- Email: [email protected]
Automation of the use of fluorescent microspheres for the determination of blood flow
Computer Methods and Programs in Biomedicine 61 (2000) 11 – 21 www.elsevier.com/locate/cmpb
Automation of the use of fluorescent microspheres for the determination of blood flow E. Thein *, S. Raab, A.G. Harris, K. Messmer Institute for Surgical Research, Klinikum Großhadern, Uni6ersity of Munich, Marchioninistr 15, 81366 Munich, Germany Received 15 December 1998; received in revised form 16 March 1999; accepted 25 March 1999
Abstract Fluorescent-labeled microspheres (FM) are a new tool for the determination of organ blood flow. However, the FM-method is labor intensive, because of the necessity to recover the microspheres from the tissue samples. The aim of this study was to automate the FM-method. A Zymate-Robotic System (Zymark, Idstein, Germany) was modified to handle a novel filtration device. The robot is surrounded by 12 different stations which are necessary to process the samples. It performs the sequential steps which are needed to recover the microspheres from the samples. The dyes are finally released from the FM with a solvent and their fluorescent intensity is measured online using a spectrophotometer (Perkin Elmer, U8 berlingen, Germany). The robotic system is able to recover the FM through digestion and filtration of the tissue samples using the new filter, to dissolve the FM and to release the dyes so that their fluorescent intensities can be measured for the calculation of organ blood flow. © 2000 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Robot; Spectrometer; Instrumentation; Radioactive microspheres; Online measurement; Blood flow
1. Introduction Microspheres (MS) have become a common tool for the determination of regional organ perfusion since Rudolph and Heyman first introduced the method in 1967 [1].The method is based on the principle that biologically inert microspheres will be trapped due to their diameter in the microvasculature [2] after left-side intraatrial or intraventricular injection. An arterial blood sample [3], drawn during and after the injection of MS at a regular and known rate using a Harvard* Corresponding author. Tel.: +49-89-70954404.
pump serves as an ‘artificial organ’ with preset flow to calculate the perfusion of the individual organs. At the end of the experiment the animal is sacrificed and the organs of interest are removed and dissected according to predetermined hierarchy schemes. After the individual samples are weighed, the number of MS is determined by different methods depending on the type of MS used. At present there are three types of MS available: radioactive labeled (RM), colored (CM) and fluorescent (FM) microspheres. The first regional blood flow measurements using the microsphere technique were done in the
0169-2607/00/$ - see front matter © 2000 Elsevier Science Ireland Ltd. All rights reserved. PII: S 0 1 6 9 - 2 6 0 7 ( 9 9 ) 0 0 0 2 4 - 3
12
E. Thein et al. / Computer Methods and Programs in Biomedicine 61 (2000) 11–21
fetus of sheep in utero by Rudolph and Heyman [1] using RM. These MS are labeled with radioactive isotopes, their radioactivity is proportional to the number of MS. The radioactivity of the organ tissue samples is measured in a g-counter as counts per minute. Soon after their introduction RM became the standard tool to determine organ blood flow. But the RM method does have certain disadvantages: Due to the short half life of most of the isotopes, RM are not suitable for chronic experiments. The researcher is in permanent exposure to radiation. Because the research-personal is working with radioactive substances, specialized training and facilities are required. The cadaver and the organ samples have to be treated as radioactive waste, of which the disposal is increasingly expensive. Because of these disadvantages Kowallik et al. [4] and Hale et al. [5] introduced colored microspheres (CM), which are polystyrene beads, containing different colors. CM eliminated the problems associated with radioactivity, but this method has its own disadvantages. While using the CM-method the number of microspheres present in the sample is not measurable from outside the organ samples, as it is with the RM-method. Therefore, the CM have to be isolated from the tissue samples, so that the intensities of the colors can be measured. The CM are recovered from the sample by digestion of the tissue in a mixture of chemical and enzymatic liquids [4], followed by the separation of the MS from the digestion fluid through several steps of centrifugation, washing and aspiration of the supernatant [6,7]. This method is known as the sedimentation method. The intensity of the colors is measured with a photometer after they have been released from the MS. In addition to the problem of a loss of MS during the aspiration of the supernatant, the method has the disadvantage that only five colors are available and their spectra are not separable when all five colors are used in the same experiment. To cope with the problem of unseparable spectra Glenny [8] introduced fluorescent micro-
spheres (FM) in 1993, which contain fluorescent dyes. The recovery of these MS from the organ tissue samples is achieved in a similar manner to that of the CM-method. The tissue is digested for 24 h in 4 M KOH, the digestate is filtered through a 10-mm pore-sized polyester filter, the dyes are released from the spheres and their fluorescent intensity is determined with a fluorescence spectrophotometer. This procedure is known as the filtration method. Since the samples are handled in several different containers this method also bares the risk of a loss of microspheres during processing, consequently leading to false blood flow values [8]. Thus, both the CM- and the FM-methods have clear disadvantages in comparison to RM. They are more labor and time intensive due to the necessity for the recovery of the MS from the tissue; e.g. digestion time using the sedimentation method is 48 h [6,7] and 24 h using the filtration method [8]. The digestion method requires different containers to digest the tissue, and to recover the MS by filtering the digestion fluid. MS may, therefore, be lost during the transfer from one container to another leading to unreliable and invalid results in the calculation of the blood flow. The sedimentation method requires the aspiration of the supernatant to recover the MS. Therefore, MS may get lost during the aspiration. The fluorescent dyes are not stable at non-neutral pH-values. It is therefore necessary to assure that no KOH-residues are attached to the MS, which is especially difficult to achieve in the sedimentation method. We have recently developed a new filter device which allows processing of the sample in a single container throughout the whole procedure of weighing, digesting, rinsing, filtering and releasing the fluorescent dyes from the MS [9]. Since the entire FM recovery can be performed in a single container, it should be possible to automate the whole procedure and to eliminate the disadvantages of the time and labor intensity. Therefore, the aim of this study was the complete automation of the time and labor intensive FM-method.
E. Thein et al. / Computer Methods and Programs in Biomedicine 61 (2000) 11–21
13
2. Materials and methods
2.1. Hardware The newly designed filter device (Sample Processing Unit (SPU)), which was developed at the Institute for Surgical Research (ISR) has been described by Raab et al [9] (Fig. 1). The robot consists of commercially available parts (Zymark, Idstein, Germany) and of components that were designed and constructed at our institute. The Zymark robotic system was supplemented with the luminescence spectrophotometer LS50B and an automated sampling unit (Perkin Elmer, U8 berlingen, Germany). Fig. 2 shows a scheme of how the components are connected. All components of the system except those listed in the next paragraph are fixed to different pie-sections, which are arranged in a circle around the centrally positioned robot (Fig. 3.). The Master-Laboratory-Station (MLS), the deposit stations, all pumps for dilution or suction and the Power-and-Event-Controller are located below the robot’s working area. The four entrance racks (Fig. 3) were designed at the ISR and consist of PVC-plates with a size
Fig. 1. The filtration device ‘Sample-Processing-Unit’ (SPU). The SPU (1) consists of three stages; the filter (2) the filter holder (3) and the sample tube (4) in which the fluorescent dyes are collected. During digestion of the tissue samples the beaker is closed with a Delrin cap (5). The sample tube can be closed with a screw cap (6)
Fig. 2. The connection of the different components of the system. The System V-Controller (3) is connected to the AS90/91-Controller for direct communication between the robot (6) and the automated luminescence spectrometer LS50B (5). One computer (1) is used to control the robot. The only modules of the robot shown in this figure are the barcode reader (8) and the balance (7) because they are the only components which need to deliver data to the computer (1). The other computer (2) is used to control the LS50B and to store the data of the fluorescence measurement.
of 21.5 × 35.5 cm, on which aluminium pipes are vertically fixed in a 8× 5 grid pattern 0.3 cm apart. The aluminium pipes are 7 cm high and have a diameter of 3.5 cm. The PVC-plates are mounted on frames which are fixed to the pie-sections. Each rack is capable of holding 40 SPUs. The core of the whole system is the filtration and wash station. It was newly constructed by Zymark according to our design. A sled, which can be moved up and down pneumatically along a track, is fitted to a vertical column. This sled carries a shower head (ISR), which is used to rinse the filters with buffer and to handle the filters throughout the whole procedure. To handle the filters a small part of the top of the shower head is spreadable (Fig. 4). If a filter is to be removed from a container, the robot positions the container including the filter so that the closed shower head protrudes into the filter. When the head is opened two spikes which are fixed to the shower head are pressed into the filter’s wall. Thus, the filter is held by the shower head and by moving the container downwards the robot can remove the filter from the container.
14
E. Thein et al. / Computer Methods and Programs in Biomedicine 61 (2000) 11–21
Fig. 3. Overview of the robot. The central positioned robot is surrounded by the different components that are needed for the procedure. (1) entrance racks; (2) pipetting station with vortex (arrow); (3) cap dispenser; (4) separation station; (5) heater blocks; (6) centrifuge with belonging rack (arrow); (7) digital balance; (8) autosampler of the luminescence spectrometer; (9) computer for the robot. To the right of the robot a part of the luminescence spectrometer can be seen.
The second function of the filtration and wash station is to filter the digestate through the filter’s mesh work after the digestion of the sample. To do this the filter is pressed down by the shower head onto a suction-mug, which is placed at the lower end of the column and which is connected via a reception container to a vacuum. With a negative pressure of 400 kPa the digestate is drawn through the filter within about 5 – 10 s. The third function of the filtration and wash station is the rinsing of the filter. Since it is important to rinse the inner walls of the filter the shower head is laterally perforated, so that the buffer is sprayed outward onto the wall. The rinsing of the filter from the inside assures a neutral pH-value in the contents of the filter. Furthermore, the filters are washed in the buffer from the outside to remove any KOH residues at the same station. To do so the robot takes the filter after it is rinsed from the shower head and swivels it for 10 s in a container which is placed next to the suction-mug. This container (ISR) is equipped with an inlet at its bottom and an outlet at its top. Fresh buffer (10 ml) is added
to the container through the inlet via the MLS before each washing procedure to assure that there is no KOH in the container.
Fig. 4. Shower-head. The shower-head (1) is laterally perforated to rinse the walls of the filter which is necessary to wash out all KOH-residues. A smaller part (2) of it is spreadable and a spike (3) is screwed into it. This part, including the spike is used to remove the filters from any container.
E. Thein et al. / Computer Methods and Programs in Biomedicine 61 (2000) 11–21
For the digestion of the tissue samples the filters are placed into high-grade steel containers (beaker) that were designed at the ISR and consist of a height of 6.5 cm and a diameter of 3 cm. The heater block (Perkin Elmer) used is capable of holding 40 beakers. It is made of aluminium, which is coated with PVC. In the heater block the digestion of the tissue samples at a constant temperature of 60°C takes place. Since it is necessary for the robot to retrieve tubes from a specific position in the centrifuge, the centrifuge is not only equipped with a motor for the rotation, but also with a stepping motor. The robot uses this stepping motor to move the rotor to the position where the desired tube has to be located. Once there, the tube to be retrieved can be grasped by the robot and removed from the centrifuge. The rack (ISR) of the autosampler (Perkin Elmer) of the measuring unit is a PVC-plate (size 21.5× 35.5×3.0 cm) with three cavities for the sample tubes. One is for the sample tube, whose contents are to be measured, and the other two are for exchanging the measured sample tube for the next sample to be measured. The pipette station (Zymark) is connected via two hoses with two syringes of the Master-Laboratory-Station (Zymark) for the pipetting of the organic solvent and isopropanol. Furthermore there are two connections to pumps, one for the pipetting of KOH and the other to suck off KOH-residues from the beakers. A capping station (Zymark) is used to separate the filter holder including the filter from the attached sample tube containing the fluorescent dye before the measurement of the dye-solution. For this the robot places the SPU between two jaws that are located on a platform that can be rotated. The jaws close and fix the sample tube in this manner. Another pair of jaws moves over the filter holder, grasps it and while the lower platform starts rotating, moves up again. In this way filter holder and sample tube are separated. The cap-dispenser (Zymark) used is capable of holding 400 screw caps which are needed to close the sample tubes after the fluorescent measurement is completed. It consists of four pipes of 130 cm height and a diameter of 2.5 cm. The lower
15
Table 1 Components of the robotic system Commercial apparatus (Zymark) XP Z830 ZP411 Z710 ZP510 Z620 CP
Robot Power and event controller System V controller Capping and separation station Centrifuge station Master laboratory station Vortex station Cap dispenser Hand B
Commercial apparatus (Perkin Elmer) AS-90/AS-91 Controller AS-91 Autosampler Dil-1 Diluter station LS50B Luminescence spectrometer Heater blocks Commercial apparatus (Data logic) DS40 Bar-code reader Commercial apparatus (Mettler) PM200 Digital balance Apparatus designed at the Institute for Surgical Research Filtration and wash station All racks Wash container
end of the tubes are open over a platform, from which the robot gets the caps. The caps are piled up inside the pipes. If one of the pipes is empty, the robot realizes via the strength of its hand’s grip that there is no cap under this pipe on the platform. It then moves to the next pipe and gets a cap from there. A digital balance (Mettler, Giessen, Germany) is used to determine both the weight of the empty SPU and the weight of the SPU containing tissue samples. Both weights are stored in an Excel database automatically. The weight of the tissue samples is determined by subtracting both weights. The single SPUs are identified with the help of a bar-code reader (Datalogic S.P.A., Lipo de Calderara, Italy) that reads the bar-code with which the sample tube of the SPU is labeled. All components of the system are listed in Table 1.
16
E. Thein et al. / Computer Methods and Programs in Biomedicine 61 (2000) 11–21
3. Software The program used to control the robot is written in a Zymark Assembler Code. The system is divided into predefined pie-sections, onto to which the individual hardware components of the robot are fixed. All properties of each component are defined in the computer program. The different coordinates the robot needs for correct positioning are defined with the help of a remote control. With this remote control the robot’s hand is moved to the desired position and the coordinates are stored in the program. The precision of the positioning of the hand is 1mm. Besides the coordinates of the different components, clear and wait positions for each procedure are defined in the program. The program is designed as a pipeline system, meaning that the program is divided into nine subsections called ‘Luos’. During each work cycle the robot brings a sample into the next Luo. Therefore there is one sample entering and one sample leaving the robotic system at each cycle. Error routines are present in each procedure. The program compares defined coordinates as well as the strength of the hand’s grip with the actual data sent from the robot. If the actual and defined data diverge the system is stopped automatically and a correction routine is presented to the user. The robot’s program is connected to an Excel database (Microsoft, Seattle, WA, USA) via a program called Easylink (Zymark) allowing for the storage of data such as the bar-code or the weight of the SPU that are determined during the procedure.
4. Procedure for automated sample processing
4.1. Preparation The sample units are labeled with bar-codes and the SPUs are put together. After they are placed into the entrance racks, the robot is started and the SPUs are tarred. For this the robot takes the SPU from the entrance rack, reads its barcode and weighs it. Both the tare-weight and the
bar-code are stored in an Excel-sheet through a program called Easylink (Zymark). This program is needed to create a connection between the robot and Excel via the power and event controller. The bar-code is used to identify the samples. After the SPUs have been tarred the organ samples are placed into the filters and the SPUs are closed with caps (ISR) made of Delrin, which is inert to almost all chemicals. This is the final step which has to be done by hand.
4.2. Digestion The number of samples to be processed is entered into the start-menu and the robot is started. The Delrin cap from the first SPU is placed upon a special deposit station by the robot and the SPU is weighed again in the previously described manner to determine the weight of the tissue sample. When weighing is completed, the robot brings the SPU to the filtration and wash station where the filter is separated from the filter holder. The filter holder and attached sample tube are placed back into their original position in the entrance rack. The robot then gets a beaker from the heater rack, moves to the filtration and wash station, and places the filter into the beaker. In the next step the robot brings the beaker and filter to the pipetting station, where a maximum of 15 ml of 4 N KOH (the amount depends on the weight of the sample) and 1.5 ml of isopropanol are added into the filter The beaker and filter are placed back into the heater block and are capped with the Delrin cap to avoid evaporation of the liquid [9]. The organ sample is digested for 6 h at a constant temperature of 60°C. Excluding the period of 6 h for the digestion of the tissue, the robot needs 12.5 min to completely process a sample. Therefore the robot repeats the above described procedure every 12.5 min with the next SPU.
4.3. Filtration and washing When the digestion period of 6 h for the sample is over, the robot takes the Delrin cap off the
E. Thein et al. / Computer Methods and Programs in Biomedicine 61 (2000) 11–21
beaker and drops it into a waste container. It takes the beaker off the heater block and brings it to the filtration and wash station, where the beaker and filter are separated. The robot then moves the beaker to the pipetting station where the remnants of the digestion fluid are aspirated. After this is completed the beaker is replaced into the heater block where it remains uncapped and ready for the next filter. Concurrently the filter is lowered onto the suction mug and the digestate is filtered. The filter is then rinsed twice by sending 15 ml of a phosphate buffer through the shower head. Following this, the outside of the filter is washed for 10 s by swiveling it in a container which is filled with the above mentioned buffer. To avoid the pollution of the buffer inside the container by KOH-residues, 10 ml of fresh buffer are added to the container before each washing procedure. The filter is then put into a 50-ml centrifugation tube. In order to prevent imbalances of the centrifuge while drying the filter, the centrifugation tube and the filter are weighed before they are centrifuged. If the weight is above the preset limit of 17 g, meaning that the filter is not completely evacuated, the unit is brought back to the filtration and wash station where the filter is removed from the centrifugation tube and put back into its SPU in the entrance rack. Samples that are singled out in this manner have to be processed further manually. If the filter has been successfully evacuated the centrifugation tube including the filter is put into the centrifuge and the filter is dried by centrifugation at 2800 rpm for 1 min. This procedure of rinsing, washing and drying the filter is repeated once again to assure that all KOH-residues are removed and that all FM are collected on the filter. The filter is then removed from the centrifugation tube at the filtration and wash station and placed back into the corresponding SPU.
4.4. Dye reco6ery In order to release the fluorescent dyes from the FM they have to be dissolved.
17
The robot moves the SPU into the pipetting station where 1 ml of organic solvent (cellosolve) is pipetted into the filter. The SPU is then vortexed softly for 30 s. The robot returns to the pipetting station where another 1 ml of the solvent is added to the filter. After another 30 s of soft vortexing the SPU is placed into the centrifuge where the fluorescent dye solution is extracted from the filter by spinning the SPU for 1 min at 2800 rpm. The fluorescent dye has now been collected in the sample tube. The SPU is taken from the centrifuge and brought into the capping station, where the sample tube is fixed between two jaws. The gripper of the capping station then moves over the top of the SPU, grips it and separates the filter holder including the filter from the sample tube by moving up again.
4.5. Fluorescence measurement The sample tube with the dye-solution is gripped by the robot and released by the jaws. The robot then places the sample tube into the autosampler (Perkin Elmer). An electronic signal sent from the robot to the controller of the measuring unit starts it. The pipette of the autosampler is driven over the sample tube, 550 ml of the dye are sucked into the pipette by a syringe of the diluter station and transferred into the flow cuvette of the LS50B. The fluorescent intensity is measured and stored on a F1 twinlab-default database (Perkin Elmer) automatically. During the measurement the robot retrieves the filter holder and filter, which are still held in the gripper at the capping station, and disposes them in another waste container. After the measurement of the sample the whole pipe system of the diluter station and the autosampler are rinsed with 10 ml of 100% ethanol and then dried for 20 s with air by a positive pressure of 3 bar. When this is finished, the controller of the measuring unit sends an electronic signal to the robot and thus directs the robot to retrieve the sample tube from the autosampler. The sample tube is again moved to the capping station, where it is placed between the two jaws. The robot then gets a screw cap from
18
E. Thein et al. / Computer Methods and Programs in Biomedicine 61 (2000) 11–21
the cap dispenser and presses it onto the sample tube. The jaws start rotating and the sample tube is capped. The closed sample tube is gripped by the robot and finally dropped into a collection box.
4.6. Calculation of the blood flow The intensities of fluorescence of the tissue samples and of the arterial blood reference are used to calculate the sample blood flow, according to the following formula: Orgflow = Corg(Pf/Cref) where Orgflow is the blood flow to the tissue sample in ml/min; Corg is the number of MS in the organ; Cref is the number of MS in the reference blood-sample; Pf is the flow-speed of the Harvard-pump Since all of the data are stored on an Excelsheet the blood flows can be calculated quickly and easily.
5. Experiments Experiments were performed to determine precision and reliability of the robot, as well as the stability of the fluorescent dyes for both dyes enclosed in the FM and dyes released from the FM.
5.1. Experiment 1 To test the precision of the pipetting of the organic solvent, a software program was written allowing to use the pipetting station repeatedly. Organic solvent was pipetted into a 50-ml centrifuge tube in 50 individual steps (1 ml each) by the robot. Following each step the robot weighed the tube and the weight was stored. The same procedure was done manually, to compare these results with the results of the robot.
5.2. Experiment 2 To determine if the automated procedure leads to a loss of fluorescent intensities the following
experiment was performed. One ml of each of the seven available types of FM-solutions was added to 20 ml distilled water. Of this solution, 40 ml or about 2000 FM were pipetted into 340 filters manually. These filters were processed by the robot and measured by the spectrophotometer according to the protocol for tissue samples. The same number of FM was pipetted manually into 25 glass test tubes and dried in an incubator at 60°C for 24 h. The pipetting of the organic solvent for this control group was performed by the robot, to avoid the inaccuracies of manual pipetting. As the FM and the fluorescent dyes of this group do not come into contact with KOH and there can be no loss of FM, the standard deviation of this group reflects the pipetting error of the manual pipetting of the FM-solution.
5.3. Experiment 3 An experiment was performed to assure that system breakdowns which remain uncorrected for a long time do not harm the fluorescent dyes inside the FM, in particular for those samples which stay in the heater blocks for a prolonged period. Of a red FM solution, 50 ml were pipetted into 60 filters and processed by the robot according to the protocol for tissue samples. When 30 filters were completely processed and the fluorescent intensity of the dye content of these samples was determined, a hardware failure was simulated by switching off the centrifuge. The 30 filters which were not completed were left in the heater block for 24 h and were then processed manually. The fluorescent intensities of both groups were compared using the students t-test
6. Results
6.1. Experiment 1 The density of cellosolve-acetate is 0.9725 g/ml. Pipetting exactly 1 ml should therefore result in an increase of the weight of the test tube of 0.9725 g per each pipetting step. The mean weight of 50
E. Thein et al. / Computer Methods and Programs in Biomedicine 61 (2000) 11–21
individual pipettations processed by the robot is 0.964 g, with a standard deviation of 0.003 g, a maximum of 0.972 g and a minimum of 0.957 g, whereas the manual pipettation results in a mean of 0.948 g with a standard deviation of 0.018 g, a maximum of 0.964 g and a minimum of 0.886 g. The robot is able to pipet the organic solvent more precisely and constantly than that could be done manually.
19
vortexing 0.5 min, second pipetting of the organic solvent 0.5 min, second vortexing 0.5 min, centrifugation 2 min, measurement (seven colors) 6 min. The total time that is needed to work one sample manually is at least 18.5 min, whereas the robot needs 12.5 min for the same procedure. (Table 2) A total of 6 min are saved with the automated method for each sample, which is significant when greater numbers of tissue samples are processed.
6.2. Experiment 2 The mean fluorescent intensity of the control is 46.26 arbitrary units (AU) with a standard deviation of 5.76 AU (= 12.47%). This standard deviation gives the error of the pipetting of the FM-solution. The mean fluorescent intensity of the group which was processed by the robot is 47.98 AU with a standard deviation of 4.78 AU (= 9.97%). This result indicates that the robot is processing the samples reliably and that there is no loss of FM or fluorescence in the dye solutions due to residual KOH.
6.3. Experiment 3 The mean fluorescent intensity of the filters which were processed under normal conditions is 27.53 AU with a standard deviation of 1.77 AU ( = 6.42%). The mean fluorescent intensity of the filters which remained in the heater block for 24 h is 26.56 AU with a standard deviation of 1.25 AU ( = 3.76%). As the results show, prolonged digestion times caused by system breakdowns do not lead to a loss of fluorescent intensity at least as long the dyes are still enclosed in the FM.
6.4. Comparison of time consumption manually/automatically To work one sample manually the following times are needed for the different steps: dissection and weighing 2 min, pipetting of the digestion fluid and isopropanol 1.5 min, filtering, rinsing and washing 3 min, centrifugation 2 min, first pipetting of the organic solvent 0.5 min, first
7. Discussion The system described here represents a complete automation of the FM-method which is accurate, time saving and reliable. In order to achieve this aim several modifications of the hard- and software were necessary to assure that the system completes the different steps of the procedure without mistakes. Using glue to connect any of the robots hard ware components appeared to be a problem. Due to the chemical vapors and the high temperature these connections dissolved and resulted in system breakdowns. The fingers of the robot are now coated with PVC-hoses, because of the problem that four PVC-pads that are glued into the finger’s inside were lost during the procedure. The loss of the pads resulted in a divergence in the distance between the fingers and the one defined in the software, which resulted in the breakdown of the system. Table 2 Comparison of time consumption for 1000 samples FM automatically FM manually RM Weighing/ sample Digestion Processing/ sample Measurement/ sample Total time
Automatically
2 min
2 min
360 min 12.5 min
360 min 14.5 min
None None
online
4 min
3 min
214 h
308 h
83.33 h
20
E. Thein et al. / Computer Methods and Programs in Biomedicine 61 (2000) 11–21
Emery paper was originally glued to the top of the shower head in order to increase the friction when the filters are removed from the different containers. This connection was dissolved, leading to the loss of filters and consequently to the stop of the routine. Therefore, the emery paper was replaced by two pointed screws, which are fixed to the shower head with a thread. This assures that the removal of the filters is done properly and this connection is not affected by the chemicals used. A major difficulty was to find a workable design for the different racks. All of the racks used now consist of vertically positioned, uninterrupted aluminium pipes of 7.5 cm height and a diameter of 3 cm. This design guarantees permanent guidance of the containers when they are placed into the racks. The permanent guidance is necessary due to the fact that vortexing and centrifuging may lead to non-vertical positioning of the containers. Because of the fact that the robot is not able to correct these inaccuracies by itself, the rack must be designed to avoid this problem when the robot attempts to place the tubes into the rack. The uninterrupted aluminium pipes serve to guide the containers back into a fully vertical position. The dye solutions must have neutral pH since the colors are not stable against non-neutral pHs after the FM have been dissolved. Therefore the entire filter—not only its mesh work — needs to be rinsed well with phosphate buffer. The shower head is therefore perforated centrally as well as laterally, so that the walls of the filters are also rinsed well. To ensure the elimination of KOHresidues the filters are centrifuged before they are rinsed another time. This centrifuging presses all KOH-residues that might still be attached to the filter’s wall onto the mesh-work, from where they are rinsed off during the second rinsing. Total costs of the robotic system, including the automated fluorescence spectrometer and the software are about $280 000. But the robot assisted FM-method saves manpower, time, costs for disposal of radioactive waste and eliminates the personal and environmental risk of harmful radioactive materials.
8. Conclusions Processing the tissue samples—according to the protocol of the FM-method—is completely automated by a robot which is accurate and reliable. It is possible to combine the robotic system with an automated spectrometer, thus allowing the online measurement of the fluorescent dyes. The most important steps of the method, for example pipetting the organic solvent are carried out more precisely and constantly by the system than that could be done manually. The automation of the time and labor intensive FM-method is 33% faster than processing the samples manually. In addition, the robot is able to work 24 h per day, 7 days per week with the same consistency and precision and is therefore able to process a higher number of samples than could be achieved manually. In the unlikely event of a system breakdown there is no loss of fluorescence from the dyes due to the fact that the fluorescent dyes are still enclosed in the FM. Furthermore the researcher is no longer constantly exposed to hazardous vapors of the chemicals used. Thus, the automation of the FM-method as presented here is a valid and promising alternative to both the radioactive method and the manual FM-method.
References [1] A.M. Rudolph, M.A Heyman, The circulation of the fetus in utero; methods for studying distribution of blood flow, cardiac output and organ blood flow, Circ. Res. 21 (1967) 163 – 184. [2] J.R.S. Hales, W.J. Cliff, Direct observations of the behavior of microspheres in the microvascaluture, Bibliotheca Anatom. 15 (1977) 87 – 91. [3] G.D. Buckberg, Studies of regional coronary flow using radioactive microspheres, Ann. Thoracic Surg. 20 (1975) 46 – 51. [4] P. Kowallik, R. Schulz, B.D. Guth, A. Schade, W. Pfaffhausen, R. Gross, G. Heusch, Measurement of regional myocardial blood flow with multiple colored microspheres, Circ. Res. 83 (1991) 974 – 982. [5] S.L. Hale, K.J. Alker, R.A. Kloner, Evaluation of nonradioactive, colored microspheres for measurement of regional myocardial blood flow in dogs, Circ. Res. 78 (1988) 428 – 434.
E. Thein et al. / Computer Methods and Programs in Biomedicine 61 (2000) 11–21 [6] New sedimentation method for determination of organ blood flow with fluorescent microspheres. Product information sheet, Molecular Probes, Eugene, OR, USA, 1996. [7] M.F.M. Van Oosterhout, H.M.M. Willigers, R.S. Reneman, F.W. Prinzen. Fluorescent microspheres to measure organ perfusion: validation of a simplified sample processing technique. Am. J. Physiol. 269 (1995) H725–H733.
21
[8] R.W. Glenny, S. Bernard, M. Brinkley. Validation of fluorescent-labeled microspheres for measurement of regional organ perfusion, J. Appl. Physiol. 74 (1993) 2585 – 2597. [9] S. Raab, E. Thein, A.G. Harris, K. Messmer. A new sample processing unit for the fluorescent microsphere method, Am. J. Physiol. 276 (1999) H1806 – H1807.
.
Automation of the use of fluorescent microspheres for the determination of blood flow E. Thein *, S. Raab, A.G. Harris, K. Messmer Institute for Surgical Research, Klinikum Großhadern, Uni6ersity of Munich, Marchioninistr 15, 81366 Munich, Germany Received 15 December 1998; received in revised form 16 March 1999; accepted 25 March 1999
Abstract Fluorescent-labeled microspheres (FM) are a new tool for the determination of organ blood flow. However, the FM-method is labor intensive, because of the necessity to recover the microspheres from the tissue samples. The aim of this study was to automate the FM-method. A Zymate-Robotic System (Zymark, Idstein, Germany) was modified to handle a novel filtration device. The robot is surrounded by 12 different stations which are necessary to process the samples. It performs the sequential steps which are needed to recover the microspheres from the samples. The dyes are finally released from the FM with a solvent and their fluorescent intensity is measured online using a spectrophotometer (Perkin Elmer, U8 berlingen, Germany). The robotic system is able to recover the FM through digestion and filtration of the tissue samples using the new filter, to dissolve the FM and to release the dyes so that their fluorescent intensities can be measured for the calculation of organ blood flow. © 2000 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Robot; Spectrometer; Instrumentation; Radioactive microspheres; Online measurement; Blood flow
1. Introduction Microspheres (MS) have become a common tool for the determination of regional organ perfusion since Rudolph and Heyman first introduced the method in 1967 [1].The method is based on the principle that biologically inert microspheres will be trapped due to their diameter in the microvasculature [2] after left-side intraatrial or intraventricular injection. An arterial blood sample [3], drawn during and after the injection of MS at a regular and known rate using a Harvard* Corresponding author. Tel.: +49-89-70954404.
pump serves as an ‘artificial organ’ with preset flow to calculate the perfusion of the individual organs. At the end of the experiment the animal is sacrificed and the organs of interest are removed and dissected according to predetermined hierarchy schemes. After the individual samples are weighed, the number of MS is determined by different methods depending on the type of MS used. At present there are three types of MS available: radioactive labeled (RM), colored (CM) and fluorescent (FM) microspheres. The first regional blood flow measurements using the microsphere technique were done in the
0169-2607/00/$ - see front matter © 2000 Elsevier Science Ireland Ltd. All rights reserved. PII: S 0 1 6 9 - 2 6 0 7 ( 9 9 ) 0 0 0 2 4 - 3
12
E. Thein et al. / Computer Methods and Programs in Biomedicine 61 (2000) 11–21
fetus of sheep in utero by Rudolph and Heyman [1] using RM. These MS are labeled with radioactive isotopes, their radioactivity is proportional to the number of MS. The radioactivity of the organ tissue samples is measured in a g-counter as counts per minute. Soon after their introduction RM became the standard tool to determine organ blood flow. But the RM method does have certain disadvantages: Due to the short half life of most of the isotopes, RM are not suitable for chronic experiments. The researcher is in permanent exposure to radiation. Because the research-personal is working with radioactive substances, specialized training and facilities are required. The cadaver and the organ samples have to be treated as radioactive waste, of which the disposal is increasingly expensive. Because of these disadvantages Kowallik et al. [4] and Hale et al. [5] introduced colored microspheres (CM), which are polystyrene beads, containing different colors. CM eliminated the problems associated with radioactivity, but this method has its own disadvantages. While using the CM-method the number of microspheres present in the sample is not measurable from outside the organ samples, as it is with the RM-method. Therefore, the CM have to be isolated from the tissue samples, so that the intensities of the colors can be measured. The CM are recovered from the sample by digestion of the tissue in a mixture of chemical and enzymatic liquids [4], followed by the separation of the MS from the digestion fluid through several steps of centrifugation, washing and aspiration of the supernatant [6,7]. This method is known as the sedimentation method. The intensity of the colors is measured with a photometer after they have been released from the MS. In addition to the problem of a loss of MS during the aspiration of the supernatant, the method has the disadvantage that only five colors are available and their spectra are not separable when all five colors are used in the same experiment. To cope with the problem of unseparable spectra Glenny [8] introduced fluorescent micro-
spheres (FM) in 1993, which contain fluorescent dyes. The recovery of these MS from the organ tissue samples is achieved in a similar manner to that of the CM-method. The tissue is digested for 24 h in 4 M KOH, the digestate is filtered through a 10-mm pore-sized polyester filter, the dyes are released from the spheres and their fluorescent intensity is determined with a fluorescence spectrophotometer. This procedure is known as the filtration method. Since the samples are handled in several different containers this method also bares the risk of a loss of microspheres during processing, consequently leading to false blood flow values [8]. Thus, both the CM- and the FM-methods have clear disadvantages in comparison to RM. They are more labor and time intensive due to the necessity for the recovery of the MS from the tissue; e.g. digestion time using the sedimentation method is 48 h [6,7] and 24 h using the filtration method [8]. The digestion method requires different containers to digest the tissue, and to recover the MS by filtering the digestion fluid. MS may, therefore, be lost during the transfer from one container to another leading to unreliable and invalid results in the calculation of the blood flow. The sedimentation method requires the aspiration of the supernatant to recover the MS. Therefore, MS may get lost during the aspiration. The fluorescent dyes are not stable at non-neutral pH-values. It is therefore necessary to assure that no KOH-residues are attached to the MS, which is especially difficult to achieve in the sedimentation method. We have recently developed a new filter device which allows processing of the sample in a single container throughout the whole procedure of weighing, digesting, rinsing, filtering and releasing the fluorescent dyes from the MS [9]. Since the entire FM recovery can be performed in a single container, it should be possible to automate the whole procedure and to eliminate the disadvantages of the time and labor intensity. Therefore, the aim of this study was the complete automation of the time and labor intensive FM-method.
E. Thein et al. / Computer Methods and Programs in Biomedicine 61 (2000) 11–21
13
2. Materials and methods
2.1. Hardware The newly designed filter device (Sample Processing Unit (SPU)), which was developed at the Institute for Surgical Research (ISR) has been described by Raab et al [9] (Fig. 1). The robot consists of commercially available parts (Zymark, Idstein, Germany) and of components that were designed and constructed at our institute. The Zymark robotic system was supplemented with the luminescence spectrophotometer LS50B and an automated sampling unit (Perkin Elmer, U8 berlingen, Germany). Fig. 2 shows a scheme of how the components are connected. All components of the system except those listed in the next paragraph are fixed to different pie-sections, which are arranged in a circle around the centrally positioned robot (Fig. 3.). The Master-Laboratory-Station (MLS), the deposit stations, all pumps for dilution or suction and the Power-and-Event-Controller are located below the robot’s working area. The four entrance racks (Fig. 3) were designed at the ISR and consist of PVC-plates with a size
Fig. 1. The filtration device ‘Sample-Processing-Unit’ (SPU). The SPU (1) consists of three stages; the filter (2) the filter holder (3) and the sample tube (4) in which the fluorescent dyes are collected. During digestion of the tissue samples the beaker is closed with a Delrin cap (5). The sample tube can be closed with a screw cap (6)
Fig. 2. The connection of the different components of the system. The System V-Controller (3) is connected to the AS90/91-Controller for direct communication between the robot (6) and the automated luminescence spectrometer LS50B (5). One computer (1) is used to control the robot. The only modules of the robot shown in this figure are the barcode reader (8) and the balance (7) because they are the only components which need to deliver data to the computer (1). The other computer (2) is used to control the LS50B and to store the data of the fluorescence measurement.
of 21.5 × 35.5 cm, on which aluminium pipes are vertically fixed in a 8× 5 grid pattern 0.3 cm apart. The aluminium pipes are 7 cm high and have a diameter of 3.5 cm. The PVC-plates are mounted on frames which are fixed to the pie-sections. Each rack is capable of holding 40 SPUs. The core of the whole system is the filtration and wash station. It was newly constructed by Zymark according to our design. A sled, which can be moved up and down pneumatically along a track, is fitted to a vertical column. This sled carries a shower head (ISR), which is used to rinse the filters with buffer and to handle the filters throughout the whole procedure. To handle the filters a small part of the top of the shower head is spreadable (Fig. 4). If a filter is to be removed from a container, the robot positions the container including the filter so that the closed shower head protrudes into the filter. When the head is opened two spikes which are fixed to the shower head are pressed into the filter’s wall. Thus, the filter is held by the shower head and by moving the container downwards the robot can remove the filter from the container.
14
E. Thein et al. / Computer Methods and Programs in Biomedicine 61 (2000) 11–21
Fig. 3. Overview of the robot. The central positioned robot is surrounded by the different components that are needed for the procedure. (1) entrance racks; (2) pipetting station with vortex (arrow); (3) cap dispenser; (4) separation station; (5) heater blocks; (6) centrifuge with belonging rack (arrow); (7) digital balance; (8) autosampler of the luminescence spectrometer; (9) computer for the robot. To the right of the robot a part of the luminescence spectrometer can be seen.
The second function of the filtration and wash station is to filter the digestate through the filter’s mesh work after the digestion of the sample. To do this the filter is pressed down by the shower head onto a suction-mug, which is placed at the lower end of the column and which is connected via a reception container to a vacuum. With a negative pressure of 400 kPa the digestate is drawn through the filter within about 5 – 10 s. The third function of the filtration and wash station is the rinsing of the filter. Since it is important to rinse the inner walls of the filter the shower head is laterally perforated, so that the buffer is sprayed outward onto the wall. The rinsing of the filter from the inside assures a neutral pH-value in the contents of the filter. Furthermore, the filters are washed in the buffer from the outside to remove any KOH residues at the same station. To do so the robot takes the filter after it is rinsed from the shower head and swivels it for 10 s in a container which is placed next to the suction-mug. This container (ISR) is equipped with an inlet at its bottom and an outlet at its top. Fresh buffer (10 ml) is added
to the container through the inlet via the MLS before each washing procedure to assure that there is no KOH in the container.
Fig. 4. Shower-head. The shower-head (1) is laterally perforated to rinse the walls of the filter which is necessary to wash out all KOH-residues. A smaller part (2) of it is spreadable and a spike (3) is screwed into it. This part, including the spike is used to remove the filters from any container.
E. Thein et al. / Computer Methods and Programs in Biomedicine 61 (2000) 11–21
For the digestion of the tissue samples the filters are placed into high-grade steel containers (beaker) that were designed at the ISR and consist of a height of 6.5 cm and a diameter of 3 cm. The heater block (Perkin Elmer) used is capable of holding 40 beakers. It is made of aluminium, which is coated with PVC. In the heater block the digestion of the tissue samples at a constant temperature of 60°C takes place. Since it is necessary for the robot to retrieve tubes from a specific position in the centrifuge, the centrifuge is not only equipped with a motor for the rotation, but also with a stepping motor. The robot uses this stepping motor to move the rotor to the position where the desired tube has to be located. Once there, the tube to be retrieved can be grasped by the robot and removed from the centrifuge. The rack (ISR) of the autosampler (Perkin Elmer) of the measuring unit is a PVC-plate (size 21.5× 35.5×3.0 cm) with three cavities for the sample tubes. One is for the sample tube, whose contents are to be measured, and the other two are for exchanging the measured sample tube for the next sample to be measured. The pipette station (Zymark) is connected via two hoses with two syringes of the Master-Laboratory-Station (Zymark) for the pipetting of the organic solvent and isopropanol. Furthermore there are two connections to pumps, one for the pipetting of KOH and the other to suck off KOH-residues from the beakers. A capping station (Zymark) is used to separate the filter holder including the filter from the attached sample tube containing the fluorescent dye before the measurement of the dye-solution. For this the robot places the SPU between two jaws that are located on a platform that can be rotated. The jaws close and fix the sample tube in this manner. Another pair of jaws moves over the filter holder, grasps it and while the lower platform starts rotating, moves up again. In this way filter holder and sample tube are separated. The cap-dispenser (Zymark) used is capable of holding 400 screw caps which are needed to close the sample tubes after the fluorescent measurement is completed. It consists of four pipes of 130 cm height and a diameter of 2.5 cm. The lower
15
Table 1 Components of the robotic system Commercial apparatus (Zymark) XP Z830 ZP411 Z710 ZP510 Z620 CP
Robot Power and event controller System V controller Capping and separation station Centrifuge station Master laboratory station Vortex station Cap dispenser Hand B
Commercial apparatus (Perkin Elmer) AS-90/AS-91 Controller AS-91 Autosampler Dil-1 Diluter station LS50B Luminescence spectrometer Heater blocks Commercial apparatus (Data logic) DS40 Bar-code reader Commercial apparatus (Mettler) PM200 Digital balance Apparatus designed at the Institute for Surgical Research Filtration and wash station All racks Wash container
end of the tubes are open over a platform, from which the robot gets the caps. The caps are piled up inside the pipes. If one of the pipes is empty, the robot realizes via the strength of its hand’s grip that there is no cap under this pipe on the platform. It then moves to the next pipe and gets a cap from there. A digital balance (Mettler, Giessen, Germany) is used to determine both the weight of the empty SPU and the weight of the SPU containing tissue samples. Both weights are stored in an Excel database automatically. The weight of the tissue samples is determined by subtracting both weights. The single SPUs are identified with the help of a bar-code reader (Datalogic S.P.A., Lipo de Calderara, Italy) that reads the bar-code with which the sample tube of the SPU is labeled. All components of the system are listed in Table 1.
16
E. Thein et al. / Computer Methods and Programs in Biomedicine 61 (2000) 11–21
3. Software The program used to control the robot is written in a Zymark Assembler Code. The system is divided into predefined pie-sections, onto to which the individual hardware components of the robot are fixed. All properties of each component are defined in the computer program. The different coordinates the robot needs for correct positioning are defined with the help of a remote control. With this remote control the robot’s hand is moved to the desired position and the coordinates are stored in the program. The precision of the positioning of the hand is 1mm. Besides the coordinates of the different components, clear and wait positions for each procedure are defined in the program. The program is designed as a pipeline system, meaning that the program is divided into nine subsections called ‘Luos’. During each work cycle the robot brings a sample into the next Luo. Therefore there is one sample entering and one sample leaving the robotic system at each cycle. Error routines are present in each procedure. The program compares defined coordinates as well as the strength of the hand’s grip with the actual data sent from the robot. If the actual and defined data diverge the system is stopped automatically and a correction routine is presented to the user. The robot’s program is connected to an Excel database (Microsoft, Seattle, WA, USA) via a program called Easylink (Zymark) allowing for the storage of data such as the bar-code or the weight of the SPU that are determined during the procedure.
4. Procedure for automated sample processing
4.1. Preparation The sample units are labeled with bar-codes and the SPUs are put together. After they are placed into the entrance racks, the robot is started and the SPUs are tarred. For this the robot takes the SPU from the entrance rack, reads its barcode and weighs it. Both the tare-weight and the
bar-code are stored in an Excel-sheet through a program called Easylink (Zymark). This program is needed to create a connection between the robot and Excel via the power and event controller. The bar-code is used to identify the samples. After the SPUs have been tarred the organ samples are placed into the filters and the SPUs are closed with caps (ISR) made of Delrin, which is inert to almost all chemicals. This is the final step which has to be done by hand.
4.2. Digestion The number of samples to be processed is entered into the start-menu and the robot is started. The Delrin cap from the first SPU is placed upon a special deposit station by the robot and the SPU is weighed again in the previously described manner to determine the weight of the tissue sample. When weighing is completed, the robot brings the SPU to the filtration and wash station where the filter is separated from the filter holder. The filter holder and attached sample tube are placed back into their original position in the entrance rack. The robot then gets a beaker from the heater rack, moves to the filtration and wash station, and places the filter into the beaker. In the next step the robot brings the beaker and filter to the pipetting station, where a maximum of 15 ml of 4 N KOH (the amount depends on the weight of the sample) and 1.5 ml of isopropanol are added into the filter The beaker and filter are placed back into the heater block and are capped with the Delrin cap to avoid evaporation of the liquid [9]. The organ sample is digested for 6 h at a constant temperature of 60°C. Excluding the period of 6 h for the digestion of the tissue, the robot needs 12.5 min to completely process a sample. Therefore the robot repeats the above described procedure every 12.5 min with the next SPU.
4.3. Filtration and washing When the digestion period of 6 h for the sample is over, the robot takes the Delrin cap off the
E. Thein et al. / Computer Methods and Programs in Biomedicine 61 (2000) 11–21
beaker and drops it into a waste container. It takes the beaker off the heater block and brings it to the filtration and wash station, where the beaker and filter are separated. The robot then moves the beaker to the pipetting station where the remnants of the digestion fluid are aspirated. After this is completed the beaker is replaced into the heater block where it remains uncapped and ready for the next filter. Concurrently the filter is lowered onto the suction mug and the digestate is filtered. The filter is then rinsed twice by sending 15 ml of a phosphate buffer through the shower head. Following this, the outside of the filter is washed for 10 s by swiveling it in a container which is filled with the above mentioned buffer. To avoid the pollution of the buffer inside the container by KOH-residues, 10 ml of fresh buffer are added to the container before each washing procedure. The filter is then put into a 50-ml centrifugation tube. In order to prevent imbalances of the centrifuge while drying the filter, the centrifugation tube and the filter are weighed before they are centrifuged. If the weight is above the preset limit of 17 g, meaning that the filter is not completely evacuated, the unit is brought back to the filtration and wash station where the filter is removed from the centrifugation tube and put back into its SPU in the entrance rack. Samples that are singled out in this manner have to be processed further manually. If the filter has been successfully evacuated the centrifugation tube including the filter is put into the centrifuge and the filter is dried by centrifugation at 2800 rpm for 1 min. This procedure of rinsing, washing and drying the filter is repeated once again to assure that all KOH-residues are removed and that all FM are collected on the filter. The filter is then removed from the centrifugation tube at the filtration and wash station and placed back into the corresponding SPU.
4.4. Dye reco6ery In order to release the fluorescent dyes from the FM they have to be dissolved.
17
The robot moves the SPU into the pipetting station where 1 ml of organic solvent (cellosolve) is pipetted into the filter. The SPU is then vortexed softly for 30 s. The robot returns to the pipetting station where another 1 ml of the solvent is added to the filter. After another 30 s of soft vortexing the SPU is placed into the centrifuge where the fluorescent dye solution is extracted from the filter by spinning the SPU for 1 min at 2800 rpm. The fluorescent dye has now been collected in the sample tube. The SPU is taken from the centrifuge and brought into the capping station, where the sample tube is fixed between two jaws. The gripper of the capping station then moves over the top of the SPU, grips it and separates the filter holder including the filter from the sample tube by moving up again.
4.5. Fluorescence measurement The sample tube with the dye-solution is gripped by the robot and released by the jaws. The robot then places the sample tube into the autosampler (Perkin Elmer). An electronic signal sent from the robot to the controller of the measuring unit starts it. The pipette of the autosampler is driven over the sample tube, 550 ml of the dye are sucked into the pipette by a syringe of the diluter station and transferred into the flow cuvette of the LS50B. The fluorescent intensity is measured and stored on a F1 twinlab-default database (Perkin Elmer) automatically. During the measurement the robot retrieves the filter holder and filter, which are still held in the gripper at the capping station, and disposes them in another waste container. After the measurement of the sample the whole pipe system of the diluter station and the autosampler are rinsed with 10 ml of 100% ethanol and then dried for 20 s with air by a positive pressure of 3 bar. When this is finished, the controller of the measuring unit sends an electronic signal to the robot and thus directs the robot to retrieve the sample tube from the autosampler. The sample tube is again moved to the capping station, where it is placed between the two jaws. The robot then gets a screw cap from
18
E. Thein et al. / Computer Methods and Programs in Biomedicine 61 (2000) 11–21
the cap dispenser and presses it onto the sample tube. The jaws start rotating and the sample tube is capped. The closed sample tube is gripped by the robot and finally dropped into a collection box.
4.6. Calculation of the blood flow The intensities of fluorescence of the tissue samples and of the arterial blood reference are used to calculate the sample blood flow, according to the following formula: Orgflow = Corg(Pf/Cref) where Orgflow is the blood flow to the tissue sample in ml/min; Corg is the number of MS in the organ; Cref is the number of MS in the reference blood-sample; Pf is the flow-speed of the Harvard-pump Since all of the data are stored on an Excelsheet the blood flows can be calculated quickly and easily.
5. Experiments Experiments were performed to determine precision and reliability of the robot, as well as the stability of the fluorescent dyes for both dyes enclosed in the FM and dyes released from the FM.
5.1. Experiment 1 To test the precision of the pipetting of the organic solvent, a software program was written allowing to use the pipetting station repeatedly. Organic solvent was pipetted into a 50-ml centrifuge tube in 50 individual steps (1 ml each) by the robot. Following each step the robot weighed the tube and the weight was stored. The same procedure was done manually, to compare these results with the results of the robot.
5.2. Experiment 2 To determine if the automated procedure leads to a loss of fluorescent intensities the following
experiment was performed. One ml of each of the seven available types of FM-solutions was added to 20 ml distilled water. Of this solution, 40 ml or about 2000 FM were pipetted into 340 filters manually. These filters were processed by the robot and measured by the spectrophotometer according to the protocol for tissue samples. The same number of FM was pipetted manually into 25 glass test tubes and dried in an incubator at 60°C for 24 h. The pipetting of the organic solvent for this control group was performed by the robot, to avoid the inaccuracies of manual pipetting. As the FM and the fluorescent dyes of this group do not come into contact with KOH and there can be no loss of FM, the standard deviation of this group reflects the pipetting error of the manual pipetting of the FM-solution.
5.3. Experiment 3 An experiment was performed to assure that system breakdowns which remain uncorrected for a long time do not harm the fluorescent dyes inside the FM, in particular for those samples which stay in the heater blocks for a prolonged period. Of a red FM solution, 50 ml were pipetted into 60 filters and processed by the robot according to the protocol for tissue samples. When 30 filters were completely processed and the fluorescent intensity of the dye content of these samples was determined, a hardware failure was simulated by switching off the centrifuge. The 30 filters which were not completed were left in the heater block for 24 h and were then processed manually. The fluorescent intensities of both groups were compared using the students t-test
6. Results
6.1. Experiment 1 The density of cellosolve-acetate is 0.9725 g/ml. Pipetting exactly 1 ml should therefore result in an increase of the weight of the test tube of 0.9725 g per each pipetting step. The mean weight of 50
E. Thein et al. / Computer Methods and Programs in Biomedicine 61 (2000) 11–21
individual pipettations processed by the robot is 0.964 g, with a standard deviation of 0.003 g, a maximum of 0.972 g and a minimum of 0.957 g, whereas the manual pipettation results in a mean of 0.948 g with a standard deviation of 0.018 g, a maximum of 0.964 g and a minimum of 0.886 g. The robot is able to pipet the organic solvent more precisely and constantly than that could be done manually.
19
vortexing 0.5 min, second pipetting of the organic solvent 0.5 min, second vortexing 0.5 min, centrifugation 2 min, measurement (seven colors) 6 min. The total time that is needed to work one sample manually is at least 18.5 min, whereas the robot needs 12.5 min for the same procedure. (Table 2) A total of 6 min are saved with the automated method for each sample, which is significant when greater numbers of tissue samples are processed.
6.2. Experiment 2 The mean fluorescent intensity of the control is 46.26 arbitrary units (AU) with a standard deviation of 5.76 AU (= 12.47%). This standard deviation gives the error of the pipetting of the FM-solution. The mean fluorescent intensity of the group which was processed by the robot is 47.98 AU with a standard deviation of 4.78 AU (= 9.97%). This result indicates that the robot is processing the samples reliably and that there is no loss of FM or fluorescence in the dye solutions due to residual KOH.
6.3. Experiment 3 The mean fluorescent intensity of the filters which were processed under normal conditions is 27.53 AU with a standard deviation of 1.77 AU ( = 6.42%). The mean fluorescent intensity of the filters which remained in the heater block for 24 h is 26.56 AU with a standard deviation of 1.25 AU ( = 3.76%). As the results show, prolonged digestion times caused by system breakdowns do not lead to a loss of fluorescent intensity at least as long the dyes are still enclosed in the FM.
6.4. Comparison of time consumption manually/automatically To work one sample manually the following times are needed for the different steps: dissection and weighing 2 min, pipetting of the digestion fluid and isopropanol 1.5 min, filtering, rinsing and washing 3 min, centrifugation 2 min, first pipetting of the organic solvent 0.5 min, first
7. Discussion The system described here represents a complete automation of the FM-method which is accurate, time saving and reliable. In order to achieve this aim several modifications of the hard- and software were necessary to assure that the system completes the different steps of the procedure without mistakes. Using glue to connect any of the robots hard ware components appeared to be a problem. Due to the chemical vapors and the high temperature these connections dissolved and resulted in system breakdowns. The fingers of the robot are now coated with PVC-hoses, because of the problem that four PVC-pads that are glued into the finger’s inside were lost during the procedure. The loss of the pads resulted in a divergence in the distance between the fingers and the one defined in the software, which resulted in the breakdown of the system. Table 2 Comparison of time consumption for 1000 samples FM automatically FM manually RM Weighing/ sample Digestion Processing/ sample Measurement/ sample Total time
Automatically
2 min
2 min
360 min 12.5 min
360 min 14.5 min
None None
online
4 min
3 min
214 h
308 h
83.33 h
20
E. Thein et al. / Computer Methods and Programs in Biomedicine 61 (2000) 11–21
Emery paper was originally glued to the top of the shower head in order to increase the friction when the filters are removed from the different containers. This connection was dissolved, leading to the loss of filters and consequently to the stop of the routine. Therefore, the emery paper was replaced by two pointed screws, which are fixed to the shower head with a thread. This assures that the removal of the filters is done properly and this connection is not affected by the chemicals used. A major difficulty was to find a workable design for the different racks. All of the racks used now consist of vertically positioned, uninterrupted aluminium pipes of 7.5 cm height and a diameter of 3 cm. This design guarantees permanent guidance of the containers when they are placed into the racks. The permanent guidance is necessary due to the fact that vortexing and centrifuging may lead to non-vertical positioning of the containers. Because of the fact that the robot is not able to correct these inaccuracies by itself, the rack must be designed to avoid this problem when the robot attempts to place the tubes into the rack. The uninterrupted aluminium pipes serve to guide the containers back into a fully vertical position. The dye solutions must have neutral pH since the colors are not stable against non-neutral pHs after the FM have been dissolved. Therefore the entire filter—not only its mesh work — needs to be rinsed well with phosphate buffer. The shower head is therefore perforated centrally as well as laterally, so that the walls of the filters are also rinsed well. To ensure the elimination of KOHresidues the filters are centrifuged before they are rinsed another time. This centrifuging presses all KOH-residues that might still be attached to the filter’s wall onto the mesh-work, from where they are rinsed off during the second rinsing. Total costs of the robotic system, including the automated fluorescence spectrometer and the software are about $280 000. But the robot assisted FM-method saves manpower, time, costs for disposal of radioactive waste and eliminates the personal and environmental risk of harmful radioactive materials.
8. Conclusions Processing the tissue samples—according to the protocol of the FM-method—is completely automated by a robot which is accurate and reliable. It is possible to combine the robotic system with an automated spectrometer, thus allowing the online measurement of the fluorescent dyes. The most important steps of the method, for example pipetting the organic solvent are carried out more precisely and constantly by the system than that could be done manually. The automation of the time and labor intensive FM-method is 33% faster than processing the samples manually. In addition, the robot is able to work 24 h per day, 7 days per week with the same consistency and precision and is therefore able to process a higher number of samples than could be achieved manually. In the unlikely event of a system breakdown there is no loss of fluorescence from the dyes due to the fact that the fluorescent dyes are still enclosed in the FM. Furthermore the researcher is no longer constantly exposed to hazardous vapors of the chemicals used. Thus, the automation of the FM-method as presented here is a valid and promising alternative to both the radioactive method and the manual FM-method.
References [1] A.M. Rudolph, M.A Heyman, The circulation of the fetus in utero; methods for studying distribution of blood flow, cardiac output and organ blood flow, Circ. Res. 21 (1967) 163 – 184. [2] J.R.S. Hales, W.J. Cliff, Direct observations of the behavior of microspheres in the microvascaluture, Bibliotheca Anatom. 15 (1977) 87 – 91. [3] G.D. Buckberg, Studies of regional coronary flow using radioactive microspheres, Ann. Thoracic Surg. 20 (1975) 46 – 51. [4] P. Kowallik, R. Schulz, B.D. Guth, A. Schade, W. Pfaffhausen, R. Gross, G. Heusch, Measurement of regional myocardial blood flow with multiple colored microspheres, Circ. Res. 83 (1991) 974 – 982. [5] S.L. Hale, K.J. Alker, R.A. Kloner, Evaluation of nonradioactive, colored microspheres for measurement of regional myocardial blood flow in dogs, Circ. Res. 78 (1988) 428 – 434.
E. Thein et al. / Computer Methods and Programs in Biomedicine 61 (2000) 11–21 [6] New sedimentation method for determination of organ blood flow with fluorescent microspheres. Product information sheet, Molecular Probes, Eugene, OR, USA, 1996. [7] M.F.M. Van Oosterhout, H.M.M. Willigers, R.S. Reneman, F.W. Prinzen. Fluorescent microspheres to measure organ perfusion: validation of a simplified sample processing technique. Am. J. Physiol. 269 (1995) H725–H733.
21
[8] R.W. Glenny, S. Bernard, M. Brinkley. Validation of fluorescent-labeled microspheres for measurement of regional organ perfusion, J. Appl. Physiol. 74 (1993) 2585 – 2597. [9] S. Raab, E. Thein, A.G. Harris, K. Messmer. A new sample processing unit for the fluorescent microsphere method, Am. J. Physiol. 276 (1999) H1806 – H1807.
.