Proceeding of IWCPB-HMF 99 (International Workshop on Chemical, Physical and Biological Processes under High Magnetic Fields), November 24 26, 1999,

Transcription

1 Proceeding of IWCPB-HMF 99 (International Workshop on Chemical, Physical and Biological Processes under High s), November 24 26, 1999, Omiya, Saitama, Japan. Submitted January 27, 2

2 Feasibility of Direct Separation of White Cells and Plasma from Whole Blood Makoto Takayasu 1, David R. Kelland 2, Joseph V. Minervini 1, Fritz J. Friedlaender 3, and Stephen R. Ash 4 1 PSFC, 2 FBML, MIT, Cambridge, MA, USA 3 Purdue University, West Lafayette, IN, USA 4 HemoCleanse, Inc., West Lafayette, IN, USA Abstract Continuous magnetic separation on the basis of high gradient magnetic separation (HGMS) provides a sensitive and selective separation technique. This method is desirable for selective separation for white blood cells and plasma from whole blood. Any additives such as magnetic tagging agents or reducing agents are not required. The oxygenated and deoxygenated states of red blood cells are controlled by a gas-permeable tubular membrane with oxygen (air) and nitrogen gas, respectively. Blood components separated by this method could be returned to the original or another patient directly after treatment. The device will be applicable to clinical usage as well as for diagnostic purposes. cell separations are unique, since the separation is based on the magnetic susceptibility (which depends on the chemical composition) of the cell. separation with a high field superconducting magnet makes possible biological cell separation even if the magnetic susceptibility is on the order of only 1-6. Introduction Blood component transfusion is very important in therapy of various diseases such as anemia, cancer, and immune disease. The principal methods of blood cell component separation for therapeutic purposes are conventional mechanical filtration using fiber and mesh, and centrifugation. The centrifugation method, which has been well developed, is most widely used for various cell separations. Continuous centrifuge, very high-speed ultracentrifugation, density gradient centrifugation and counterflow centrifugation are available for biological cell separations. However, the centrifugal force to which cells are subjected is huge ( times the force of gravity) and it causes various complications in separated cells such as cell damage and aggregation 1, 2. For white cell separation by a centrifuge, several stages of centrifugation or time consuming operation of density gradient centrifugation are required 3. In the late 196's high gradient magnetic separation (HGMS) was introduced 4, 5, using ferromagnetic fine wires placed in a magnetic field to obtain a high field gradient. The fundamental work of HGMS for the removal of red blood cells from whole blood was reported by Melville 6-8, Graham 9, 1, and Owen Most of the work, however, was done with additives to enhance the magnetic state of red blood cells and to obtain effective magnetic separation. Direct magnetic separation of red and white blood cells from whole blood without any additives was confirmed in 1982 in a quasi-static state of very low flow conditions using an optical observation method 14. In those experiments, the magnetic states of the red blood cells were controlled by passing blood through a gas-permeable tubular membrane with oxygen or nitrogen gas on the outside. Blood components separated by this method could be returned to the original or another patient directly after treatment. A continuous magnetic separation method has been developed on the basis of high gradient magnetic separation The separator does not trap particles in the filter, but separates continuously during passage through a separation channel in a magnetic field. The continuous magnetic separator is capable of separating even very weakly and very small magnetic particles. This method is very useful for biological cell separations. Continuous direct red blood cell separation from whole blood by combining the gas-

3 permeable membrane and continuous separation techniques has been performed 19. Here we review these previous works and discuss the feasibility of magnetic separation of white blood cells and plasma from whole blood. Direct Separation of Red Blood Cells 14 A schematic of the direct magnetic separation apparatus for blood cells using a gaspermeable tubular membrane is shown in Fig. 1. Whole blood is fed to a magnetic separation chamber through a gas-permeable membrane tube of silicon tubing (.9 m long,.5 mm ID and.9 mm OD, Silastic, Dow Corning Corporation). The oxygenated and deoxygenated states of red blood cells are obtained during passage through a gas-permeable tubing with oxygen (air) or nitrogen gas, respectively. Any additives such as magnetic tagging agents or reducing agents are not required. This process of oxygenation is similar to that of the human body. By using an HGMS observation chamber made of a very thin rectangular glass tube (.1 mm x 1 mm cross-section), magnetic susceptibilities of blood cells and blood cell separation processes in whole blood have been investigated 14. Table 1 shows the magnetic susceptibilities of blood cells and plasma by a trajectory method. The magnetic susceptibility of plasma was measured by trajectory analysis of a H 2 gas bubble in plasma. As seen in the table, red blood cells of both oxygenated and deoxygenated states are themselves diamagnetic. However, the deoxygenated red blood cells in plasma are paramagnetic. The relative magnetic susceptibility of +3.9 x 1-6 (SI) for fully reduced red blood cells has been reported by Melville 8. Oxygen or Nitrogen Gas Blood Gas Permeable Membrane Tube Separation Observation Chamber Ferromagnetic Wire FIG. 1. Schematic of the direct magnetic separation apparatus for blood cells using a gas-permeable tube with an HGMS observation chamber. 14 TABLE 1 Susceptibilities of Blood Cells and Plasma. 14 Deoxygenated Red Blood Cells Susceptibility (SI units) χ p or χ f χ p χ f in Plasma -3.8 x 1-6 Diamagnetic +3.5 x 1-6 (+3.1 to +3.9 x 1-6 ) +3.9 x 1-6 calculated 8 Paramagnetic Diamagnetic Oxygenated Diamagnetic Red Blood Cells White Blood Cells Diamagnetic Diamagnetic Plasma -7.7 x 1-6 Diamagnetic -

4 Fig. 2 shows collected blood cells on a ferromagnetic nickel wire of 1 µm diameter in a quasi-static state of very low flow conditions using an optical observation method. Fig. 2(a) shows deoxygenated red blood cells in whole blood. Red blood cells are collected in the paramagnetic capture sites, that is, on both sides of the magnetic wire in the direction of the magnetic field. The paramagnetic blood cells are repelled from a region on the top edge of the wire. Buildup behaviors of oxygenated red blood cells and white blood cells are shown in Figs. 2(b) and (c), respectively. Both collection profiles are typical diamagnetic capture where particles are repelled from the sides of the magnetic wire in the direction of the magnetic field, and attracted to the front and back sides and on the edge of the wire. The experimental observations show clearly a paramagnetic separation of deoxygenated red cells and diamagnetic separations of oxygenated red cells and white cells in whole blood. (a) Deoxygenated Red Blood Cells (b) Oxygenated Red Blood Cells FIG. 2. Blood cell collections on a 1 µm Ni wire in a magnetic field of 1.25 T. 14 (a) Paramagnetic collection of deoxygenated red blood cells. (b) Diamagnetic collection of oxygenated red blood cells. (c) White blood cells, which show the typical diamagnetic buildup profile as seen in (b). (c) White Blood Cells S INLET x L FLOW CELL X3 X2 X1 X a z OUTLET #3 #2 #1 MAGNETIC FIELD FLOW CELL P D GRAVITY MAGNETIC WIRE MAGNETIC WIRE (a) PARAMAGNETIC CAPTURE MODE FIG. 3. Schematic of a continuous magnetic separator having one inlet and three outlets. 19 The magnetic field is applied vertically. Paramagnetic particles are attracted and diamagnetic particles are repelled. (b)

5 Continuous Separation Fig. 3 shows a schematic of a continuous magnetic separator having one inlet and three outlets. The magnetic separator consists of a long flow channel composed of a thin rectangular cross-section tube and a ferromagnetic wire. The magnetic field is applied vertically and parallel to the gravitational force. Paramagnetic particles are attracted toward the magnetic wire while diamagnetic particles are repelled. The continuous magnetic separation method does not trap particles in the separator. Therefore it does not require a wash-out cycle. The continuous magnetic separation method provides a very sensitive and selective separation technique This technique is especially desirable for selective separation between very weakly paramagnetic and diamagnetic particles. This separator provides a paramagnetic particle rich slurry from outlet #1 (the nearest outlet to the magnetic wire) and a diamagnetic particle rich slurry from outlet #3 (the farthest outlet from the magnetic wire). Fig. 4 shows a fabricated continuous separator used for blood cell separation. Table 2 gives the design parameters of the separator. The channel of 3.6 m flow length was made of plastic tubing which was co-wound with a ferromagnetic wire in a 14-turn, single-spiral, rectangulargroove on a Plexiglas tube. The flow channel tube has one inlet and three outlets. TABLE 2 Design parameters of 1-inlet 3-outlet continuous HGMS separator. Flow cell dimensions Separator winding bobbin Separator length L 3.58 m Diameter x length 82 mm x 1 mm X.76 mm Rectangular groove 3.48 mm x 1.9 mm X mm Groove turn number 14 X mm Material Plexiglas X mm Flow channel plastic tube S.61 mm (original circular dimensions) Ferromagnetic wire Outer diameter 1.96 mm Material Stainless steel SUS 43 Inner diameter 1.47 mm Radius a.52 mm Saturation magnetization 1.7 T (a) FIG. 4. Fabricated continuous separator of 3.6 m flow length. 19 (a) Parts before assembling, plastic tube (left bottom), Plexiglas tube with a magnetic wire in a single-spiral, rectangular-groove of 82 mm diameter, 1 mm long (left top) and covers. (b) The flow channel made of plastic tubing is wound with a ferromagnetic wire in a 14-turn groove on the Plexiglas tube. The flow channel tube has one inlet and three outlets. (b)

6 The cell motion in the magnetic separation process is found from the force equilibrium equation; F g + F d + F m = (1) here F m is the magnetic force on the cell to be separated, F g is the gravitational force on the cell and F d is the hydrodynamic drag force. The gravitational force on the cell of a volume V p is given by F g = -g(ρ p - ρ f )V p x where x is the unit vector in the vertical direction (the direction of the gravitational force), g is the gravitational constant, and ρ p and ρ f are the densities of the cells and fluid, respectively. The hydrodynamic drag force at zero flow is given in the Stokes region by F d = -6πηbv p where η is the fluid viscosity, b is the radius of the cell, and v p is the velocity of the cell. If the thickness S of the separation flow channel is small enough to neglect the effect of the azimuthal force, the magnetic force is approximated for a high magnetic field by F m -{(χ p χ f ) a 2 µ MH V p /x 3 }x (2) where a is the radius of the ferromagnetic wire, M is the magnetization of the wire, and H is the applied magnetic field, and χ p and χ f are the magnetic susceptibilities of the particles (cells) and fluid (plasma), respectively. The cell traveling time T t required for the cell to move from the entering position x to the position x 1 at the separator outlet is obtained from Eq. (1). The time T t should be equal to L/v where L is the separator length, and v is the average flow velocity. Consequently, the equation to describe the cell motion through the separator is given as: L V g / (a v ) = G(x a ) - G(x 1a ), (3) here G(x) = x - X c [(1/6) log (x + X c ) 3 /(x 3 + X c 3 ) + 3-1/2 arctan{(2x - X c )/3 1/2 X c }] (4) X c = ( V m /V g ) 1/3 = {(µ χmh /(gρa)} 1/3 (5) V g = 2gρb 2 /(9η) (Gravitational velocity) (6) V m = 2µ χmh b 2 /(9ηa) ( velocity) (7) ρ = ρ p - ρ f (Relative density) (8) χ = χ p - χ f (Relative magnetic susceptibility) (9) where x a = x /a (x is the entering position.), and x 1a = x 1 /a (x 1 is the position at the outlet.). The separation capacity of the throughput Q ( the cross-section x the average flow velocity ) is given by Q = ( X - X 3 )( S a a )( L a ) a V 2 g (1) G(x a ) G(x 1a ) where S is the canister thickness, and X and X 3 are the distances from the walls of the canister to the wire axis as shown in Fig. 3. If the separator dimensions are scaled by the magnetic wire radius a, the throughput Q is proportional to a 2. The throughput is proportional to the particle radius squared (b 2 ) since V g is proportional to b 2 and G(x) is not a function of b. Continuous Direct Separation of Red Blood Cells 19 Human venous blood samples used for the experiment were drawn at a local medical department into evacuated glass tubes containing sodium citrate. The hematocrit (cell concentration) was reduced to about 13% by dilution with the plasma of the drawn blood. To obtain the deoxygenated state of the red blood cells, a gas permeable membrane method was used in the same way as shown in Fig. 1. Nitrogen gas flow was provided around the permeable membrane tube. The blood flow rate was controlled by a syringe pump. The separator was operated in a magnetic field of 2 T which was generated with a warm-bore superconducting solenoid magnet. Red blood cell volume concentrations of the samples obtained from the outlets were evaluated by sedimentation using a glass capillary tube (.8 mm ID, 1.1 mm OD, and 9 mm long).

7 Fig. 5 shows both experimental and calculated results of red blood cell separations. Relative concentrations obtained for the three outlets at the average flow velocity v = 6.4 mm/s and at magnetic fields of zero and 2 T are plotted. For analytic calculations, the magnetic susceptibilities χ p = -3.8x1-6 (in SI units) for deoxygenated red blood cells, χ f = -7.7x1-6 for plasma 14, and the relative density ρ = ρ p - ρ f = 1 kg/m 3 (ρ p = 11 kg/m 3 and ρ f = 1 kg/m 3 ) 2 were used. The parameter b 2 /η = 3.57x1-9 m/kg s was chosen for the best fit of the experimental data obtained for the various flow velocities of 6.4 mm/s, 9.5 mm/s and 12.6 mm/s at magnetic fields of zero and 2 T. This value of b 2 /η was confirmed to agree well with the experimental value obtained from a sedimentation test for the same blood as that used for the magnetic separation experiment. In Fig. 6, the experimental results are compared with calculated results as a function of the flow velocity. The figure shows the relative concentrations of red blood cells at a magnetic field of 2 T. Overall the experimental results agree well with the results calculated from the equations taking into account the gravitational force. The red blood cell separation experiments were carried out using blood of 13% hematocrit which was diluted with the plasma. The hematocrit of human blood is approximately 4% 45%. The separation efficiency may decrease with increasing hematocrit. However, the separation efficiencies could be improved by optimizing the separator design in various ways, such as by using a multi-stage operation and multiple-ferromagnetic-wire arrangements to combine diamagnetic and paramagnetic capture modes. RELATIVE CONCENTRATION OF RED BLOOD CELLS (%) Flow Velocity = 6.4 mm/s EXPERIMENT B = 2 T B = T CALCULATION B = 2 T B = T.5 1#1 #2 # OUT LET FIG. 5. Experimental and theoretical calculation results of the relative concentrations for the three outlets at the flow velocity v = 6.4 mm/s. 19 RELATIVE CONCENTRATION OF RED BLOOD CELLS (%) B = 2 T EXPERIMENT CALCULATION FLOW VELOCITY (mm/s) FIG. 6. Experimental and theoretical calculation results as a function of the average flow velocities at a magnetic field of 2 T. 19 ATTRACTIVE FORCE PARAMAGNETIC CAPTURE MODE Flow Cell Wire P Gravity DIAMAGNETIC CAPTURE MODE Flow Cell D Wire Gravity Wire Wire REPULSIVE FORCE D P Flow Cell Gravity Flow Cell Gravity Paramagnetic Particle Diamagnetic Particle Alternate Diamagnetic Capture Method Wires D P Gravity

8 FIG. 7. Cross-section view of the continuous magnetic separator having the magnetic wire at the top. Flow direction is perpendicular to the drawing plane. The magnetic field is applied vertically for the paramagnetic capture mode and horizontally for the diamagnetic capture mode. Paramagnetic particles are attracted toward the wire and diamagnetic particles are repelled from the wire in the paramagnetic capture modes. On the other hand, diamagnetic particles are attracted and paramagnetic particles are repelled from the wire in the diamagnetic capture mode. Alternatively, the diamagnetic capture mode can be obtained by mounting magnetic wires at the side of the long wall of the flow cell in a vertical field. Simulations of White Cell and Plasma Separations 19 For a selective separation of white blood cells and plasma, we consider a continuous magnetic separator having a ferromagnetic wire above the flow channel as shown in Fig. 7. If the magnetic field is applied vertically as shown in the left column (Paramagnetic Capture Mode), paramagnetic particles are attracted toward the wire and diamagnetic particles are repelled. On the other hand, if the magnetic field is applied horizontally (Diamagnetic Capture Mode), diamagnetic particles are attracted toward the wire and paramagnetic particles are repelled 15. In these configurations having a magnetic wire above the flow cell, the particle separation profiles are calculated from Eq. (3) by replacing Eq. (4) with Eq. (11), G(x) = x + X c [(1/6) log (x - X c ) 3 /(x 3 - X c 3 ) - 3-1/2 arctan{(2x + X c )/3 1/2 X c }] (11) White blood cell and plasma separations by the continuous separator are analyzed using Eq. (3) for the separator shown in Fig. 4. The magnetic susceptibilities of these cells have not been found in the literature. Therefore, we used χ p = -11.6x1-6 for both white blood cells and oxygenated red cells for the calculation. This value gives the same absolute value χ p - χ f as that of the deoxygenated red blood cells, but the sign is negative. White blood cells are classified mainly into two groups and five different kinds of cells, agranulocytes (lymphocyte and monocyte) and granulocytes (neutrophil, eosinophil and basophil). Their sizes are between about 6 µm and 15 µm in diameter. However, the fitting parameter b 2 /η = 3.57x1-9 m/kg s obtained from the red blood cell separation was used for the following calculated simulations. Fig. 8 shows relative concentrations calculated for white and oxygenated red blood cells in the diamagnetic capture mode shown in Fig. 7. Fig. 8 is also applicable to deoxygenated red cell separation in the paramagnetic capture mode. That is, Fig. 8 is for the case of attractive force. On the other hand, Fig. 9 illustrates calculated results for repulsive forces which are applicable to deoxygenated red blood cells in the diamagnetic capture mode or white and oxygenated red blood cells in the paramagnetic capture mode. RELATIVE CONCENTRATION (%) ATTRACTIVE FORCE v = 6.4 mm/s v = 15 mm/s v = 3 mm/s MAGNETIC FIELD (T) RELATIVE CONCENTRATION (%) REPULSIVE FORCE v = 6.4 mm/s v = 15 mm/s v = 3 mm/s MAGNETIC FIELD (T) FIG. 8. Simulation of blood cell separation under attractive forces: white cells and oxygenated red blood cells in the diamagnetic capture mode, and deoxygenated red blood cells in the paramagnetic capture mode, calculated on the basis of the separation of deoxygenated red blood cells. 19 FIG. 9. Simulation of blood cell separation under repulsive forces: white cells and oxygenated red blood cells in the paramagnetic capture mode, and deoxygenated red blood cells in the diamagnetic capture mode, calculated on the basis of the separation of deoxygenated red blood cells. 19

9 In general, a high magnetic field is obtained more conveniently by a solenoid magnet rather than by a dipole magnet. It is desirable for the continuous magnetic separator to have a vertical magnetic field in a solenoid magnet. To obtain the diamagnetic capture mode in the vertical field arrangement, the magnetic wire will be mounted at the side of the long wall of the flow channel as shown in Fig. 7. The magnetic wires can be placed at one side or both sides to enhance the magnetic force. These simulations of white blood cell and plasma separations were performed with the estimated magnetic susceptibility χ p = 11.6x1-6 of the white cells and oxygenated red blood cells. The actual relative susceptibility χ p - χ f might be smaller than this value. If we use the magnetic susceptibility of water (-9.2x1-6 ) for white cells, the relative susceptibility χ p - χ f becomes about half of the value we used for the above simulation. However, white cells are much larger than red blood cells. The volumes of lymphocytes, monocytes and granulocytes are approximately 3, 8, and 6 times the red blood cell volume, respectively 21. force is proportional to the volume of the cell. In the continuous magnetic separation process, the throughput Q is proportional to the particle radius squared (b 2 ) as discussed with Eq. (1). These two factors of the magnetic susceptibility and the volume are approximately canceled out. Therefore, the simulation results are acceptable as a first approximation of white cell and plasma separations. White Cell Separation Red-blood-cell free white-blood-cell collection To obtain white cells without red blood cells, whole blood should be deoxygenated by passing it through a permeable membrane tube in nitrogen gas. When deoxygenated whole blood continuously passes through the separator in the diamagnetic capture mode shown in Fig. 7, red-cell free white-cells and red-cells are obtained from outlet #1 (the nearest outlet to the wire) and outlet #3 (the farthest outlet from the wire), respectively. The separator having three outlets provides one third of the treated blood from each outlet. As seen in Fig. 9, deoxygenated red blood cells are completely repelled from the region of outlet #1 at 3 mm/s and magnetic fields greater than 6 T. White cell recoveries at 6 T are 54%, 46% and 41% at the flow velocities of 6.4 mm/s, 15 mm/s and 3 mm/s, respectively, as seen in Fig. 8. At 1 T white cell recoveries can increase to 66%, 55% and 47% at the flow velocities of 6.4 mm/s, 15 mm/s and 3 mm/s, respectively. Those flow rates correspond to the throughput of 33 ml/h, 7 ml/h and 15 ml/h, respectively. The treatment capacities can be increased by increasing the magnetic field and the separator length. From Eq. (1), if the magnetic field and all the separator dimensions including the wire size are doubled, the throughput can be increased by four times with the same flow velocity. If this doubled-dimension separator is extended to 15 m, it might be possible to increase the capacity to 1.2 L/h with white cell recovery of about 5% at the field of 12 T and the flow velocity of 3 mm/s. White-blood-cell free red-blood-cell collection To collect red blood cell without white cells, deoxygenated whole blood would be separated in the paramagnetic capture mode in Fig. 7, white-cell free red-cells are obtained from outlet #1 (the nearest outlet to the wire). The separation efficiencies of white-cell free red-cell collection will be similar to those of the above operation. Therefore, the doubled-dimension separator of 15 m long at the field of 12 T and the average flow velocity of 3 mm/s might obtain white-cell free red-blood-cells at 1.2 L/h with red cell recovery of about 5%. Plasma Separation To separate plasma from whole blood, the continuous magnetic separator can be operated with an oxygenated blood state in the paramagnetic capture mode in Fig. 7. Both red and

10 white blood cells are diamagnetic in oxygenated blood. Therefore, they are repelled from the region of outlet #1 (the nearest outlet to the wire) and plasma without red and white cells is obtained from outlet #1. In this case calculated results of the simulation are shown in Fig. 9. The separator in Fig. 4 may treat whole blood of 15 ml/h (3 mm/s) at a field of 6 T to separate plasma at 5 ml/h (1/3 of whole blood). Using the doubled-dimension separator of 15 m length discussed above, the operation capacity can be increased to 1.2 L/h at a field of 6 T and the average flow velocity of 3 mm/s. Conclusions Experimental results of continuous direct magnetic separation of red blood cells and simulation studies of white cell and plasma separation indicate that it is possible to develop a therapeutic device for magnetic separation of blood components from whole blood without using any additives such as magnetic tagging agents or reducing agents. The continuous direct magnetic separation technique would make possible approximately 1 L/h processing of whole blood for plasma and white blood cell separations in a high field superconducting magnet of 6 T to 12 T. This magnetic separation method has the following features: no magnetic additives required, continuous operation, quiet, no moving parts, easy operation, and completely closed system. High magnetic field applications have been widely expanded in the last decade. In medical fields, superconducting magnets have been successfully accepted for magnetic resonance imaging (MRI). cell separations are unique, since the separation is based on the magnetic susceptibility of the cell. separation with a high field superconducting magnet makes possible biological cell separation even if the magnetic susceptibility is in the order of only 1-6 (SI). References 1) R.J. Sanderson, Separation of different kinds of nucleated cells from blood by centrifugal elutriation, Cell Separation Methods and Applications, Vol. 1, T. P. Pretlow II and T. G. Pretlow, ed., , Academic Press, New York, ) I. Bertoncello, A comparison of cell separations obtained with centrifugal elutriation and sedimentation at unit gravity, Cell Separation Methods and Applications, Vol. 4, T. G. Pretlow II and T. P. Pretlow, ed., 89-18, Academic Press, New York, ) J.J. Fournel, J. Zingsem, J. Riggert, L. Muylle, N. Muller, M. Kohler, J. L. Beaumont, M. Baeten, R. Eckstein, and G. Van Waeg, A multicenter evaluation of the routine use of a new white cell-reduction apheresis system for collection of platelets, Transfusion, 37, , ) H.H. Kolm, Device, U. S. Patent 3,567,26 (March 2, 1971). 5) J.A. Oberteuffer, separation: A review of principles, devices, and applications, IEEE Transactions on s, Mag-1, , ) D.F. Melville, F. Paul and S. Roath, "Direct Separation of Red Cells from Whole Blood," Nature, 255, 76, ) D.F. Melville F. Paul and S. Roath, "High Gradient Separation of Red Cells from Whole Blood," IEEE Transactions on s, MAG-11, 171, ) D. Melville, F. Paul and S. Roath, "Red Blood Cells in High Gradient Separation," in Industrial Applications of Separation, Y. A. Liu, ed., Publication No. IEEE 78CH1447-2, 39, ) M.D. Graham, "Blood-Cell Separation by High-Gradient s," Bibliotheca Anntomica, 16, 345, ) M.D. Graham, "Efficiency Comparison of Two Preparative Mechanisms for Separation of Erythrocytes from Whole Blood," J. Appl. Phys., 52, 2578, ) C.S. Owen, "High Gradient Separation of Erythrocytes," Biphy. J., 22, 171, ) C.S. Owen, Magnetite cells sorting, Cell Separation Methods and Applications, Vol. 2, T.G. Pretlow II and T.P. Pretlow, ed., , Academic Press, New York, ) C.S. Owen and P. A. Liberti, Magnetite-protein conjugates for the separation of cells by high gradient magnetic filtration, Cell Separation Methods and Applications, vol. 4, T. G. Pretlow II and T. P. Pretlow, ed., , Academic Press, New York, 1987.

11 14) M. Takayasu, N. Duske, S.R. Ash, and F.J. Friedlaender, HGMS Studies of Blood Cell Behavior in Plasma, IEEE Trans. Magn. Mag-18, No. 6, 152, ) M. Takayasu, E. Maxwell, and D.R. Kelland, Continuous Selective HGMS in the Repulsive Force Mode, IEEE Trans. Magn. Mag-2, No. 5, 1186, ) M. Takayasu and D.R. Kelland, "Selective continuous magnetic separation of two-component particulate suspensions," IEEE Transactions on s, Mag-22, No. 5, , ) D.R. Kelland and M. Takayasu, "Increased Selectivity in Continuous Axial HGMS Separators," IEEE Trans. Magn., Mag-24, No. 6, , ) D.R. Kelland, " Separation of Nanoparticles," IEEE Trans. Magn., Mag-34, No. 4, , ) M. Takayasu, D.R. Kelland, and J.V. Minervini, Continuous Separation of Blood Components from Whole Blood, 16 th International Conference on Magnet Technology, Florida, USA, ) H. Pertoft and T.C. Laurent, Sedimentation of cells in colloidal silica (Percoll), Cell Separation Methods and Applications, vol. 1, T.P. Pretlow II and T.G. Pretlow, ed., , Academic Press, New York, ) R.J. Sanderson, Separation of different kinds of nucleated cells from blood by centrifugal elutriation, Cell Separation Methods and Applications, vol. 1, T. P. Pretlow II and T. G. Pretlow, ed., , Academic Press, New York, 1982.