tissue is sucked in the outer tube and cut by the rotating inner tube. Irrigation using only the scope–sheath combination was discarded. That is, 80% of the irrigation fluid would not enter the joint phantom, if both stopcocks of the sheath were open simultaneously. Bleedings were simulated by injection of 2 ml of blue colored ink directly in the compliant rubber sleeve at three locations indicated by arrows (Table 1). The three-dimensionally shaped surface consisted of two hemisphere glass plates placed in the centers of each flat glass plate (Table 1). From fluid mechanics theory, it was derived that quick irrigation is achieved or at least stimulated by a high flow, and the flow being turbulent [12,13]. In this study, we tested the effect of the direction of the inflow (DIR) as well as the presence of an instrument in the inflow stream (INST), because it was assumed that they would stimulate the creation of a turbulent flow (Table 2). Finally, the effect of DIR, INST and 2D vs. 3D were assessed at two different locations (Table 2). The video recordings were digitized with Adobe Premiere Pro, Version 1.5 (Adobe Systems Incorporated, San Jose, California) to an AVI-format with image resolution of 720 × 576, a frame rate of 25 images per second, and a Cinepak Codec by Radius compression. Semi-automated blue color segmentation was performed to determine the irrigation time per experiment with customized software programmed in Matlab, Version 7.0.4.365 (R14) (The Mathworks, Natick, USA). The objective of the segmentation is to classify each RGB pixel in a given image as having a color in the specified range or not [14]. Two items were assessed manually. Firstly, the blue color range for segmentation in the RGB vector space was determined by visual inspection of original images versus segmented images using at least 50 frames of three different conditions (Fig. 2). For computational efficiency, a blue colored bounding box was chosen [14]. The box was centred on the average blue color. Its
dimension along each of the RGB-color axes was set by visual inspection of the blue color levels of key areas in the reference frame set. For the processing of the experiments, two bounding boxes had to be defined: one for blue detection in the shadow zone along the border of the glass plates, and one for the remaining centre area (Fig. 2). Secondly, an ellipse shaped mask was manually indicated in the first frame of each movie to include only the area of the transparent glass plates for processing. The percentage of blue colored area was determined for every fifth frame to speed up the data processing, which was sufficiently accurate. The resulting curve of percentages was smoothed with a triangular smoothing filter in order to exclude outliners caused by corrupted frames from digitizing. Since progress of blue color fading was sometimes slow, it was difficult to stop the analogue video recording at the time of complete image clarity. Thereto, a safe margin of 15% remaining blue colored area was set as the threshold level for complete clarity. One condition (Table 1, IO: condition 1) was repeated ten times. This condition had a mean irrigation time of 74.5 s (standard deviation 4.6 s). Considering design and clinical perspectives, a difference between conditions of at least 10 s would be worthwhile to adjust a protocol or design. This required five repetitions per condition with the given standard deviation of 4.6 s, and a power of 90% (α = 0.05). One-way analysis of variance tests (ANOVA, p < 0.05) were performed to assess significant differences (SPSS 12.0.2, SPSS Inc., Chicago, IL, USA). Posthoc Bonferroni-tests were additionally performed to highlight significant internal differences of within conditions. The effect of the presence of an instrument in the inflow stream, and two-dimensionally versus threedimensionally shaped joint space was analyzed for statistical differences by means of a univariate analysis of variance for two factors (p < 0.05), since these datasets contained both DIR condition 1 and 3 within their conditions.
3. Results Fig. 2 shows a subset of original and their corresponding segmented frames for PF conditions 2 and 6. The flow initially follows the direction of the scope–sheath combination with a cone shaped widening of the stream. Via the borders of the joint phantom, the course of the flow is redirected and flows back along the borders towards the inflow location. At this point, the blue colored stream is integrated with the now colorless inflow stream. This pattern is repeated until distribution through the entire joint phantom is achieved. The blue color gradually fades due to the continuous supply of colorless irrigation fluid. The color segmentation is good with some small imperfections at the transition from the shadow area at the border to the central area of the joint phantom (Fig. 2). The time until the entire joint phantom is covered maximally with blue colored ink is not significantly different for the two conditions in Fig. 2.
Table 2 The conditions and results of the second set of items that were tested: three directions of inflow (DIR) compared with IO condition 1 (Table 1), the presence of an instrument in the inflow stream (INST), and the effect of DIR, INST and 2D vs. 3D at two different locations Condition
1
2
3
4
Results
74.5(s) ± 4.6
93.0(s) ± 5.4
83.8(s) ± 2.1
77.2(s) ± 5.1
DIR: p < 0.05 INST: p < 0.05 DIR and INST combined: p = 0.06
No instrument vs. instrument (INST)
F(3.18)=7.181, p < 0.05 Posthoc: only condition 2 and 3 p < 0.05
Location 1
77.7 (s) ± 4.5
70.6 (s) ± 5.8
86.3 (s) ± 4.5
77.2 (s) ± 5.1
F(3.19)=7.084, p < 0.05 Posthoc: only condition 4 vs. p < 0.05
Location 2
81.8(s) ± 5.1
78.5(s) ± 6.0
83.2(s) ± 4.2
G.J.M. Tuijthof et al. / Medical Engineering & Physics 30 (2008) 1071–1078
F(3.24) = 20.4 p < 0.05 Posthoc: condition 1 and 4 p = 1.000 condition 3 and 4 p = 0.176
Direction inflow (DIR)
93.0(s) ± 5.4
Below each condition, the mean irrigation time and the standard deviation are given in seconds. In the last column, the results of the ANOVA-tests are shown with the additional outcome of the Posthoc Bonferroni-tests (see ‘Posthoc:’). Exception was the effect of INST which was analyzed with a univariate test.
1075
1076
G.J.M. Tuijthof et al. / Medical Engineering & Physics 30 (2008) 1071–1078
Fig. 2. Results for two different experiments: on the left PF condition 2 with a pressure of 15 kPa, and a flow of 1.5 ml/s, and on the right PF condition 6 with a pressure of 15 kPa, and a flow of 4.2 ml/s (Table 1). The pictures show the original and segmented frames indicating the flow patterns at 1, 3, 5, 10 and 15 s after the blue ink injection into the joint phantom.
Fig. 3. Results of 10 experiments repeated for PF condition 2 (which is the same condition as IO condition 1) with a pressure of 15 kPa, and a flow of 1.5 ml/s (Table 1). The graph shows the smoothed curves of the percentage of blue colored area in time. Two times are marked by a filled circle: the time for which blue color percentage is maximum, and the time for which the percentage decreases below the threshold level of 15% (bold line).
G.J.M. Tuijthof et al. / Medical Engineering & Physics 30 (2008) 1071–1078
The percentage of blue colored area is shown for all repetitions of PF condition 2 (Fig. 3). The shape of the curves matches the expected exponential response behavior caused by the simulated impulse input of the blue colored ink. The overlap of the curves suggests that the experiments are reproducible. Some variation is present in the time at which maximum blue colored area is achieved, and the time till at which less than 15% of blue colored area is left (both indicated by solid circles). The pressure and flow, the configuration of in- and outflow portals, the direction of the inflow stream, and the presence of an instrument have significant effect on the irrigation time (Tables 1 and 2). The former two show the largest differences in irrigation time. A maximum average decrease is achieved of 64% of the irrigation time between PF conditions 1 and 6, and 77% between IO conditions 1 and 2. The location of a bleeding does not effect the irrigation time significantly, but the irrigation times of BL conditions 1 and 2 are significantly larger compared to PF condition 2 (Table 1). The three-dimensionally shaped joint space shows no significant results in comparison with the two-dimensionally shaped joint space (Table 1). The effect of the presence of an instrument is significant, but depends on the location of the inflow (Table 2).
4. Discussion A detailed analysis was performed to assess the influence of different items on the irrigation performance during arthroscopic operations. The proposed experimental approach of using a simple joint phantom in combination with the colorization of flow patterns offers the possibility to analyze isolated conditions. A key step in the experimental comparison was the definition of the irrigation time, which has a direct relation with the arthroscopic view and is an objective and quantitative outcome measure. The results show the influence of pressure and flow levels on the irrigation time is significant, especially the creation of high flows. This was expected and corresponds with other experimental joint irrigation studies [3,8]. Apart from this, the hypothesis is supported that other aspects have an equally dominant effect on the irrigation time: the configuration of the in- and outflow portals. In fact, IO condition 2 gives the fastest average irrigation time for all conditions (Table 1). From fluid mechanics theory, it is known that the flow is determined by division of the total pressure drop along the irrigation system, and the sum of its restrictions [1,5]. Thus, if a restriction is decreased, a higher flow can be created at the same pressure level. For pipes and tubing, the cross-sectional area determines the size of the restriction to a large extent [1,12]. The cross-sectional area of the scope–sheath combination is around one-fifth of the cannula. When comparing IO conditions 1 and 2, the inflow restriction for IO condition 2 is significantly lower, causing a higher local flow. Additionally, the larger diameter increases the
1077
Reynolds number, and contributes to the flow becoming turbulent [12,13]. This causes quicker distribution and mixing of the inflow stream entering the joint phantom. Concluding, a turbulent inflow stream significantly reduces the irrigation time. The fact that turbulence is beneficial is supported by the results of the direction of the inflow stream (DIR), and the presence of an instrument in the inflow stream (INST) (Table 2). In case of DIR, the inflow stream in condition 1 is smoothly directed along the border of the joint phantom thereby maintaining a laminar stream (Table 2). As a result, the blue ink is gradually mixed with the colorless fluid in the joint phantom, which increases the irrigation time. In case of INST, the instrument that is placed in the inflow stream blocks a continuous course of the stream, and forces it to be distributed in different directions which results in quicker mixing of fluid. All other items that were tested did not have major impact on the irrigation time. If a larger volume of blue colored ink would have been injected or a larger number of repetitions would have been performed, the irrigation times would probably show less variation, and other items could possibly have shown significant differences. However, the differences in irrigation time would probably be too small to be of value in clinical practice. An issue that needs discussion is the fact that in this study not the arthroscopic view itself, but the overall view in the joint phantom was analyzed. We believed that this gave more insight in the flow patterns and helped in the explanation of certain phenomena. In general, the expectation is that a quicker irrigation time for the overall view corresponds to a quicker irrigation time of the arthroscopic view. The assessment of the type of correlation between the arthroscopic view and the overall view remains to be determined. From the results, important design criteria can be formulated for new cannula or sheaths: they should have a large as possible cross-sectional area, when used as inflow they should stimulate a turbulent nature of the flow, and when operating with a two-portal technique they should enable continuous separated in- and outflow streams to ensure that all fluid reaches the joint [4].
5. Conclusions The presented experimental approach enables the study of joint irrigation in detail thereby imitating the clinical practice sufficiently within a controlled setting. This study lead to new insights in joint irrigation, and made it possible to formulate design criteria that are directly applicable for optimization of irrigation by means of hardware redesign. In the near future, we aim at development of new sheaths and cannulae that will be evaluated technically with the same protocol. Finally, the proposed method is a good candidate to be used as a standardized platform and protocol for evaluation of different types of arthroscopic devices and instruments.
1078
G.J.M. Tuijthof et al. / Medical Engineering & Physics 30 (2008) 1071–1078
Acknowledgments The authors wish to thank A. van der Pijl for his efforts in constructing the physical joint phantom, P. Heeman of the Medical Technology Department (AMC, NL) for his contributions in acquiring all data in a digital format, and I. Paardekoper and A.-J. Punt for the performance of a pilot study. Finally, Arsis Medical bv (De Bilt, NL) is especially thanked for providing the FMS Duo pump used in the experiments. Role of the funding source: This research was co-funded by the Minimally Invasive Surgery and Interventional Techniques Program, Delft University of Technology, Delft, The Netherlands (http://mms.tudelft.nl/misit/index.htm), and by the Technology Foundation STW, applied science division of NOW and the technology program of the Ministry of Economic Affairs, The Netherlands. Both grants had no involvement in the study design, the collection, analysis and interpretation of data; in the writing of the manuscript; and in the decision to submit the manuscript for publication. The same holds for Arsis Medical bv.
Conflict of interest All authors disclose any financial and personal relationships with other people or organization that could inappropriately influence (bias) their work.
References [1] Morgan CD. Fluid delivery systems for arthroscopy. Arthroscopy 1987;3(4):288–91. [2] Ogilvie-Harris DJ, Weisleder L. Fluid pump systems for arthroscopy: a comparison of pressure control versus pressure and flow control. Arthroscopy 1995;11(5):591–5. [3] Muellner T, Menth-Chiari WA, Reihsner R, Eberhardsteiner J, Engebretsen L. Accuracy of pressure and flow capacities of four arthroscopic fluid management systems. Arthroscopy 2001;17(7):760–4. [4] Tuijthof GJM, Technical improvement of arthroscopic techniques. PhD Delft University of Technology, Delft, The Netherlands; 2003. [5] Tuijthof GJ, Dusee L, Herder JL, van Dijk CN, Pistecky PV. Behavior of arthroscopic irrigation systems. Knee Surg Sports Traumatol Arthrosc 2005;13(3):238–46. [6] Noyes FR, Spievack ES. Extraarticular fluid dissection in tissues during arthroscopy. Am J Sports Med 1982;10(6):346–51. [7] Ampat G, Bruguera J, Copeland SA. Aquaflo pump vs. FMS 4 pump for shoulder arthroscopic surgery. Ann Royal Coll Surg Engl 1997;79:341–4. [8] Dolk T, Augustini BG. Three irrigation systems for motorized arthroscopic surgery: a comparative experimental and clinical study. Arthroscopy 1989;5(4):307–14. [9] Dolk T, Augustini BG. Elevated position of the irrigation outflow during arthroscopy: an experimental study. Arthroscopy 1989;5(2):93–6. [10] Ewing JW, Noe DA, Kitaoka HB, Askew MJ. Intra-articular pressures during arthroscopic knee surgery. Arthroscopy 1986;2(4):264–9. [11] Tuijthof GJM, Sierevelt IN, van Dijk CN. Disturbances in the arthroscopic view defined with video analysis. Knee Surg Sports Traumatol Arthrosc 2007;15(9):1101–6. [12] White FM. Fluid mechanics. 4th ed. Boston, USA: McGraw-Hill; 1999. [13] Leijdens H. Stroming en warmteoverdracht I. The Netherlands: DUT, Delft; 1992. [14] Gonzalez R, Woods R. Color image processing. In: Hattori R, Woods R, editors. Digital image processing. 2nd ed. Upper Saddle River, New Jersey, USA: Prentice Hall; 2002. p. 282–348.
Medical Engineering & Physics 30 (2008) 1071–1078
Experimental approach to study arthroscopic irrigation G.J.M. Tuijthof a,b,∗ , J.L. Herder b , C.N. van Dijk a a
Orthopedic Research Center Amsterdam, Department of Orthopedic Surgery, Academic Medical Centre, Meibergdreef 9, 1105 AZ Amsterdam, The Netherlands b Department of Biomechanical Engineering, Delft University of Technology, Delft, The Netherlands Received 22 October 2007; received in revised form 6 December 2007; accepted 9 January 2008
Abstract The view during arthroscopic operations is kept clear by means of irrigation. The purpose was to determine dominant parameters on irrigation performance from which design considerations were formulated for optimization of joint irrigation. An experimental approach was chosen. The set up consisted of a human joint phantom with normal operative equipment for irrigation. Disturbances of the view were simulated with blue colored ink. With this, an objective and quantitative outcome measure was defined as the time from ink injection till complete clear view (irrigation time). The irrigation times for varying parameters were evaluated: pressure and flow, configuration of in- and outflow portals, location of bleeding, two- versus three-dimensionally shaped joint space, direction and location of inflow, and presence of an instrument. Apart from the level of pressure and flow (F(5,34) = 245, p < 0.05), the configuration of in- and outflow portals had a dominant significant influence on the irrigation time (F(2,23) = 69, p < 0.05) achieving a decrease of up to 64% and 77%, respectively. The experimental approach resulted in formulation of design criteria for new sheaths: cross-sectional area as large as possible, and stimulation of a turbulent inflow. The method can be used as a standard testing protocol for new arthroscopic devices and instruments. © 2008 IPEM. Published by Elsevier Ltd. All rights reserved. Keywords: Irrigation; Experiments; Arthroscopy; Design; Comparative study; Sheath
1. Introduction Most surgeons agree that optimal view is important to perform an arthroscopic operation (minimally invasive surgical treatment in joints) safely and fast [1–4]. During arthroscopic procedures, the view is kept clear from debris and blood by irrigation of the joint with a saline solution. The irrigation is performed with an irrigation system that consists of a pump connected in series with in- and outflow tubing, a scope–sheath combination, a cannula and the joint to be operated [5]. The scope–sheath combination consists of the arthroscope which is inserted in a sheath (hollow tube) that allows irrigation fluid to flow between the inner surface of the tube and the arthroscope. In practice, maintaining a clear view ∗ Corresponding author at: Orthopedic Research Center Amsterdam, Department of Orthopedic Surgery, Academic Medical Centre, Meibergdreef 9, 1105 AZ Amsterdam, The Netherlands. Tel.: +31 205662173; fax: +31 205669117. E-mail address: [email protected] (G.J.M. Tuijthof).
is often difficult due to the condition of the joint, soft tissue blocking the view [6], occurrence of bleedings, air bubbles or debris, and inadequate use of irrigation systems [4]. In literature, little attention has been paid to an integral approach for joint irrigation optimization. Studies in this area have mainly focused on comparing the performance of arthroscopic pumps. Some of them have used measures with little distinguishing capacity such as the operation time, and number of fluid bags used [2,7]. Other studies have documented pressure and flows experimentally and clinically when using different pump systems [3,5,8–10]. However, apart from the study of the type of pump other aspects involving irrigation have not been studied, such as the contour of the joint and the choice of inflow and outflow portals. These can be equally important in achieving optimal irrigation [4]. Additionally, none of these studies included the quality of the arthroscopic view as an objective and quantitative outcome measure. From this, the hypothesis is that aspects other than pressure and flow can have an equally significant effect on the irrigation time. Thereto, a detailed analysis on irrigation performance
1350-4533/$ – see front matter © 2008 IPEM. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.medengphy.2008.01.002
1072
G.J.M. Tuijthof et al. / Medical Engineering & Physics 30 (2008) 1071–1078
has been performed, which resulted in design considerations to optimize joint irrigation. An experimental approach was chosen, because fluid mechanics depend to a large extent on empirical input in daily practice, and the operative setting could be imitated.
2. Materials and methods When treating a joint by arthroscopy, the main interest of surgeons regarding joint irrigation is its ability to obtain and maintain a certain quality of the arthroscopic view. The arthroscopic view is defined as good if no disturbances are present [11]. In practice, a disturbance cannot always be prevented. Thereto, we defined an optimal performance for a certain irrigation configuration if this permits the shortest time to clear a disturbed arthroscopic view. This definition was translated into the primary outcome measure by creating one preset type of disturbed view. This existed of the injection of 2 ml of blue colored ink in the inflow tubing. From this, the primary outcome measure was specified as the time from colored ink injection till complete clarity (irrigation time). For a fair comparison of different irrigation settings, a physical phantom was preferred over the use of human cadaver material. The use of a phantom excludes any time constraints, and enhances reproducibility. In addition, it enables the visualization of disturbed view, and therefore the study of the irrigation time in a controlled setting. The joint phantom should resemble reality on one hand, but consist of a simple configuration on the other hand. This latter requirement is important to attribute differences between conditions to the parameter that is tested. To meet both requirements,
a joint phantom was constructed of two circular glass plates (diameter 95 mm) that were positioned at a slight angle relative to one another with a distance ranging from about 6 mm to a maximum of 12 mm (Fig. 1). The glass plates were surrounded by a compliant rubber sleeve in which three portals were placed that could be closed off if required. The height of the inner phantom joint space and its volume (64 ml) were in the same order of magnitude as those of the larger joints such as the knee, the shoulder and possibly also the ankle joint. The phantom was deliberately fabricated of transparent material to visualize the distribution of the blue colored ink in the entire phantom (flow pattern), and not solely at the location just in front of the arthroscope (Fig. 1). Although the rubber sleeve allowed increased volumes when the phantom was pressurized, the actual capacity of the phantom was not important, because the experiments were performed under steady state conditions. Additionally, leakage along the portals was prevented during the experiments. Thereto, the fluid resistance of the phantom was solely determined by the inner diameter of the outflow portal. The experimental set up was completed by connecting the physical phantom to an irrigation system consisting of an automated pump (FMS Duo+© , FMSGroup, Nice, France) with its complementary tubing, and outflow cannula (inner diameter ∅ 4.5 mm). A ∅ 4 mm arthroscope (7200BW, Karl Storz Endoscopes, Tuttlingen, Germany) and matching sheath (inner diameter ∅ 4.5 mm) with two stopcocks were used. The pressure and flow were kept constant during the experiments. The flow patterns in the entire phantom were recorded by an analogue CCD camera (Fig. 1). Different items were tested for their effect on the irrigation time: pressure and flow (PF), configuration of in- and outflow
Fig. 1. Experimental set up. Left: side view of the joint phantom showing the portal locations and the compliant rubber sleeve that is connected with hose clamps. Middle: frontal view of the joint model implemented in the complete experimental set up. A video camera is positioned in front of the transparent glass plates. Inflow and the blue colored ink injection take place just before one of the upper portals. Outflow takes place via the lower portal. Right: automated pump that permits irrigation.
Table 1 The conditions and results of the first set of items that were tested: pressure and flow (PF), inflow/outflow combination (IO), location of bleeding (BL) and two-dimensionally vs. three-dimensionally shaped joint space (2D vs. 3D) Condition
2
3
4
5
6
Results
15.0 (kPa) 1.5 (ml/s)
18 (kPa) 1.5 (ml/s)
24 (kPa) 1.5 (ml/s)
15.0 (kPa) 3.3 (ml/s)
15.0 (kPa) 4.2 (ml/s)
F(5,34) = 245, p < 0.05 Posthoc: condition 1 and 2 p = 0.081 condition 5 and 6 p = 0.330
79.9(s) ± 3.3
74.5(s) ± 4.6
64.1(s) ± 2.7
47.4(s) ± 2.2
33.9(s) ± 2.5
28.7(s) ± 1.2
F(2.23)=69, p < 0.05 Posthoc: condition 1 and 3 p=0.069
In and outflow combination (IO) 74.5(s) ± 4.6
17.4(s) ± 4.1
64.3(s) ± 13.5
F(2,14) = 0.243, p = 0.788
Location bleeding (BL) 83.1(s) ± 3.8
Two-dimensionally vs. three-dimensionally shape (2D vs. 3D)
83.2(s) ± 5.4
81.4(s) ± 4.5
DIR: p < 0.05 2D vs. 3D; p = 0.89 DIR and 2D vs. 3D combined: p < 0.05
G.J.M. Tuijthof et al. / Medical Engineering & Physics 30 (2008) 1071–1078
Pressure and flow (PF)
1 9.0 (kPa) 1.5 (ml/s)
See photographs in table. Below each condition, the mean irrigation time and the standard deviation are given in seconds. In the last column, the results of the ANOVA-tests are shown with the additional outcome of the Posthoc Bonferroni-tests (see ‘Posthoc:’). The last item where the two-dimensionally shaped joint space is compared with a three-dimensionally shaped joint space was analyzed with a two-factor univariate test.
1073
1074
G.J.M. Tuijthof et al. / Medical Engineering & Physics 30 (2008) 1071–1078
portals (IO), location of bleeding (BL), two-dimensionally versus three-dimensionally shaped joint space (2D vs. 3D) (Table 1). Unless indicated otherwise, the configuration of the in- and outflow stream for all conditions (both Tables 1 and 2) was equal to IO: condition 1 (Table 1) where the advised preset pressure and flow levels were used (pressure of 15 kPa (113 mmHg), and flow of 1.5 ml/s (90 ml/min)). For IO: condition 1, inflow takes place through the scope–sheath combination, and outflow through a separate cannula. The flow and pressure were taken as one condition, because in steady state the flow is physically dependent on the pressure difference. Their testing range was similar to the values used in Dolk and Augustini [9]. The three flow levels corresponded with the predefined minimum, medial and maximum flow levels on the automated pump. The other two conditions for IO were inflow through the separate cannula and outflow through the scope–sheath combination (IO: condition 2), inflow through the scope–sheath combination and outflow through active suction by an oscillating shaver (IO: condition 3). The shaver is a cutting device that consists of two concentrically placed tubes having both an opening at the tip with sharp edges. Tissue is sucked in the outer tube and cut by the rotating inner tube. Irrigation using only the scope–sheath combination was discarded. That is, 80% of the irrigation fluid would not enter the joint phantom, if both stopcocks of the sheath were open simultaneously. Bleedings were simulated by injection of 2 ml of blue colored ink directly in the compliant rubber sleeve at three locations indicated by arrows (Table 1). The three-dimensionally shaped surface consisted of two hemisphere glass plates placed in the centers of each flat glass plate (Table 1). From fluid mechanics theory, it was derived that quick irrigation is achieved or at least stimulated by a high flow, and the flow being turbulent [12,13]. In this study, we tested the effect of the direction of the inflow (DIR) as well as the presence of an instrument in the inflow stream (INST), because it was assumed that they would stimulate the creation of a turbulent flow (Table 2). Finally, the effect of DIR, INST and 2D vs. 3D were assessed at two different locations (Table 2). The video recordings were digitized with Adobe Premiere Pro, Version 1.5 (Adobe Systems Incorporated, San Jose, California) to an AVI-format with image resolution of 720 × 576, a frame rate of 25 images per second, and a Cinepak Codec by Radius compression. Semi-automated blue color segmentation was performed to determine the irrigation time per experiment with customized software programmed in Matlab, Version 7.0.4.365 (R14) (The Mathworks, Natick, USA). The objective of the segmentation is to classify each RGB pixel in a given image as having a color in the specified range or not [14]. Two items were assessed manually. Firstly, the blue color range for segmentation in the RGB vector space was determined by visual inspection of original images versus segmented images using at least 50 frames of three different conditions (Fig. 2). For computational efficiency, a blue colored bounding box was chosen [14]. The box was centred on the average blue color. Its
dimension along each of the RGB-color axes was set by visual inspection of the blue color levels of key areas in the reference frame set. For the processing of the experiments, two bounding boxes had to be defined: one for blue detection in the shadow zone along the border of the glass plates, and one for the remaining centre area (Fig. 2). Secondly, an ellipse shaped mask was manually indicated in the first frame of each movie to include only the area of the transparent glass plates for processing. The percentage of blue colored area was determined for every fifth frame to speed up the data processing, which was sufficiently accurate. The resulting curve of percentages was smoothed with a triangular smoothing filter in order to exclude outliners caused by corrupted frames from digitizing. Since progress of blue color fading was sometimes slow, it was difficult to stop the analogue video recording at the time of complete image clarity. Thereto, a safe margin of 15% remaining blue colored area was set as the threshold level for complete clarity. One condition (Table 1, IO: condition 1) was repeated ten times. This condition had a mean irrigation time of 74.5 s (standard deviation 4.6 s). Considering design and clinical perspectives, a difference between conditions of at least 10 s would be worthwhile to adjust a protocol or design. This required five repetitions per condition with the given standard deviation of 4.6 s, and a power of 90% (α = 0.05). One-way analysis of variance tests (ANOVA, p < 0.05) were performed to assess significant differences (SPSS 12.0.2, SPSS Inc., Chicago, IL, USA). Posthoc Bonferroni-tests were additionally performed to highlight significant internal differences of within conditions. The effect of the presence of an instrument in the inflow stream, and two-dimensionally versus threedimensionally shaped joint space was analyzed for statistical differences by means of a univariate analysis of variance for two factors (p < 0.05), since these datasets contained both DIR condition 1 and 3 within their conditions.
3. Results Fig. 2 shows a subset of original and their corresponding segmented frames for PF conditions 2 and 6. The flow initially follows the direction of the scope–sheath combination with a cone shaped widening of the stream. Via the borders of the joint phantom, the course of the flow is redirected and flows back along the borders towards the inflow location. At this point, the blue colored stream is integrated with the now colorless inflow stream. This pattern is repeated until distribution through the entire joint phantom is achieved. The blue color gradually fades due to the continuous supply of colorless irrigation fluid. The color segmentation is good with some small imperfections at the transition from the shadow area at the border to the central area of the joint phantom (Fig. 2). The time until the entire joint phantom is covered maximally with blue colored ink is not significantly different for the two conditions in Fig. 2.
Table 2 The conditions and results of the second set of items that were tested: three directions of inflow (DIR) compared with IO condition 1 (Table 1), the presence of an instrument in the inflow stream (INST), and the effect of DIR, INST and 2D vs. 3D at two different locations Condition
1
2
3
4
Results
74.5(s) ± 4.6
93.0(s) ± 5.4
83.8(s) ± 2.1
77.2(s) ± 5.1
DIR: p < 0.05 INST: p < 0.05 DIR and INST combined: p = 0.06
No instrument vs. instrument (INST)
F(3.18)=7.181, p < 0.05 Posthoc: only condition 2 and 3 p < 0.05
Location 1
77.7 (s) ± 4.5
70.6 (s) ± 5.8
86.3 (s) ± 4.5
77.2 (s) ± 5.1
F(3.19)=7.084, p < 0.05 Posthoc: only condition 4 vs. p < 0.05
Location 2
81.8(s) ± 5.1
78.5(s) ± 6.0
83.2(s) ± 4.2
G.J.M. Tuijthof et al. / Medical Engineering & Physics 30 (2008) 1071–1078
F(3.24) = 20.4 p < 0.05 Posthoc: condition 1 and 4 p = 1.000 condition 3 and 4 p = 0.176
Direction inflow (DIR)
93.0(s) ± 5.4
Below each condition, the mean irrigation time and the standard deviation are given in seconds. In the last column, the results of the ANOVA-tests are shown with the additional outcome of the Posthoc Bonferroni-tests (see ‘Posthoc:’). Exception was the effect of INST which was analyzed with a univariate test.
1075
1076
G.J.M. Tuijthof et al. / Medical Engineering & Physics 30 (2008) 1071–1078
Fig. 2. Results for two different experiments: on the left PF condition 2 with a pressure of 15 kPa, and a flow of 1.5 ml/s, and on the right PF condition 6 with a pressure of 15 kPa, and a flow of 4.2 ml/s (Table 1). The pictures show the original and segmented frames indicating the flow patterns at 1, 3, 5, 10 and 15 s after the blue ink injection into the joint phantom.
Fig. 3. Results of 10 experiments repeated for PF condition 2 (which is the same condition as IO condition 1) with a pressure of 15 kPa, and a flow of 1.5 ml/s (Table 1). The graph shows the smoothed curves of the percentage of blue colored area in time. Two times are marked by a filled circle: the time for which blue color percentage is maximum, and the time for which the percentage decreases below the threshold level of 15% (bold line).
G.J.M. Tuijthof et al. / Medical Engineering & Physics 30 (2008) 1071–1078
The percentage of blue colored area is shown for all repetitions of PF condition 2 (Fig. 3). The shape of the curves matches the expected exponential response behavior caused by the simulated impulse input of the blue colored ink. The overlap of the curves suggests that the experiments are reproducible. Some variation is present in the time at which maximum blue colored area is achieved, and the time till at which less than 15% of blue colored area is left (both indicated by solid circles). The pressure and flow, the configuration of in- and outflow portals, the direction of the inflow stream, and the presence of an instrument have significant effect on the irrigation time (Tables 1 and 2). The former two show the largest differences in irrigation time. A maximum average decrease is achieved of 64% of the irrigation time between PF conditions 1 and 6, and 77% between IO conditions 1 and 2. The location of a bleeding does not effect the irrigation time significantly, but the irrigation times of BL conditions 1 and 2 are significantly larger compared to PF condition 2 (Table 1). The three-dimensionally shaped joint space shows no significant results in comparison with the two-dimensionally shaped joint space (Table 1). The effect of the presence of an instrument is significant, but depends on the location of the inflow (Table 2).
4. Discussion A detailed analysis was performed to assess the influence of different items on the irrigation performance during arthroscopic operations. The proposed experimental approach of using a simple joint phantom in combination with the colorization of flow patterns offers the possibility to analyze isolated conditions. A key step in the experimental comparison was the definition of the irrigation time, which has a direct relation with the arthroscopic view and is an objective and quantitative outcome measure. The results show the influence of pressure and flow levels on the irrigation time is significant, especially the creation of high flows. This was expected and corresponds with other experimental joint irrigation studies [3,8]. Apart from this, the hypothesis is supported that other aspects have an equally dominant effect on the irrigation time: the configuration of the in- and outflow portals. In fact, IO condition 2 gives the fastest average irrigation time for all conditions (Table 1). From fluid mechanics theory, it is known that the flow is determined by division of the total pressure drop along the irrigation system, and the sum of its restrictions [1,5]. Thus, if a restriction is decreased, a higher flow can be created at the same pressure level. For pipes and tubing, the cross-sectional area determines the size of the restriction to a large extent [1,12]. The cross-sectional area of the scope–sheath combination is around one-fifth of the cannula. When comparing IO conditions 1 and 2, the inflow restriction for IO condition 2 is significantly lower, causing a higher local flow. Additionally, the larger diameter increases the
1077
Reynolds number, and contributes to the flow becoming turbulent [12,13]. This causes quicker distribution and mixing of the inflow stream entering the joint phantom. Concluding, a turbulent inflow stream significantly reduces the irrigation time. The fact that turbulence is beneficial is supported by the results of the direction of the inflow stream (DIR), and the presence of an instrument in the inflow stream (INST) (Table 2). In case of DIR, the inflow stream in condition 1 is smoothly directed along the border of the joint phantom thereby maintaining a laminar stream (Table 2). As a result, the blue ink is gradually mixed with the colorless fluid in the joint phantom, which increases the irrigation time. In case of INST, the instrument that is placed in the inflow stream blocks a continuous course of the stream, and forces it to be distributed in different directions which results in quicker mixing of fluid. All other items that were tested did not have major impact on the irrigation time. If a larger volume of blue colored ink would have been injected or a larger number of repetitions would have been performed, the irrigation times would probably show less variation, and other items could possibly have shown significant differences. However, the differences in irrigation time would probably be too small to be of value in clinical practice. An issue that needs discussion is the fact that in this study not the arthroscopic view itself, but the overall view in the joint phantom was analyzed. We believed that this gave more insight in the flow patterns and helped in the explanation of certain phenomena. In general, the expectation is that a quicker irrigation time for the overall view corresponds to a quicker irrigation time of the arthroscopic view. The assessment of the type of correlation between the arthroscopic view and the overall view remains to be determined. From the results, important design criteria can be formulated for new cannula or sheaths: they should have a large as possible cross-sectional area, when used as inflow they should stimulate a turbulent nature of the flow, and when operating with a two-portal technique they should enable continuous separated in- and outflow streams to ensure that all fluid reaches the joint [4].
5. Conclusions The presented experimental approach enables the study of joint irrigation in detail thereby imitating the clinical practice sufficiently within a controlled setting. This study lead to new insights in joint irrigation, and made it possible to formulate design criteria that are directly applicable for optimization of irrigation by means of hardware redesign. In the near future, we aim at development of new sheaths and cannulae that will be evaluated technically with the same protocol. Finally, the proposed method is a good candidate to be used as a standardized platform and protocol for evaluation of different types of arthroscopic devices and instruments.
1078
G.J.M. Tuijthof et al. / Medical Engineering & Physics 30 (2008) 1071–1078
Acknowledgments The authors wish to thank A. van der Pijl for his efforts in constructing the physical joint phantom, P. Heeman of the Medical Technology Department (AMC, NL) for his contributions in acquiring all data in a digital format, and I. Paardekoper and A.-J. Punt for the performance of a pilot study. Finally, Arsis Medical bv (De Bilt, NL) is especially thanked for providing the FMS Duo pump used in the experiments. Role of the funding source: This research was co-funded by the Minimally Invasive Surgery and Interventional Techniques Program, Delft University of Technology, Delft, The Netherlands (http://mms.tudelft.nl/misit/index.htm), and by the Technology Foundation STW, applied science division of NOW and the technology program of the Ministry of Economic Affairs, The Netherlands. Both grants had no involvement in the study design, the collection, analysis and interpretation of data; in the writing of the manuscript; and in the decision to submit the manuscript for publication. The same holds for Arsis Medical bv.
Conflict of interest All authors disclose any financial and personal relationships with other people or organization that could inappropriately influence (bias) their work.
References [1] Morgan CD. Fluid delivery systems for arthroscopy. Arthroscopy 1987;3(4):288–91. [2] Ogilvie-Harris DJ, Weisleder L. Fluid pump systems for arthroscopy: a comparison of pressure control versus pressure and flow control. Arthroscopy 1995;11(5):591–5. [3] Muellner T, Menth-Chiari WA, Reihsner R, Eberhardsteiner J, Engebretsen L. Accuracy of pressure and flow capacities of four arthroscopic fluid management systems. Arthroscopy 2001;17(7):760–4. [4] Tuijthof GJM, Technical improvement of arthroscopic techniques. PhD Delft University of Technology, Delft, The Netherlands; 2003. [5] Tuijthof GJ, Dusee L, Herder JL, van Dijk CN, Pistecky PV. Behavior of arthroscopic irrigation systems. Knee Surg Sports Traumatol Arthrosc 2005;13(3):238–46. [6] Noyes FR, Spievack ES. Extraarticular fluid dissection in tissues during arthroscopy. Am J Sports Med 1982;10(6):346–51. [7] Ampat G, Bruguera J, Copeland SA. Aquaflo pump vs. FMS 4 pump for shoulder arthroscopic surgery. Ann Royal Coll Surg Engl 1997;79:341–4. [8] Dolk T, Augustini BG. Three irrigation systems for motorized arthroscopic surgery: a comparative experimental and clinical study. Arthroscopy 1989;5(4):307–14. [9] Dolk T, Augustini BG. Elevated position of the irrigation outflow during arthroscopy: an experimental study. Arthroscopy 1989;5(2):93–6. [10] Ewing JW, Noe DA, Kitaoka HB, Askew MJ. Intra-articular pressures during arthroscopic knee surgery. Arthroscopy 1986;2(4):264–9. [11] Tuijthof GJM, Sierevelt IN, van Dijk CN. Disturbances in the arthroscopic view defined with video analysis. Knee Surg Sports Traumatol Arthrosc 2007;15(9):1101–6. [12] White FM. Fluid mechanics. 4th ed. Boston, USA: McGraw-Hill; 1999. [13] Leijdens H. Stroming en warmteoverdracht I. The Netherlands: DUT, Delft; 1992. [14] Gonzalez R, Woods R. Color image processing. In: Hattori R, Woods R, editors. Digital image processing. 2nd ed. Upper Saddle River, New Jersey, USA: Prentice Hall; 2002. p. 282–348.