bs_bs_banner
Copyright © 2014 International Center for Artificial Organs and Transplantation and Wiley Periodicals, Inc.
Feasibility of the Optical Imaging of Thrombus Formation in a Rotary Blood Pump by Near-Infrared Light *Daisuke Sakota, †Tomotaka Murashige, *Ryo Kosaka, *Masahiro Nishida, and *Osamu Maruyama *National Institute of Advanced Industrial Science and Technology, Tsukuba; and †Graduate School of Science and Technology, Tokyo University of Science, Noda, Japan
Abstract: Blood coagulation is one of the primary concerns when using mechanical circulatory support devices such as blood pumps. Noninvasive detection and imaging of thrombus formation is useful not only for the development of more hemocompatible devices but also for the management of blood coagulation to avoid risk of infarction. The objective of this study is to investigate the use of near-infrared light for imaging of thrombus formation in a rotary blood pump. The optical properties of a thrombus at wavelengths ranging from 600 to 750 nm were analyzed using a hyperspectral imaging (HSI) system. A specially designed hydrodynamically levitated centrifugal blood pump with a visible bottom area was used. In vitro antithrombogenic testing was conducted five times with the pump using bovine whole blood in which the activated
blood clotting time was adjusted to 200 s prior to the experiment. Two halogen lights were used for the light sources. The forward scattering through the pump and backward scattering on the pump bottom area were imaged using the HSI system. HSI showed an increase in forward scattering at wavelengths ranging from 670 to 750 nm in the location of thrombus formation. The time at which the thrombus began to form in the impeller rotating at 2780 rpm could be detected. The spectral difference between the whole blood and the thrombus was utilized to image thrombus formation. The results indicate the feasibility of dynamically detecting and imaging thrombus formation in a rotary blood pump. Key Words: Hemocompatibility—Hyperspectral imaging—Nearinfrared light—Rotary blood pump—Thrombus.
Blood coagulation is one of the primary concerns when using mechanical circulatory support devices such as blood pumps and oxygenators, from both research-and-development and clinical-use perspectives. Improved techniques for detecting or imaging thrombus formation will contribute enormously not only to the development of more hemocompatible devices but also to the management of blood coagulation to obviate cerebral (1,2) or renal (3) infarction in patients requiring mechanical circulatory support.
Ultrasound techniques to detect coronary or arterial microemboli have been proposed (4,5). However, the proposed methods are inadequate because ultrasound does not penetrate the rigid surface of mechanical devices. Near-infrared light has been used as a noninvasive diagnostic tool for blood. Light attenuation in blood results from both scattering and absorption; however, at near-infrared wavelengths from 700 to 900 nm, the light absorption of hemoglobin and water decreases (6). Therefore, the light can be used to obtain information from blood and deeper tissues. Light-scattering microemboli detectors (LSMDs), which apply a near-infrared laser or a light-emitting diode at wavelengths of 624 (7), 632 (8), 805 (9), 810 (10), or 830 (7,11–13) nm, can detect microthrombi flowing in the extracorporeal circuit. However, no spectroscopic analytical method has been investigated. In addition, as these LSMDs obtain a single-point measurement, they are not applicable for detecting and imaging thrombus formation within the pump.
doi:10.1111/aor.12377 Received February 2014; revised June 2014. Address correspondence and reprint requests to Dr. Daisuke Sakota, Artificial Organ Group, Human Technology Research Institute, National Institute of Advanced Industrial Science and Technology, Namiki 1-2-1 Tsukuba, Ibaraki 305-8564, Japan. E-mail: [email protected] Presented in part at the 21st Congress the International Society for Rotary Blood Pumps held September 26–28, 2013 in Yokohama, Japan.
Artificial Organs 2014, 38(9):733–740
734
D. SAKOTA ET AL.
To further advance the evaluation of device hemocompatibility and techniques for diagnosing risk of thrombosis, this study proposes hyperspectral imaging (HSI) of a rotary blood pump. Although multispectral imaging deals with several images at discrete bands, HSI produces contiguous spectral images at narrow spectral bandwidths. In biomedical optics, HSI has mainly been applied to quantitative imaging of the oxygen saturation and hemoglobin concentration in the vascular network of tissues or tumors (14–16). The optical properties of blood mainly depend on physiological and biochemical parameters such as hematocrit, flow, osmolarity, hemolysis, and oxygen saturation (17), as these parameters are significantly affected by red blood cells (RBCs), which are the main source of light scattering and absorption in blood. However, little is known about the spectral difference between normal whole blood and thrombus. The objective of this study is to develop an optical imaging system to visualize thrombus formation in a rotary blood pump during in vitro antithrombogenic testing (18). To achieve this objective, the difference in optical characteristics between normal whole blood and thrombus was investigated using an HSI system in the wavelength range from 600 to 750 nm. Individual wavelength images obtained by the HSI system were analyzed, and the optimal wavelength to obtain the clearest image of thrombus formation was determined. MATERIALS AND METHODS In vitro antithrombogenic testing In vitro antithrombogenic testing as proposed by Maruyama et al. (18) was conducted five times using bovine whole blood (Funakoshi Co., Ltd., Tokyo, Japan). The blood was mixed with anticoagulant (100 mL sodium citrate/1000 mL blood) immediately after being collected at a slaughterhouse. The hematocrit value from the five tests was 35.4 ± 1.5% (mean ± SD). Prior to the experiment, the oxygen saturation of the blood was adjusted to 100% using an oxygenator and a rotary blood pump to unify and maintain the optical absorption characteristics of blood during all experiments. The oxygenated blood was then added to the closed-loop extracorporeal circulation circuit. The test pump was a hydrodynamically levitated centrifugal blood pump developed by our group. The basic structure and levitation principle were very similar to those of a previously reported pump (19,20). This pump comprises three parts: the top casing, the bottom casing, and the impeller. The stator coil is assembled in the top Artif Organs, Vol. 38, No. 9, 2014
casing. Therefore, the bottom of the fabricated pump is visible. The inner blood layer thickness of the impeller, the thickness of six vanes, is 4.9 mm. The hydrodynamic force generated on the shroud with the spiral groove attached to the vanes levitates the impeller. Before the study, the location of the levitated impeller was confirmed by a mock circulation loop experiment using a 38% (wt/wt) glycerol–water solution as the working fluid. Results from the mock circulation loop experiment indicated a bottom bearing gap of 72 μm at a rotational speed of 2780 rpm, while the top bearing gap was 262 μm. The pump and a heparin-coated 500-mL volumetric blood reservoir (specially ordered products; Senko Medical Instrument Mfg. Co., Ltd., Tokyo, Japan) were connected with a heparin-coated polyvinyl chloride tube (MERA Exceline-H 3/8 × 3/32, Senko Medical Instrument Mfg. Co., Ltd.), which included heparin-coated sampling ports (3/8 × 3/8, Senko Medical Instrument Mfg. Co., Ltd). A pressure gauge (Toyoda Machine Works Ltd., Aichi, Japan) and flowmeter (T106, Transonic Systems, Inc., Ithaca, NY, USA) were used. The priming volume was 500 mL. At a rotational speed of 2780 rpm, the pump circulated the blood at a flow rate of 1 L/min against a head pressure of 120 mm Hg. Because the aim of the experiment was to image thrombus formation in the rotary blood pump, the activated clotting time (ACT) of the blood was intentionally reduced by adding 2% w/v calcium chloride solution (Otsuka Calcium Chloride Injection 2%, Otsuka Pharmaceutical Co., Ltd., Tokyo, Japan). The initial ACT value was 197 ± 48 s. The ACT value was reduced and the final value was adjusted to ≤130 s within the experimental time from 60 to 120 min. Consequently, the final concentration of the calcium chloride solution was 0.15 ± 0.04 w/v%. When the flow rate became less than 0.8 L/min, 3 mL heparin sodium (1000 units/ mL; Heparin Sodium Injection-Wf, Mitsubishi Pharma Corporation, Osaka, Japan) was injected into the circuit immediately to stop thrombus formation, and the experiments were terminated. The mean time until termination was 95 ± 24 min. After the experiments, the pump was flushed with saline, and the thrombus remaining in the pump was confirmed. Optical setup The optical imaging setup is shown in Fig. 1. Two halogen lights (Megalight100, Schott Moritex, Inc., Saitama, Japan) with the wavelength of emitted light ranging from 400 to 800 nm were used as the light sources for transmitted- and reflected-light scattering imaging, respectively. For transmitted-light imaging,
OPTICAL IMAGING OF THROMBUS FORMATION
735
two-dimensional spectroscopy. In this study, the ability to detect the thrombogenic process in the pump was investigated by the HSI analyses explained below.
FIG. 1. Optical setup of HSI in the rotary blood pump.
the ring light guide (inner diameter = 18 mm, outer diameter = 38 mm; Schott Moritex, Inc.) was attached to the top housing. For reflected-light imaging, a light beam with a diameter of 12 mm was emitted from the light guide. While a majority of the light passed through the bottom housing, some light was partially reflected on the housing surface. The light reflected by the housing surface was filtered using the polarizer. The forward- and backward-scattering light was imaged by the HSI system (HSi-300, Gooch & Housego, Inc., Ilminster, UK). The HSI system obtains spectral images of contiguous wide bands. Therefore, the technique produces images similar to
Real-time imaging of thrombogenic process via stroboscopic effect The HSI system extracted a single image of 700-nm light at a bandwidth of 27.3 nm from the forwardand backward-scattered light. To achieve real-time imaging of the rotating impeller, the imaging cycle was synchronized with the rotational frequency of the pump (2780 rpm = 46.3 Hz). The exposure time was 1 ms and the imaging resolution was 125 × 125 pixels. Stationary images of both forward- and backward-scattered light obtained by the stroboscopic effect were analyzed. Hyperspectral imaging of thrombus formation in a rotary blood pump To investigate the optical properties of a thrombus in the near-infrared wavelength region, HSI was conducted at wavelengths ranging from 600 to 750 nm at increments of 2 nm. The imaging was conducted at rotational speeds of 0 and 2780 rpm before the injection of calcium chloride solution and after the injection of 3000 units of heparin sodium. At 0 rpm, to achieve the best HSI performance, the imaging resolution was 1002 × 1002 pixels, the bandwidth was 2.6 nm, and the exposure time was 200 ms. At 2780 rpm, to investigate the feasibility of dynamically imaging the thrombogenic process, the imaging was
FIG. 2. The forward and backward scattering images (FSI and BSI) at 700 nm.
Artif Organs, Vol. 38, No. 9, 2014
736
D. SAKOTA ET AL.
FIG. 3. Changes in the brightness value of FSI during in vitro antithrombogenic testing.
conducted under the same conditions used for realtime imaging. The results were compared with the thrombus remaining in the pump after being flushed with saline. RESULTS Detection of thrombus formation Figure 2 shows an example of changes in the stroboscopic image during the experiments. In the forward scattering image (FSI), the imaging area was mainly restricted to the vanes because the permanent magnet ring embedded in the impeller blocked the
forward light. In contrast, the backward scattering image (BSI) showed the entire bottom area, including the rotating impeller, vanes, shroud patterns, and blood stream patterns of the outer circumferential region of the impeller. There were no changes in either FSI or BSI for the first 95 min. However, after 96 min, although the pump flow rate remained constant, the FSI began to flicker. Conversely, the contrast of the backward scattering image decreased and the front edges of the vanes became clearer. After 98 min, the brightness of FSI apparently began to increase, as shown in Fig. 3, and the BSI became blurry.
FIG. 4. (a) FSI and BSI by HSI before the injection of calcium chloride solution. (b) FSI and BSI by HSI after the termination of the experiment.
Artif Organs, Vol. 38, No. 9, 2014
OPTICAL IMAGING OF THROMBUS FORMATION
737
FIG. 5. (a) Light spectrum of FSI and BSI in ROI1, ROI2, and ROI3 shown in Fig. 4a. (b) Light spectrum of FSI and BSI in ROI1, ROI2, and ROI3 shown in Fig. 4b.
Spectral change with respect to thrombus formation The FSI and BSI shown in Fig. 4a are the total spectral images before the injection of calcium chloride solution. The intensity at each pixel I was calculated using the following equation.
I=
IS IR
(1)
where IS is the sample image value and IR is the reference image value obtained from the
FIG. 6. (a) FSI by HSI of thrombus remaining in the pump after being flushed with saline. (b) RGB color picture of the thrombus formation. (c) The thrombus mainly formed within the inner blood flow path of the impeller, and a very thin, membranous thrombus formed on the shroud. (d) Picture of the thrombus. Artif Organs, Vol. 38, No. 9, 2014
738
D. SAKOTA ET AL.
transmittance or reflectance image on a polycarbonate plate. As shown in Fig. 4b, the image changed after the experiment was terminated, which indicates that the spectral gravity shifted to a longer wavelength. Figure 5 shows the mean spectral values in three different regions of interest (ROIs): ROI1, ROI2, and ROI3. The transmitted light intensity was especially increased at wavelengths ≥670 nm. On the other hand, the reflected light intensity decreased, and the spectral gravity shifted to a longer wavelength. The index of the spectral shift is defined as τ, which is the wavelength when its intensity equals the intensity at 750 nm. Before the experiments, τFSI and τBSI were 648 nm and 644 nm, respectively. Both values were increased in all ROIs. Figure 6 shows images of the thrombus that remained in the pump after flushing with saline. The thrombus mainly formed within the inner blood flow path of the impeller. Moreover, a very thin, membranous thrombus formed on the shroud. Figure 7 shows the τ level distribution in the BSI. The white solid line and dotted line indicate the visualized thick and transparent thrombus shapes, respectively, fitted with the imaged shape. The results indicated that the τ level increased within the area of the thrombus. As a result of the dynamic HSI at 2780 rpm, the impeller rotated 343 ± 3.51° during capturing of the respective wavelength image. Therefore, to obtain images of the same impeller phase, two images at 680 and 780 nm were selected for the thrombus imaging. The rotational phase difference between 680 and 780 nm was 2.84°, so the image at 680 nm was rotated by 2.84° in the impeller area by affine transformation. Finally, image processing was conducted using the following equation:
IT =
I 720 nm I 680 nm
(2)
where IT is the threshold value, and I680 nm and I720 nm are the intensities calculated by Eq. 1. The image binarization was conducted with IT = 1. Figure 8 shows the binarized images before the injection of calcium chloride solution and after the injection of heparin sodium, respectively. The white dots indicate areas where IT > 1.0. DISCUSSION We successfully detected and imaged thrombus formation noninvasively in a hydrodynamically levitated centrifugal blood pump. The spectrum of noncoagulated blood was especially changed by thrombus formation in the near-infrared wavelength range from 670 to 750 nm. This wavelength range is Artif Organs, Vol. 38, No. 9, 2014
FIG. 7. (a) τBSI distribution in BSI shown in Fig. 6a. τBSI < 644 nm in black area. (b) τBSI distribution in BSI shown in Fig. 6b.
optimal for detecting and imaging thrombus. The spectrum of normal whole blood shown in Fig. 5 can be explained by the absorption spectrum of an oxyhemoglobin molecule (6). The spectrum of whole blood correlated with the absorption curve; therefore, it was considered that the obtained spectrum was primarily affected by the light absorption of oxyhemoglobin. However, the thrombus would reduce light scattering in the wavelength range of 670 to 750 nm. This property induces the shift of the spectral gravity to a longer wavelength and is a unique effect that can be utilized in the imaging of thrombus formation. Based on this knowledge, we propose dual-wavelength imaging of thrombus formation, as
OPTICAL IMAGING OF THROMBUS FORMATION
739
FIG. 8. (a) Images before the injection of calcium chloride solution. (b) Images after the termination of the experiment. The left pictures are images of the rotating pump at 680 nm. The middle pictures are images of the rotating pump at 720 nm. The right pictures are the images binarized by the index IT shown in Eq. 2.
shown in Fig. 8. The light scattered by the thrombus strongly increased the ratio of the 720-nm light compared with the 680-nm light. As a result, a relative difference in the light intensity between the two wavelengths was produced in the area of thrombus formation during high-speed rotation of the pump. Evaluation of the antithrombogenicity of mechanical circulatory support devices is required. In most cases, animal experiments using a calf or goat have been conducted by designing and prototyping different pumps (21,22). Our group proposed a simple in vitro antithrombogenic testing method using a mock circulation system (18). However, dynamic evaluation of thrombus formation in the devices has been difficult because the inside of the devices is covered by whole blood. As shown in Fig. 3, near-infrared light is able to detect the start of thrombus formation at blood layer thicknesses of 5.2 mm. Recently, extracorporeal circulation therapies extending for several weeks and months have been reported for bridge to transplantation (23) and bridge to decision (24). The noninvasive and continuous monitoring of blood conditions contributes to the safety and reliability of long-term circulation support systems. Because of scattering and absorption problems, optical techniques to monitor blood conditions have limited penetration depth and are unable to reach locations deep within extracorporeal circulation devices. However, recent contact-free pumps have become simpler and more compact. The develop-
ment of simpler and more compact devices would provide better opportunities to use optical diagnostic techniques. The present study indicated the feasibility of imaging the blood flow stream and thrombus formation in a rotary blood pump by HSI. Recently, computational modeling of thrombosis to optimize the design of cardiovascular devices has been developed (25). The HSI could be used in conjunction with the simulation technique to develop more complex devices. Moreover, HSI can be used for precise quantification of hemoglobin concentration noninvasively by obtaining the consecutive light spectrum. In the present study, a red thrombus formed in the rotary blood pump. The hemoglobin concentration in thrombus depends on the surrounding flow conditions. In some areas of high shear stress, a white thrombus without red blood cells is formed (26). Therefore, it is speculated that quantifying the hemoglobin concentration in a thrombus can reveal information about the cause of thrombus formation. Future studies will examine whether HSI can be used to evaluate not only the quantity but also the quality of thrombi that form in mechanical circulatory support devices. CONCLUSION Application of hyperspectral imaging to noninvasive imaging of thrombus formation in a rotary blood pump is proposed. The optical properties of the Artif Organs, Vol. 38, No. 9, 2014
740
D. SAKOTA ET AL.
thrombus mainly changed in the near-infrared wavelength range from 670 to 750 nm. The detection and imaging of thrombus formation in rotary blood pumps is feasible using near-infrared light. Acknowledgment: This work was supported by a Grant-in-Aid for Young Scientists (B) (#23700575, recipient Daisuke Sakota, April 2011 to present) from the Japan Society for the Promotion of Science.
REFERENCES 1. O’Brien JJ, Butterworth J, Hammon JW, Morris KJ, Phillips JM, Stump DA. Cerebral emboli during cardiac surgery in children. Anesthesiology 1997;87:1063–9. 2. Braekken SK, Russell D, Brucher R, Abdelnoor M, Svennevig JL. Cerebral microembolic signals during cardiopulmonary bypass surgery. Frequency, time of occurrence, and association with patient and surgical characteristics. Stroke 1997;28:1988–92. 3. Spoelhof GD, Farchmin CJ. Renal emboli and atrial septal aneurysm. J Fam Pract 1996;42:519–22. 4. Bahrmann P, Werner GS, Heusch G, et al. Detection of coronary microembolization by Doppler ultrasound in patients with stable angina pectoris undergoing elective percutaneous coronary interventions. Circulation 2007;115:600–8. 5. Russell D, Madden KP, Clark WM, et al. Detection of arterial emboli using Doppler ultrasound in rabbits. Stroke 1991;22: 253–8. 6. Horecker BL. The absorption spectra of hemoglobin and its derivatives in the visible and near infrared regions. J Biol Chem 1943;48:173–83. 7. Solen KA, Mohammad SF, Burns GL, et al. Markers of thromboembolization in a bovine ex vivo left ventricular assist device model. ASAIO J 1994;40:M602–8. 8. Yasuda T, Sekimoto K, Taga I, Funakubo A, Fukui Y, Takatani S. New method for the detection of thrombus formation in cardiovascular devices: optical sensor evaluation in a flow chamber model. ASAIO J 2005;51:110–5. 9. Sankai Y, Tsutui T, Jikuya T, Shigeta O, Ohta M, Mitsui T. Method of noninvasive and continuous hemolysis/thrombogenesis measurement by laser photometry during artificial heart development. ASAIO J 1997;43:M682–6. 10. Oshima S, Sankai Y. Development of optical sensing system for noninvasive and dynamic monitoring of thrombogenic process. ASAIO J 2010;56:460–7. 11. Reynolds LO, Solen KA, Mohammad SF, Pantalos GM, Kim J. Differential light scattering cuvettes for the measurement of thromboemboli in high shear blood flow systems. ASAIO Trans 1990;36:M185–8.
Artif Organs, Vol. 38, No. 9, 2014
12. Sukavaneshvar S, Zheng Y, Rosa GM, et al. Thromboembolization associated with sudden increases in flow in a coronary stent ex vivo shunt model. ASAIO J 2000;46:301–4. 13. Solen K, Sukavaneshvar S, Zheng Y, et al. Light-scattering instrument to detect thromboemboli in blood. J Biomed Opt 2003;8:70–9. 14. Zuzak KJ, Schaeberle MD, Lewis EN, Levin IW. Visible reflectance hyperspectral imaging: characterization of a noninvasive, in vivo system for determining tissue perfusion. Anal Chem 2002;74:2021–8. 15. Khoobehi B, Beach JM, Kawano H. Hyperspectral imaging for measurement of oxygen saturation in the optic nerve head. Invest Ophthalmol Vis Sci 2004;45:1464–72. 16. Sorg BS, Moeller BJ, Donovan O, Cao Y, Dewhirst MW. Hyperspectral imaging of hemoglobin saturation in tumor microvasculature and tumor dypoxia development. J Biomed Opt 2005;10:44004. 17. Roggan A, Friebel M, Dorschel K, Hahn A, Muller G. Optical properties of circulating human blood in the wavelength range 400–2500 nm. J Biomed Opt 1999;4:36–46. 18. Maruyama O, Tomari Y, Sugiyama D, et al. Simple in vitro testing method for antithrombogenic evaluation of centrifugal blood pumps. ASAIO J 2009;55:314–22. 19. Kosaka R, Yada T, Nishida M, Maruyama O, Yamane T. Geometric optimization of a step bearing for a hydrodynamically levitated centrifugal blood pump for the reduction of hemolysis. Artif Organs 2013;37:778–85. 20. Yasui K, Kosaka R, Nishida M, Maruyama O, Kawaguchi Y, Yamane T. Optimal design of the hydrodynamic multi-arc bearing in a centrifugal blood pump for the improvement of bearing stiffness and hemolysis level. Artif Organs 2013;37:768–77. 21. Nagaoka E, Someya T, Kitao T, et al. Development of a disposable magnetically levitated centrifugal blood pump (MedTech Dispo) intended for bridge-to-bridge applications—two-week in vivo evaluation. Artif Organs 2010;34:778–83. 22. Ando Y, Kitao T, Nagaoka E, et al. One-month biocompatibility evaluation of the pediatric TinyPump in goats. Artif Organs 2011;35:813–8. 23. Barbone A, Malvindi PG, Sorabella RA, et al. 6 months of “temporary” support by Levitronix left ventricular assist device. Artif Organs 2012;30:639–42. 24. Russo CF, Cannata A, Lanfranconi M, et al. Veno-arterial extracorporeal membrane oxygenation using Levitronix centrifugal pump as bridge to decision for refractory cardiogenic shock. J Thorac Cardiovasc Surg 2010;140:1416–21. 25. Moiseyev G, Bar-Yoseph PZ. Computational modeling of thrombosis as a tool in the design and optimization of vascular implants. J Biomech 2013;46:248–52. 26. Mizuno T, Sugimoto M, Matsui H, Hamada M, Shida Y, Yoshioka A. Visual evaluation of blood coagulation during mural thrombogenesis under high shear blood flow. Thromb Res 2008;121:855–64.
Copyright © 2014 International Center for Artificial Organs and Transplantation and Wiley Periodicals, Inc.
Feasibility of the Optical Imaging of Thrombus Formation in a Rotary Blood Pump by Near-Infrared Light *Daisuke Sakota, †Tomotaka Murashige, *Ryo Kosaka, *Masahiro Nishida, and *Osamu Maruyama *National Institute of Advanced Industrial Science and Technology, Tsukuba; and †Graduate School of Science and Technology, Tokyo University of Science, Noda, Japan
Abstract: Blood coagulation is one of the primary concerns when using mechanical circulatory support devices such as blood pumps. Noninvasive detection and imaging of thrombus formation is useful not only for the development of more hemocompatible devices but also for the management of blood coagulation to avoid risk of infarction. The objective of this study is to investigate the use of near-infrared light for imaging of thrombus formation in a rotary blood pump. The optical properties of a thrombus at wavelengths ranging from 600 to 750 nm were analyzed using a hyperspectral imaging (HSI) system. A specially designed hydrodynamically levitated centrifugal blood pump with a visible bottom area was used. In vitro antithrombogenic testing was conducted five times with the pump using bovine whole blood in which the activated
blood clotting time was adjusted to 200 s prior to the experiment. Two halogen lights were used for the light sources. The forward scattering through the pump and backward scattering on the pump bottom area were imaged using the HSI system. HSI showed an increase in forward scattering at wavelengths ranging from 670 to 750 nm in the location of thrombus formation. The time at which the thrombus began to form in the impeller rotating at 2780 rpm could be detected. The spectral difference between the whole blood and the thrombus was utilized to image thrombus formation. The results indicate the feasibility of dynamically detecting and imaging thrombus formation in a rotary blood pump. Key Words: Hemocompatibility—Hyperspectral imaging—Nearinfrared light—Rotary blood pump—Thrombus.
Blood coagulation is one of the primary concerns when using mechanical circulatory support devices such as blood pumps and oxygenators, from both research-and-development and clinical-use perspectives. Improved techniques for detecting or imaging thrombus formation will contribute enormously not only to the development of more hemocompatible devices but also to the management of blood coagulation to obviate cerebral (1,2) or renal (3) infarction in patients requiring mechanical circulatory support.
Ultrasound techniques to detect coronary or arterial microemboli have been proposed (4,5). However, the proposed methods are inadequate because ultrasound does not penetrate the rigid surface of mechanical devices. Near-infrared light has been used as a noninvasive diagnostic tool for blood. Light attenuation in blood results from both scattering and absorption; however, at near-infrared wavelengths from 700 to 900 nm, the light absorption of hemoglobin and water decreases (6). Therefore, the light can be used to obtain information from blood and deeper tissues. Light-scattering microemboli detectors (LSMDs), which apply a near-infrared laser or a light-emitting diode at wavelengths of 624 (7), 632 (8), 805 (9), 810 (10), or 830 (7,11–13) nm, can detect microthrombi flowing in the extracorporeal circuit. However, no spectroscopic analytical method has been investigated. In addition, as these LSMDs obtain a single-point measurement, they are not applicable for detecting and imaging thrombus formation within the pump.
doi:10.1111/aor.12377 Received February 2014; revised June 2014. Address correspondence and reprint requests to Dr. Daisuke Sakota, Artificial Organ Group, Human Technology Research Institute, National Institute of Advanced Industrial Science and Technology, Namiki 1-2-1 Tsukuba, Ibaraki 305-8564, Japan. E-mail: [email protected] Presented in part at the 21st Congress the International Society for Rotary Blood Pumps held September 26–28, 2013 in Yokohama, Japan.
Artificial Organs 2014, 38(9):733–740
734
D. SAKOTA ET AL.
To further advance the evaluation of device hemocompatibility and techniques for diagnosing risk of thrombosis, this study proposes hyperspectral imaging (HSI) of a rotary blood pump. Although multispectral imaging deals with several images at discrete bands, HSI produces contiguous spectral images at narrow spectral bandwidths. In biomedical optics, HSI has mainly been applied to quantitative imaging of the oxygen saturation and hemoglobin concentration in the vascular network of tissues or tumors (14–16). The optical properties of blood mainly depend on physiological and biochemical parameters such as hematocrit, flow, osmolarity, hemolysis, and oxygen saturation (17), as these parameters are significantly affected by red blood cells (RBCs), which are the main source of light scattering and absorption in blood. However, little is known about the spectral difference between normal whole blood and thrombus. The objective of this study is to develop an optical imaging system to visualize thrombus formation in a rotary blood pump during in vitro antithrombogenic testing (18). To achieve this objective, the difference in optical characteristics between normal whole blood and thrombus was investigated using an HSI system in the wavelength range from 600 to 750 nm. Individual wavelength images obtained by the HSI system were analyzed, and the optimal wavelength to obtain the clearest image of thrombus formation was determined. MATERIALS AND METHODS In vitro antithrombogenic testing In vitro antithrombogenic testing as proposed by Maruyama et al. (18) was conducted five times using bovine whole blood (Funakoshi Co., Ltd., Tokyo, Japan). The blood was mixed with anticoagulant (100 mL sodium citrate/1000 mL blood) immediately after being collected at a slaughterhouse. The hematocrit value from the five tests was 35.4 ± 1.5% (mean ± SD). Prior to the experiment, the oxygen saturation of the blood was adjusted to 100% using an oxygenator and a rotary blood pump to unify and maintain the optical absorption characteristics of blood during all experiments. The oxygenated blood was then added to the closed-loop extracorporeal circulation circuit. The test pump was a hydrodynamically levitated centrifugal blood pump developed by our group. The basic structure and levitation principle were very similar to those of a previously reported pump (19,20). This pump comprises three parts: the top casing, the bottom casing, and the impeller. The stator coil is assembled in the top Artif Organs, Vol. 38, No. 9, 2014
casing. Therefore, the bottom of the fabricated pump is visible. The inner blood layer thickness of the impeller, the thickness of six vanes, is 4.9 mm. The hydrodynamic force generated on the shroud with the spiral groove attached to the vanes levitates the impeller. Before the study, the location of the levitated impeller was confirmed by a mock circulation loop experiment using a 38% (wt/wt) glycerol–water solution as the working fluid. Results from the mock circulation loop experiment indicated a bottom bearing gap of 72 μm at a rotational speed of 2780 rpm, while the top bearing gap was 262 μm. The pump and a heparin-coated 500-mL volumetric blood reservoir (specially ordered products; Senko Medical Instrument Mfg. Co., Ltd., Tokyo, Japan) were connected with a heparin-coated polyvinyl chloride tube (MERA Exceline-H 3/8 × 3/32, Senko Medical Instrument Mfg. Co., Ltd.), which included heparin-coated sampling ports (3/8 × 3/8, Senko Medical Instrument Mfg. Co., Ltd). A pressure gauge (Toyoda Machine Works Ltd., Aichi, Japan) and flowmeter (T106, Transonic Systems, Inc., Ithaca, NY, USA) were used. The priming volume was 500 mL. At a rotational speed of 2780 rpm, the pump circulated the blood at a flow rate of 1 L/min against a head pressure of 120 mm Hg. Because the aim of the experiment was to image thrombus formation in the rotary blood pump, the activated clotting time (ACT) of the blood was intentionally reduced by adding 2% w/v calcium chloride solution (Otsuka Calcium Chloride Injection 2%, Otsuka Pharmaceutical Co., Ltd., Tokyo, Japan). The initial ACT value was 197 ± 48 s. The ACT value was reduced and the final value was adjusted to ≤130 s within the experimental time from 60 to 120 min. Consequently, the final concentration of the calcium chloride solution was 0.15 ± 0.04 w/v%. When the flow rate became less than 0.8 L/min, 3 mL heparin sodium (1000 units/ mL; Heparin Sodium Injection-Wf, Mitsubishi Pharma Corporation, Osaka, Japan) was injected into the circuit immediately to stop thrombus formation, and the experiments were terminated. The mean time until termination was 95 ± 24 min. After the experiments, the pump was flushed with saline, and the thrombus remaining in the pump was confirmed. Optical setup The optical imaging setup is shown in Fig. 1. Two halogen lights (Megalight100, Schott Moritex, Inc., Saitama, Japan) with the wavelength of emitted light ranging from 400 to 800 nm were used as the light sources for transmitted- and reflected-light scattering imaging, respectively. For transmitted-light imaging,
OPTICAL IMAGING OF THROMBUS FORMATION
735
two-dimensional spectroscopy. In this study, the ability to detect the thrombogenic process in the pump was investigated by the HSI analyses explained below.
FIG. 1. Optical setup of HSI in the rotary blood pump.
the ring light guide (inner diameter = 18 mm, outer diameter = 38 mm; Schott Moritex, Inc.) was attached to the top housing. For reflected-light imaging, a light beam with a diameter of 12 mm was emitted from the light guide. While a majority of the light passed through the bottom housing, some light was partially reflected on the housing surface. The light reflected by the housing surface was filtered using the polarizer. The forward- and backward-scattering light was imaged by the HSI system (HSi-300, Gooch & Housego, Inc., Ilminster, UK). The HSI system obtains spectral images of contiguous wide bands. Therefore, the technique produces images similar to
Real-time imaging of thrombogenic process via stroboscopic effect The HSI system extracted a single image of 700-nm light at a bandwidth of 27.3 nm from the forwardand backward-scattered light. To achieve real-time imaging of the rotating impeller, the imaging cycle was synchronized with the rotational frequency of the pump (2780 rpm = 46.3 Hz). The exposure time was 1 ms and the imaging resolution was 125 × 125 pixels. Stationary images of both forward- and backward-scattered light obtained by the stroboscopic effect were analyzed. Hyperspectral imaging of thrombus formation in a rotary blood pump To investigate the optical properties of a thrombus in the near-infrared wavelength region, HSI was conducted at wavelengths ranging from 600 to 750 nm at increments of 2 nm. The imaging was conducted at rotational speeds of 0 and 2780 rpm before the injection of calcium chloride solution and after the injection of 3000 units of heparin sodium. At 0 rpm, to achieve the best HSI performance, the imaging resolution was 1002 × 1002 pixels, the bandwidth was 2.6 nm, and the exposure time was 200 ms. At 2780 rpm, to investigate the feasibility of dynamically imaging the thrombogenic process, the imaging was
FIG. 2. The forward and backward scattering images (FSI and BSI) at 700 nm.
Artif Organs, Vol. 38, No. 9, 2014
736
D. SAKOTA ET AL.
FIG. 3. Changes in the brightness value of FSI during in vitro antithrombogenic testing.
conducted under the same conditions used for realtime imaging. The results were compared with the thrombus remaining in the pump after being flushed with saline. RESULTS Detection of thrombus formation Figure 2 shows an example of changes in the stroboscopic image during the experiments. In the forward scattering image (FSI), the imaging area was mainly restricted to the vanes because the permanent magnet ring embedded in the impeller blocked the
forward light. In contrast, the backward scattering image (BSI) showed the entire bottom area, including the rotating impeller, vanes, shroud patterns, and blood stream patterns of the outer circumferential region of the impeller. There were no changes in either FSI or BSI for the first 95 min. However, after 96 min, although the pump flow rate remained constant, the FSI began to flicker. Conversely, the contrast of the backward scattering image decreased and the front edges of the vanes became clearer. After 98 min, the brightness of FSI apparently began to increase, as shown in Fig. 3, and the BSI became blurry.
FIG. 4. (a) FSI and BSI by HSI before the injection of calcium chloride solution. (b) FSI and BSI by HSI after the termination of the experiment.
Artif Organs, Vol. 38, No. 9, 2014
OPTICAL IMAGING OF THROMBUS FORMATION
737
FIG. 5. (a) Light spectrum of FSI and BSI in ROI1, ROI2, and ROI3 shown in Fig. 4a. (b) Light spectrum of FSI and BSI in ROI1, ROI2, and ROI3 shown in Fig. 4b.
Spectral change with respect to thrombus formation The FSI and BSI shown in Fig. 4a are the total spectral images before the injection of calcium chloride solution. The intensity at each pixel I was calculated using the following equation.
I=
IS IR
(1)
where IS is the sample image value and IR is the reference image value obtained from the
FIG. 6. (a) FSI by HSI of thrombus remaining in the pump after being flushed with saline. (b) RGB color picture of the thrombus formation. (c) The thrombus mainly formed within the inner blood flow path of the impeller, and a very thin, membranous thrombus formed on the shroud. (d) Picture of the thrombus. Artif Organs, Vol. 38, No. 9, 2014
738
D. SAKOTA ET AL.
transmittance or reflectance image on a polycarbonate plate. As shown in Fig. 4b, the image changed after the experiment was terminated, which indicates that the spectral gravity shifted to a longer wavelength. Figure 5 shows the mean spectral values in three different regions of interest (ROIs): ROI1, ROI2, and ROI3. The transmitted light intensity was especially increased at wavelengths ≥670 nm. On the other hand, the reflected light intensity decreased, and the spectral gravity shifted to a longer wavelength. The index of the spectral shift is defined as τ, which is the wavelength when its intensity equals the intensity at 750 nm. Before the experiments, τFSI and τBSI were 648 nm and 644 nm, respectively. Both values were increased in all ROIs. Figure 6 shows images of the thrombus that remained in the pump after flushing with saline. The thrombus mainly formed within the inner blood flow path of the impeller. Moreover, a very thin, membranous thrombus formed on the shroud. Figure 7 shows the τ level distribution in the BSI. The white solid line and dotted line indicate the visualized thick and transparent thrombus shapes, respectively, fitted with the imaged shape. The results indicated that the τ level increased within the area of the thrombus. As a result of the dynamic HSI at 2780 rpm, the impeller rotated 343 ± 3.51° during capturing of the respective wavelength image. Therefore, to obtain images of the same impeller phase, two images at 680 and 780 nm were selected for the thrombus imaging. The rotational phase difference between 680 and 780 nm was 2.84°, so the image at 680 nm was rotated by 2.84° in the impeller area by affine transformation. Finally, image processing was conducted using the following equation:
IT =
I 720 nm I 680 nm
(2)
where IT is the threshold value, and I680 nm and I720 nm are the intensities calculated by Eq. 1. The image binarization was conducted with IT = 1. Figure 8 shows the binarized images before the injection of calcium chloride solution and after the injection of heparin sodium, respectively. The white dots indicate areas where IT > 1.0. DISCUSSION We successfully detected and imaged thrombus formation noninvasively in a hydrodynamically levitated centrifugal blood pump. The spectrum of noncoagulated blood was especially changed by thrombus formation in the near-infrared wavelength range from 670 to 750 nm. This wavelength range is Artif Organs, Vol. 38, No. 9, 2014
FIG. 7. (a) τBSI distribution in BSI shown in Fig. 6a. τBSI < 644 nm in black area. (b) τBSI distribution in BSI shown in Fig. 6b.
optimal for detecting and imaging thrombus. The spectrum of normal whole blood shown in Fig. 5 can be explained by the absorption spectrum of an oxyhemoglobin molecule (6). The spectrum of whole blood correlated with the absorption curve; therefore, it was considered that the obtained spectrum was primarily affected by the light absorption of oxyhemoglobin. However, the thrombus would reduce light scattering in the wavelength range of 670 to 750 nm. This property induces the shift of the spectral gravity to a longer wavelength and is a unique effect that can be utilized in the imaging of thrombus formation. Based on this knowledge, we propose dual-wavelength imaging of thrombus formation, as
OPTICAL IMAGING OF THROMBUS FORMATION
739
FIG. 8. (a) Images before the injection of calcium chloride solution. (b) Images after the termination of the experiment. The left pictures are images of the rotating pump at 680 nm. The middle pictures are images of the rotating pump at 720 nm. The right pictures are the images binarized by the index IT shown in Eq. 2.
shown in Fig. 8. The light scattered by the thrombus strongly increased the ratio of the 720-nm light compared with the 680-nm light. As a result, a relative difference in the light intensity between the two wavelengths was produced in the area of thrombus formation during high-speed rotation of the pump. Evaluation of the antithrombogenicity of mechanical circulatory support devices is required. In most cases, animal experiments using a calf or goat have been conducted by designing and prototyping different pumps (21,22). Our group proposed a simple in vitro antithrombogenic testing method using a mock circulation system (18). However, dynamic evaluation of thrombus formation in the devices has been difficult because the inside of the devices is covered by whole blood. As shown in Fig. 3, near-infrared light is able to detect the start of thrombus formation at blood layer thicknesses of 5.2 mm. Recently, extracorporeal circulation therapies extending for several weeks and months have been reported for bridge to transplantation (23) and bridge to decision (24). The noninvasive and continuous monitoring of blood conditions contributes to the safety and reliability of long-term circulation support systems. Because of scattering and absorption problems, optical techniques to monitor blood conditions have limited penetration depth and are unable to reach locations deep within extracorporeal circulation devices. However, recent contact-free pumps have become simpler and more compact. The develop-
ment of simpler and more compact devices would provide better opportunities to use optical diagnostic techniques. The present study indicated the feasibility of imaging the blood flow stream and thrombus formation in a rotary blood pump by HSI. Recently, computational modeling of thrombosis to optimize the design of cardiovascular devices has been developed (25). The HSI could be used in conjunction with the simulation technique to develop more complex devices. Moreover, HSI can be used for precise quantification of hemoglobin concentration noninvasively by obtaining the consecutive light spectrum. In the present study, a red thrombus formed in the rotary blood pump. The hemoglobin concentration in thrombus depends on the surrounding flow conditions. In some areas of high shear stress, a white thrombus without red blood cells is formed (26). Therefore, it is speculated that quantifying the hemoglobin concentration in a thrombus can reveal information about the cause of thrombus formation. Future studies will examine whether HSI can be used to evaluate not only the quantity but also the quality of thrombi that form in mechanical circulatory support devices. CONCLUSION Application of hyperspectral imaging to noninvasive imaging of thrombus formation in a rotary blood pump is proposed. The optical properties of the Artif Organs, Vol. 38, No. 9, 2014
740
D. SAKOTA ET AL.
thrombus mainly changed in the near-infrared wavelength range from 670 to 750 nm. The detection and imaging of thrombus formation in rotary blood pumps is feasible using near-infrared light. Acknowledgment: This work was supported by a Grant-in-Aid for Young Scientists (B) (#23700575, recipient Daisuke Sakota, April 2011 to present) from the Japan Society for the Promotion of Science.
REFERENCES 1. O’Brien JJ, Butterworth J, Hammon JW, Morris KJ, Phillips JM, Stump DA. Cerebral emboli during cardiac surgery in children. Anesthesiology 1997;87:1063–9. 2. Braekken SK, Russell D, Brucher R, Abdelnoor M, Svennevig JL. Cerebral microembolic signals during cardiopulmonary bypass surgery. Frequency, time of occurrence, and association with patient and surgical characteristics. Stroke 1997;28:1988–92. 3. Spoelhof GD, Farchmin CJ. Renal emboli and atrial septal aneurysm. J Fam Pract 1996;42:519–22. 4. Bahrmann P, Werner GS, Heusch G, et al. Detection of coronary microembolization by Doppler ultrasound in patients with stable angina pectoris undergoing elective percutaneous coronary interventions. Circulation 2007;115:600–8. 5. Russell D, Madden KP, Clark WM, et al. Detection of arterial emboli using Doppler ultrasound in rabbits. Stroke 1991;22: 253–8. 6. Horecker BL. The absorption spectra of hemoglobin and its derivatives in the visible and near infrared regions. J Biol Chem 1943;48:173–83. 7. Solen KA, Mohammad SF, Burns GL, et al. Markers of thromboembolization in a bovine ex vivo left ventricular assist device model. ASAIO J 1994;40:M602–8. 8. Yasuda T, Sekimoto K, Taga I, Funakubo A, Fukui Y, Takatani S. New method for the detection of thrombus formation in cardiovascular devices: optical sensor evaluation in a flow chamber model. ASAIO J 2005;51:110–5. 9. Sankai Y, Tsutui T, Jikuya T, Shigeta O, Ohta M, Mitsui T. Method of noninvasive and continuous hemolysis/thrombogenesis measurement by laser photometry during artificial heart development. ASAIO J 1997;43:M682–6. 10. Oshima S, Sankai Y. Development of optical sensing system for noninvasive and dynamic monitoring of thrombogenic process. ASAIO J 2010;56:460–7. 11. Reynolds LO, Solen KA, Mohammad SF, Pantalos GM, Kim J. Differential light scattering cuvettes for the measurement of thromboemboli in high shear blood flow systems. ASAIO Trans 1990;36:M185–8.
Artif Organs, Vol. 38, No. 9, 2014
12. Sukavaneshvar S, Zheng Y, Rosa GM, et al. Thromboembolization associated with sudden increases in flow in a coronary stent ex vivo shunt model. ASAIO J 2000;46:301–4. 13. Solen K, Sukavaneshvar S, Zheng Y, et al. Light-scattering instrument to detect thromboemboli in blood. J Biomed Opt 2003;8:70–9. 14. Zuzak KJ, Schaeberle MD, Lewis EN, Levin IW. Visible reflectance hyperspectral imaging: characterization of a noninvasive, in vivo system for determining tissue perfusion. Anal Chem 2002;74:2021–8. 15. Khoobehi B, Beach JM, Kawano H. Hyperspectral imaging for measurement of oxygen saturation in the optic nerve head. Invest Ophthalmol Vis Sci 2004;45:1464–72. 16. Sorg BS, Moeller BJ, Donovan O, Cao Y, Dewhirst MW. Hyperspectral imaging of hemoglobin saturation in tumor microvasculature and tumor dypoxia development. J Biomed Opt 2005;10:44004. 17. Roggan A, Friebel M, Dorschel K, Hahn A, Muller G. Optical properties of circulating human blood in the wavelength range 400–2500 nm. J Biomed Opt 1999;4:36–46. 18. Maruyama O, Tomari Y, Sugiyama D, et al. Simple in vitro testing method for antithrombogenic evaluation of centrifugal blood pumps. ASAIO J 2009;55:314–22. 19. Kosaka R, Yada T, Nishida M, Maruyama O, Yamane T. Geometric optimization of a step bearing for a hydrodynamically levitated centrifugal blood pump for the reduction of hemolysis. Artif Organs 2013;37:778–85. 20. Yasui K, Kosaka R, Nishida M, Maruyama O, Kawaguchi Y, Yamane T. Optimal design of the hydrodynamic multi-arc bearing in a centrifugal blood pump for the improvement of bearing stiffness and hemolysis level. Artif Organs 2013;37:768–77. 21. Nagaoka E, Someya T, Kitao T, et al. Development of a disposable magnetically levitated centrifugal blood pump (MedTech Dispo) intended for bridge-to-bridge applications—two-week in vivo evaluation. Artif Organs 2010;34:778–83. 22. Ando Y, Kitao T, Nagaoka E, et al. One-month biocompatibility evaluation of the pediatric TinyPump in goats. Artif Organs 2011;35:813–8. 23. Barbone A, Malvindi PG, Sorabella RA, et al. 6 months of “temporary” support by Levitronix left ventricular assist device. Artif Organs 2012;30:639–42. 24. Russo CF, Cannata A, Lanfranconi M, et al. Veno-arterial extracorporeal membrane oxygenation using Levitronix centrifugal pump as bridge to decision for refractory cardiogenic shock. J Thorac Cardiovasc Surg 2010;140:1416–21. 25. Moiseyev G, Bar-Yoseph PZ. Computational modeling of thrombosis as a tool in the design and optimization of vascular implants. J Biomech 2013;46:248–52. 26. Mizuno T, Sugimoto M, Matsui H, Hamada M, Shida Y, Yoshioka A. Visual evaluation of blood coagulation during mural thrombogenesis under high shear blood flow. Thromb Res 2008;121:855–64.
