INOM
EXAMENSARBETE MEDICINSK TEKNIK, AVANCERAD NIVÅ, 30 HP
STOCKHOLM SVERIGE 2020,
Development of a Novel Device for Optimal Sample Blood Volume
Collection from Patients with Sepsis
RASMUS BOLTSHAUSER
KTH
Development of a Novel Device for Optimal Sample Blood Volume Collection from Patients
with Sepsis
Utveckling av ett Instrument för Insamling av Optimal Blodvolym från Sepsispatienter
Master thesis project in Medical engineering 30hp Author: Rasmus Boltshauser
External Supervisor: Volkan Özenci Supervisor at KTH: Maksims Kornevs Examiner: Sebastiaan Meijer
School of engineering sciences in chemistry biotechnology and health Royal Institute of Technology
Kungliga Tekniska Högskolan, KTH SE-141 86 Flemingsberg, Sweden TRITA-CBH-GRU-2020:082
Abstract
When performing sepsis diagnosis, the most important preanalytical variable is blood volume. Too little blood increases the risk for false negatives whereas overfilling causes increased risk for false positives. Even though this fact is known, there are case studies showing that in a majority of tests, the taken blood sample volume is not the recommended amount. As previously tried methods have been limited in their ability to tackle the problem this study aimed at creating a technical device to aid healthcare providers with blood volume sample collection.
As a base, the double diamond approach by the Design Council was used. This design approach splits up the design process in four distinctly different phases (discover, define, develop, and deliver) all using their own methods to aid the creative process.
After completing the discover and define phase it was determined that a non-contact capacitance liquid level sensor could operate as an ideal blood volume sample device. During the development and delivery phase prototypes were created and evaluated. The final results of this work could not give conclusive evidence concerning if a non-contact liquid level sensor could operate as an ideal blood volume collection device. The methodological approach used in this thesis can be used as inspiration for a designer to create a device for a similar or different purpose.
Moreover, information from this thesis can also work as reference material to develop a device to perform ideal blood volume sample collection.
Such a device would have the potential to be an essential part of the everyday workflow in sample collection from patients with sepsis worldwide and would aid in ensures effective and fast diagnostics.
Sammanfattning
När man diagnostiserar sepsis så är den viktigaste pre analytiska komponenten blodvolym. För lite blod ökar risken för ett falskt negativt resultat. För mycket blod ökar risken för ett falskt positivt resultat. Även om detta faktum är välkänt så visar studier att många tester genomförs med en ej rekommenderad blodvolym. Då tidigare testade metoder att bemöta problematiken är av begränsad förmåga, genomfördes denna studie. Studien inriktade sig på att undersöka design och utveckling av ett instrument med förmågan att hjälpa sjukhuspersonal att finna rekommenderad mängd blodvolym. Grunden i arbetet är baserat på design metoden ‘double diamond’ utvecklat av Design Council. Denna metod delar upp designprocessen i fyra olika faser (upptäcka, definiera, utveckla och leverera). Alla delar har sina egna metoder för att underlätta den kreativa processen.
Efter att upptäcka- och definiera faserna var slutförda blev det tydligt att en sensor baserad på indirekt kapacitansmätning kunde va ett idealt sätt att avläsa blodvolym. Under utveckla och leverera faserna skapades ett antal prototyper för att prova ut konceptet. Slutresultatet med det skapade instrumentet kunde inte ge ett definitivt svar huruvida instrumentet kunde användas som en ideal blodvolymmätare.
Den metodologiska processen använd vid detta arbeta kan användas av designers för att skapa en liknande eller annorlunda produkt. Tidigare genomfört arbete kan också användas som referensinformation vid framtagning av en ideal produkt för att underlätta sepsisprovtagning. En sådan produkt skulle ha potential att underlätta och säkerställa att rätt blodvolym tas vid den dagliga medicinska verksamheten.
Contents
Acknowledgement ... vi
Acronyms & abbreviations ... vii
1.Introduction ... 1
1.1 Limitations ... 1
2.Background ... 2
2.1 Process of blood sample volume collection ... 2
2.2 Issues with the current process... 3
2.3 Prior work ... 3
3.Materials and methods ... 5
3.1 Methodology ... 5
3.2 Discovery ... 6
3.2.1 Choosing a sample ... 6
3.2.2 MoSCoW prioritization... 6
3.3 Define ... 6
3.3.1 Secondary research- Literature review... 7
3.3.2 Assignment criteria ... 7
3.4Develop ... 8
3.4.1 Theory in connection to physical prototypes ... 8
3.4.2 Physical prototypes ... 8
3.4.3 Real vs measured volume ... 9
3.4.4 Feedback ... 9
3.5Deliver... 9
3.5.1 Final prototype ... 9
3.5.2 Final testing of the product ... 9
4. Results ... 10
4.1 Discovery ... 10
4.1.1 Choosing a sample ... 10
4.1.2 MoSCoW prioritization... 11
4.2 Define ... 13
4.2.1 Secondary research - literature review ... 13
4.2.2 Comparison to must have features ... 14
4.2.3 Assignment criteria ... 15
4.3 Develop ... 17
4.3.1 Theory in connection to prototypes... 17
4.3.2 Capacitive liquid level sensor ... 17
4.3.3 Shielding and the out of phase liquid level method ... 19
4.3.4 Calibration process ... 20
4.3.5 Setting parameter values for the system... 20
4.3.6 Linear correction algorithm ... 21
4.3.7 Capacitive Liquid level sensor version 1 ... 22
4.3.8 Capacitance level sensor version 2 ... 24
4.3.9 Capacitive liquid level sensor version 3... 25
4.3.10 Capacitance liquid level sensor version 4 ... 26
4.3.11 Programming / code ... 26
4.3.12 Real vs truth test ... 28
4.3.13 Feedback ... 29
4.4 Deliver... 29
4.4.1 Final product ... 29
4.4.2 Final evaluation ... 31
5. Discussion ... 32
6. Conclusions ... 35
Acknowledgement
This project has been a joyful and great learning experience even though work was performed during the COVID-19 pandemic. I feel that this comes down to both a very interesting thesis task and that I got to perform the project with excellent people around me.
I am extremely grateful to my external supervisor Associate Professor Volkan Özenci for providing me which such an interesting task and his help and support during the entire time. Also, for introducing me to the working team at Clinical Microbiology at Karolinska Hospital in Huddinge.
I also wish to thank my KTH supervisor Dr. Maksims Kornevs and my examiner professor Sebastiaan Meijer for introducing the double diamond design approach and a special thanks to Maksim for all the guidance given along the way.
I had the great pleasure of working with Mr. Jan Klingler, Dr. Alicia Wong, and Mr.
Alexander Johnsson at Clinic microbiological, thank you for creating such a fantastic work atmosphere. You made work feel enjoyable. A special thanks to Jan. K for all the discussions and help along the way. Your contribution was pivotal for prototyping and getting the work forward.
Lastly, I would like to thank my family: my parents and my brother for supporting me
spiritually throughout writing this thesis and my life in general.
Acronyms & abbreviations
BSI - Bloodstream Infection
ICU - Intensive Care Unit
BC - Blood Culture
ISDT - Initial Specimen Diversion Technique
TOF - Time Of Flight
CDC - Capacitance to Digital Converter
OOP - Out Of Phase
GUI - Graphical User Interface
DIY - Do It Yourself
IDE - Integrated Development Environment
1. Introduction
Blood stream infections (BSI) happen when a bacterial infection gets into the blood stream. Sepsis is a negative autoimmune response to the BSI attacking the human body (1). Throughout time detection and treatment of sepsis and other forms of BSI have been a large problem for the medical field. With its consistently high rate of mortality around 10 – 40 % dependent on the degree of sepsis, the autoimmune response causes the death of close to 11 million people worldwide every year (1).
When it comes to sepsis the most important factor for lowering the mortality is to reduce time before treatment (2). Due to the fact that this can save lives, it is of great importance to have fast and accurate diagnosing methods.
In suspected cases of sepsis four blood culture (BC) bottles will be sampled from a patient.
Each bottle contains liquid broth as a growth media and should be filled with 8 -10 ml of patient blood to ensure optimal preanalytical setup (3, 4). However, it has been shown that the collected sample volume many times are far from the recommended amount (3). Observing today’s recommended methods for blood collection, a recent study at the Karolinska University Hospital revealed that only 18% of bottles contained the recommended amount of blood (3). This is a major problem, because the concentration of the microorganism causing BSI in the blood tends to be very low. Because of the low concentration, if too little blood is drawn time for diagnosis to complete increases and so does the risk for false negatives (3, 5). On the other hand, if too much blood is drawn, the risk for false positives will go up (4). This harms the quality of the diagnostic method potentially leading to patient harm. To stop this from happening, researchers have previously looked at different methods to provide education and feedback for healthcare providers to improve sample volume collection (6, 7). This approach to lower the impact of the problem can help, but is limited, as hospital departments do not have the time to give out or to receive education.
Comparatively, has almost no work been done trying to find a technical device capable of helping healthcare providers with blood sample volume collection. As such, a device could be helpful in aiding healthcare staff without disrupting the normal workflow or increase workload on other hospital departments. This work researched the development and design of a technical device to aid healthcare providers with blood volume sample collection. With the specific goal in mind, to use found information to create and evaluate a product that could aid healthcare staff with blood volume sample collection. As a basis, the double diamond design approach was used (8).
1.1 Limitations
Due to the COVID-19 pandemic unfolding while work was being performed, some changes to the work were required to be done. The most relevant of these was the inability for the author to commit a field study observing healthcare staff performing blood volume sample collection in the real hospital environment. This was quite unfortunate as being able to talk to and observe the normal workflow and the hospital environment could have been very helpful when trying to design something that hopefully ends up used in the same environment.
2. Background
2.1 Process of blood sample volume collection
The recommended way to get a blood sample volume from a patient in today’s hospitals is done in the following way.
First, the nurse or the phlebotomy team member will begin by rubbing an alcohol- drenched cotton pad over the area where blood is to be collected, usually either the right or the left arm (9, 10). This is done to lower the amount of skin contamination present at the place of injection and a very important part of the process. If too much contamination gets into the BC bottle the diagnostic test will give a false positive read. Every time this happens, the medical sector will be wasting between $1000 - $8,720 in unnecessary antibiotic treatment and additional time invested (11). After this has been performed, a butterfly needle is used to first perform initial specimen diversion technique (ISDT) which is the process of diverting and containing the initial 1-2 ml of blood in a separate container before BC sampling to get rid of skin contaminants (12). After ISDT has been performed, blood is collected through the same butterfly sampling kit (9, 10).
Already before filling the BC bottle contains a liquid in the form of nutrient-rich broth allowing the, if present, microorganism causing the BSI to grow. To make sure the volume of blood collected is correct, the standard method today is to use a graduated scale placed on the bottle’s label that shows volume in 5 ml increments. Before filling the BC bottle, the healthcare provider uses this scale to measure out 10 ml above the already existent liquid surface line. This place is then marked with a pencil, since 10 ml of blood is the recommended amount in every BC bottle to ensure ideal sample quality (3-4,9-10). This process, however, can be quite cumbersome in the already hard-pressed hospital environment especially during night shifts or when there are many patients in urgent need. The process is also complicated by the fact that the pre-filled liquid surface level and label position differs on each bottle. This causes the healthcare provider to heavily rely on ocular inspection to find the 10 ml above the surface level. Figure 1 shows, how the marking process is to be performed.
Figure 1: Usage of the graduated label scale to find the 10ml above the surface level mark.
10 ml above liquid level.
Liquid level.
A total of four BC bottles is to be collected from every patient and then sent to a microbiology laboratory where the diagnostic process continues.
2.2 Issues with the current process
It has for a very long time been a well-known fact that the more blood you collect, the bigger the chance for a positive diagnosis when using blood culture diagnostic systems (4, 13). It is also known to be the most important pre-analytical variable to ensure good sample quality (3, 14).
Observation of different care facilities has shown that the actual volume of blood sampled many times are far from the recommended amount (3, 15). In a study performed at the Karolinska University laboratory a total of 10328 BC bottles had their volume measured to explore how well recommended guidelines are followed (3). The mean volume collected was 8.60 (±4.29) ml, which sounds promising, but a deeper look at the data of the collected BC bottles revealed that only 18.24% of the bottles contained the recommended volume. The vast majority, 81.76% contained too much or too little blood sample volume (3). Especially worrying was the fact that up to 47.10%
contained below the lower recommended limit of 8 ml of blood. A breakdown of the collected sample volumes can be seen in Figure 2.
Figure 2: Histogram breaking down blood sample volume collected in BC bottles at the Karolinska University Hospital. 2 ml bins with n = 10 328 (3).
Factors affecting the blood volume collected were gender, age and time of the day. Female patients had a lower mean blood volume collected, 8.36 (±4.28) ml versus 8.78 (±4.30) ml. Age and blood volume collected were negatively correlated and a lower blood volume was taken from patients that had their blood collected from 00:00 to 07:59 (3).
2.3 Prior work
Recently, researchers recommended a bundled approach for improvement of diagnosis and treatment of patients with sepsis (16). The general idea of the bundled approach is to define and improve different phases of the diagnostic workflow from the preanalytical phase to the post-
analytical phase (16). In this article, authors Lamy et al. (2020) thinks post-analytical quality could be improved by establishing more effective education and feedback methods for nurses and phlebotomy staff members (16). This to make sure recommended guidelines for blood volume sample collection are followed.
An education approach to fix the problem has already been tried, generally displaying good results (6-7). In both the studies cited, before education was employed, hospitals were consistently underfilling bottles. After an initial increase in education and by providing consistent feedback the average bottle volume came in-line with the recommended amount (6-7). However, as noted in the article by Lamy et al. education does not per se make sure that the problem of underfilling goes away (16). There is a limit to how much education can help and at the same time, it impedes on the medical staff’s natural workflow adding on extra stress. It also requires extra work for the laboratory departments that would become responsible for both information gathering, education and feedback to the on-sight departments. As talked about in the article by Khare et al. (2020) it was of great importance to gather data so departments not following the guidelines could be found (6). After being identified the specific underperforming department could be improved by consistent feedback. This means that education should not be treated as a onetime fix to the problem as the problem will come back if consistent feedback is not given (6). Moreover, as was found in the study at the Karolinska University Hospital. Even when the average volume in the BC bottles are close to the recommended amount, underfilling and overfilling can still be a problem (3).
As such, this work aimed to develop a technical device to improve the sample quality without impeding the natural workflow of the healthcare staff or add on extra work for microbiology departments.
3. Materials and methods
3.1 Methodology
As a base the double diamond design approach was used to develop the product. The double diamond approach is a simple, clear and visually intuitive methodology for creating new products established by the design council in 2004 (8). A visual representation of the approach is shown in Figure 3 below.
Figure 3: The double diamond approach where the creation of a product is done in four distinct phases:
discovery, define, develop and deliver.
The Double diamond approach is naturally split up in to 4 different phases that is to be performed when creating a new product, namely (i) discovery; (ii), define; (iii), develop; and (iv) deliver.
From the Design council web page there is a short bullet list explaining every part of the process.
i. “Discover. The first diamond helps people understand, rather than simply assume, what the problem is. It involves speaking to and spending time with people who are affected by the issues.
ii. Define. The insight gathered from the discovery phase can help you to define the challenge in a different way.
iii. Develop. The second diamond encourages people to give different answers to the clearly defined problem, seeking inspiration from elsewhere and co-designing with a range of different people.
iv. Deliver. Delivery involves testing out different solutions at small-scale, rejecting those that will not work and improving the ones that will.” (8)
Against what one might think the process is not intended to be perfectly linear. For example, a great way to discover parts of the problem is when a new dimension to the problem shows up while developing a product (8). In these cases, it is necessary to go back and perform more definition and discovery work and then try to develop a new solution to the redefined problem.
3.2 Discovery
During the discovery phase, the methods of ‘choosing a sample’ and ‘MoSCoW prioritization’
were used. Using what was found by applying the choosing a sample method and by talking to stakeholders in the product, a summary of requirements that the solution needed to fulfill was created in the form of a MoSCoW prioritization table.
3.2.1 Choosing a sample
Choosing a sample to operate and test with was simple as all the technical parts being used today (described on page 10 under the title “choosing a sample”) are standardized for medical use.
Because of this, parts making up the system cannot be changed to fix the problem. By having a clear image of the state of the art system, it was easier to understand what different solution methods could operate with (17).
3.2.2 MoSCoW prioritization
A MoSCoW prioritization is a table created to more clearly define and discover what stakeholders want in a product. This by splitting the product, into requirements and features where, requirements are unclarified demands put on the product and features are clearly defined tasks that prove that a solution can fulfil a specific requirement. Features in the MoSCow table are placed in different brackets depending on how hard they are to fulfil and to what degree they are important for the product to satisfy the requirement. The different brackets are as follows; must have, should have, could have, will not have (18). For a potential solution to be considered it needs to fulfil all the must have features and most of the should have features. Depending on how much time remains after fulfilling the must have and should have features, the designer can move on to fix the could have and will not have features (18).
After initial discussions with team members at clinical microbiology Karolinska Huddinge at the start of the project work, an initial MoSCoW prioritization table was created by the author.
After that the author and stakeholders in the product meet in a preplanned web meeting where the author presented how MoSCoW prioritization is performed. After presenting the method the table created by the author was shown and updated with the help of the people present in the meeting. Requirements and features where changed and updated until the table, was considered satisfactory in defining an ideal device for solving the problem.
3.3 Define
During the define phase, ideas and information found earlier in the discovery phase was used to better define the problem at hand and to take forward potential solutions for the problem. To do this, a literature review was performed to find systems capable of solving the problems. After different solutions were collected, an assignment criteria list was created so that the different ideas could be placed in a hierarchy (19).
3.3.1 Secondary research- Literature review
Using all the information taken from the discovery phase, the problem and what kind of device that needed to be created had become clearer. A literature review was made to find potential technical solutions able to perform the intended task. First, an initial search was made on Google to find general methods of measuring the volume inside a tank. Initial search terms - low flow measurement review, liquid volume methods review.
After finding two broad modus operandi for volume measurement, a more in-depth search was performed so that all different measurement methods relating to the modus operandi could be found. This search used the following terms. In Primo - review of flow meter systems, water level sensor review low cost.
After the performed search, different measurement methods were compared to specific requirements that our system had, as defined in the MoSCoW prioritization. For example, any method invasively measuring the volume would not be suitable here as it would not fulfil a requirement connected to contamination risk.
3.3.2 Assignment criteria
With all the ideas in place, it was important to create some form of an idea hierarchy so focus could be placed on the most promising solution. This was done using an assignment criteria list where 6 different criteria were to be rated for all the different solution methods. The criteria were;
fabrication possibility, technical possibility, clinical possibility, expected usability, environmental impact and budget possibility. A bullet point list below defines the meaning of the different criteria in terms of this dissection (19).
• Fabrication possibility - How easy it will be to get the material needed for prototyping and creating the product.
• Technical possibility - If the technical know-how is great enough and if adequate documentation exists to create a prototype and eventually a product.
• Clinical possibility - If the solution would be suitable in the clinical environment.
• Expected usability - If the device could easily be integrated to the workflow without interrupting how it is done today.
• Environmental impact - If the device leaves a big footprint on the environment.
• Budget possibility - Could we create and sell the product for a price considered in line with what a medical ward would pay for it.
The author used earlier found information about the different solutions to score the solutions in the following 5 criteria; fabrication possibility, technical ability clinical possibility, expected usability and environmental impact.
The budget possibility score was decided by first having the head of the work team at Karolinska Huddinge giving a monitory range where a multiple use product could be around. After this range was set, both the author and a design expert working in the team, examined if the products could be made in that price range. After all scores were set, members in the work group got to see the list and give input if the list seemed accurate or not.
3.4 Develop
To get practical knowledge regarding devices operating with the chosen measurement method, prototypes were created and evaluated. After a prototype with satisfactory ability was shown. Feedback was taken in so that a final prototype could be created.
3.4.1 Theory in connection to physical prototypes
To make it possible for a reader to understand how created prototypes operate a theoretical background chapter was written.
3.4.2 Physical prototypes
Prototypes of the most suitable solution were planned and constructed. After the prototypes were realized, they were tested and evaluated to see if they would be able to work as a real product (20).
3.4.2.1 Capacitive liquid level sensor version 1
To further understand if a capacitive liquid level sensor could operate with the bottle, a straightforward capacitive liquid level sensor was prototyped. The capacitive liquid level sensor follows an earlier Arduino project that is available at: (21).
Material used in the setup were two strips of copper adhesive tape functioning as sensing electrodes, one Arduino Uno working as a microcontroller, an FDC 1004 protocentral breakout board operating as a capacitance to digital converter (CDC) and two copper wires. To hold up the electrodes towards the BC bottle a plastic container was used.
3.4.2.2 Capacitance liquid level sensor version 2
The second capacitance liquid level sensor prototype used a specific method known as the out of phase (OOP) liquid method that employs a specific electrode layout to increase system robustness. More in deep information regarding the OOP method can be seen in the result section under the title “out of phase capacitance liquid level sensor” at page 19. The system used the same parts as the first prototype and hade an additional six copper adhesive tapes added to the system. Moreover, an electrical isolation layer made up of electrical tape was added to the system.
Electrical isolation tape was also placed over the wiring and the FDC1004 input pins to negate disturbing electrical signals.
3.4.2.3 Capacitive liquid level sensor version 3
After the second capacitance level sensor showed promising results, a product that could operate in the ICU environment was created. The sensor used the same electrode setup as before with the Arduino Uno controlling the FDC1004 that sensed the capacitance with copper adhesive tape operating as electrodes. The plastic container holding up the electrodes was removed and replaced with a 3D printed plastic model able to contain the electrical components inside two compartments and had two plastic pillars that could hold the copper electrodes towards the BC bottle. The prototype was tested using a cap adapter attachment normally used when a blood volume sample is collected.
3.4.2.4 Capacitance liquid level sensor version 4
Two big problems were found in the third capacitance liquid level sensor prototype. The first one was a fitting problem between the electrodes and the cap adapter attachments that is normally used while filling the BC bottles. The second problem was that as the system had changed from the earlier models the accuracy had become inadequate. To address these problems, another capacitance liquid level sensor prototype was created. Where the electrodes were fitted tighter to the bottle itself by, changing the 3D printed electrode holder. Otherwise the prototype operated with the same equipment and setup as the third one.
3.4.3 Real vs measured volume
For all the different prototypes a real vs measured volume test was made to estimate the accuracy of the prototype. To keep environment conditions constant, the BC bottle was first filled and then emptied before starting a test. The real vs measured volume test was performed by filling the emptied BC bottle in 10 ml increments until the edge of the liquid level sensors was reached.
After every increment, theoretical volume added to the BC bottle according to the sensor was written down. A syringe was used to make sure that the volume added to the BC bottle at every increment was exactly 10 ml. For the test made with the first prototype no linear correction was used while measuring but added to the measurement result later by mathematical estimation to improve consistency in the data being displayed.
3.4.4 Feedback
Feedback concerning the last prototype was gathered from different team members and the author himself. This feedback was used to improve the prototype in order to create a final prototype.
3.5 Deliver
By using the received feedback, the final prototype was created and later evaluated.
3.5.1 Final prototype
The final prototype functioned much like the fourth prototype operating with the same electrode setup and the same 3D printed electrode holders. However, more things were integrated into the prototype to better the working system following the received feedback (22).
3.5.2 Final testing of the product
The last prototype solution was moved away from its place of fabrication into an environment better resembling where it is meant to be used. Doing this allows a designer to get a feel for how the product would operate in a realistic work situation and can give valuable insight on how the product would realistically perform. A standard operation kit, following the system defined in the
“choosing a sample” title, was used to fill four BC bottles containing growth medium. The bottles were filled by a member of the clinical microbiology team while the author operated the device.
(22)
4. Results
4.1 Discovery
4.1.1 Choosing a sample
The standardized sampling kit intended for one-time use, consists of a butterfly needle, a tube going from the needle to a cap adapter attachment that is, subsequently connected to the BC bottle when blood is to be sampled. The butterfly needle is 20 mm long and will cause blood to flow as an effect of the negative pressure inside the BC bottle. The tube connecting the butterfly needle to the cap adapter has an inner diameter of 3,1 mm and is quite short, being only 190 mm long. At the other end of the tube is a rubber isolated needle, 20 mm long, which is used to break a plastic membrane on the top of BC bottle. The adapter itself is not truly a part of the standard kit but instead offered by a different company. The cap adapter attachment has a diameter of 36 mm and is 45 mm long. The top member on the BC bottle has a diameter of 9 mm and the bottle itself has an outer diameter of 34 mm and an inner diameter of 31 mm. In total the BC bottle can contain around 80 ml of fluid. Below in Figure 4 the full standard kit and all parts are showcased.
Figure 4: 4A, butterfly needle and tubing. 4B, cap adapter attachment. 4C, BS bottle 4D, top rubber membrane. 4E, rubber isolation layer and needle.
4.1.2 MoSCoW prioritization
The following MoSCoW table was created following the method described under the title
“MoScoW prioritization” page 6 in the method chapter. Potential solutions can be analyzed with the use of the MoSCoW table (Table 1). For example, if they cannot perform all the must have features, they are considered unfit as a solution for the problem. If they, apart from fulfilling all the must have features, can perform many of the should, could and will not have features, the solution would be optimal for this intended task.
Table 1: MoSCoW prioritization table categorizing what a potential solution needs to fulfill in terms of requirements and features.
Requirement Must have Should have Could have Will not have Fluid type Works with blood. Can be made to
work with different fluids with the help of recalibration and system changes.
Any user of the system can change what liquid it can operate with.
System can Self- calibration dependent on what liquid is measured.
Accuracy/drift Within 2 ml range. No measurement drift while operating.
Below 1 ml error. -
Workflow compatibility
Can handle hospital environment changes in temperature, humidity and electrical disturbances.
Can handle human handling without risk of breakage or loss of measurement ability.
Does not change the normal workflow of medical staff.
Maintain full measurement ability even with suboptimal usage.
Contamination risk
Device can measure without risk of
contaminating the blood sample volume.
Device can measure the blood volume indirectly without risk of becoming contaminated.
- Device can
automatically perform the ISDT method.
Environmental impact
- Does not
increase the usage of one-use material.
Long system survivability to keep replacement rate low.
-
Compatibility with the standard kit explained under the
“choosing a sample”
title.
Can operate with the standard kit.
Does not change how any part of the standard kit is used.
Does not change the structure of any part of the standard kit.
-
Economical feasibility
Multiple use unit around 500 Euro.
Multiple use unit cost below 500 euro.
- -
Measurement feedback
Signal when the volume added to the system reaches recommended amount of 8-10 ml.
Real time tracking of the blood sample volume added to the system.
Ability to store information about added blood sample volume.
Can document and send over where, when and how much volume has been added when measurement has been performed
Biosafety
No risk of liquid spill on healthcare staff.
No chance of personal harm by usage of device.
- -
Reproducibility Measurements on the same bottle produces similar results.
Measurements on different BC bottles gives similar results.
Measurements on different bottles containing different amount of initial liquid gives similar results.
-
Measurement range Capable of measuring the volume over the entire bottle volume.
No significant change in measurement ability over full bottle range.
- -
4.2 Define
4.2.1 Secondary research - literature review
The following articles were found in the literature review and illustrated two distinctly different methods to measure a volume coming in to a tank (23, 24). The first way, is to measure the flow coming into the tank using some form of flow meter and then multiplying by time to get volume.
The second method was to use some technique to measure the liquid level inside of the vessel and then getting the volume by mathematical calculation. The review was continued by trying to find articles and sources that could help in finding all different ways these methods could be used. The result of this search is summarized in Table 2 below (25-32).
Table 2: Different solutions found by the literature review.
Flow Meter Methods Liquid Level Methods
Electromagnetic Capacitance
Ultrasonic Displacement
Differential pressure / orifice meter Resistance transducers
Coriolis effect Level sight gauge
Laser doppler Pressure sensor
Pitot Tubes Inductive transducers
Calorimetric Flowmeter Digital / Infrared camera
Positive Displacement Flow meter TOF (time of flight) / echo methods (Radar, ultrasound, laser, microwaves, nuclear)
Vortex shedding Optical methods
Thermal Flowmeter Float methods
Open Channel Flowmeters Load cell / weight scale
Resonance sensor
After all these potential methods were found they were compared to the must features presented in the MoSCoW prioritization table found earlier removing methods unable to function for the intended application.
4.2.2 Comparison to must have features
As requirements such as, accuracy, reproducibility, measurement range and measurement feedback are hard to judge for a specific application without prototyping they will not be used in the comparison.
The first must have feature examined was the one connected to the contamination risk requirement. According to the feature, measurement methods are required not to increase the risk of contamination inside of the blood sample. This is achieved when the measurement method measures without direct contact. Many level measurements method such as floats, gauges, pressure sensors and displacement sensors require parts of the system to be in contact with the fluid and, as such, are not fit for this application (23-24). The same goes for differential pressure, positive displacement, calorimetric, thermal, open channel, pitot tubes and vortex shedding flow meter sensors that require system parts to be in contact with the fluid (22, 25-26).
The next quality examined was the one connected to the workflow compatibility requirement.
The hospital environment is not a specifically limiting environment as temperature, humidity and other conditions tend to be relatively constant. However, in connection to robustness there exists limiting environmental factors. For example, measurement system will generally not be kept still when operated and there exist potential for electrical disturbances in the form of mobile phones, human body parts and electrical equipment. Measurement systems not kept still creates sever problems for resonance and optical fiber liquid level methods as they are sensitive to mechanical movements (23, 28). Electrical disturbances in the environment will definitely cause problems for both capacitance and inductive level sensors, but in the case of capacitance sensors there are methods to lower the impact of disturbing electrical signals (29).
Another requirement to examine is independence of fluid type. The measurement method must be able to operate with the intended fluid and the expected flow rate displayed by the blood volume collection system used today. Blood as a fluid does not on a theoretical level cause a lot of problems for most remaining measurement methods. Especially for digitally controlled ones that can be recalibrated in order to keep operating with different kinds of fluids. All forms of electrical measurements systems (inductive, capacitive and electromagnetic) should theoretically be able to operate with blood because, blood like water, is a conducting fluid. However, the normal flow rate created in the BC bottle could create problems for some measurement methods that require a stable liquid surface level for example the different time of flight (TOF) methods. All these measurement methods operate in a TOF manner meaning a wave is sent out to travel along the tank until the wave encounters a reflecting surface making the wave reflect back to its original location. Dependent on how fast the wave gets back the liquid level can be calculated as this variable will correlate to the liquid level inside the tank (30). If the surface is agitated this process can encounter problems as the sent out wave starts to scatter around, leading to erroneous measurements (30). Because the flow of blood coming into the BC bottle can be quite high and user movement destabilizes the liquid surface TOF methods can be expected to not perform particularly well.
Next requirement to examine is the compatibility with the standard sampling kit, where the measurement method must be able to work without changing the already standardized system defined earlier in the report under the “choosing a sample” title. Because the tubing is quite short and needs to be easy to use, a Coriolis flow meter sensor would not be capable of working with the system as they require a complex tube set up (25,26). Also, digital camera systems can have
problems operating for this application as the liquid level will at times be hidden by the BC bottle label or the cap adapter attachment.
The last important requirement to examine is the theoretical cost of the different measurement method, whereas the cheapest option is the best. Most measurement methods left to be considered are cost effective. Both load cells and capacitance sensors are low cost devices meanwhile, electromagnetic and ultrasonic flowmeter sensors have been used in healthcare for quite some time, indicating they in general are not too costly for the field (31). However, in connection to the infrared camera measurement method, cost will probably become a concern as good quality IR cameras are comparatively expensive. This would in theory lead to fabrication and testing costs becoming higher than the other systems (32). As such, was the IR camera measurement method considered unfit for the intended application.
The measurement methods left to be considered are the non-contact capacitance level sensor, an electromagnetic or ultrasonic meter sensor and a weighting system utilizing a load cell.
As, all of these methods also passes the biosafety requirement they have theoretically passed all different must features in Table 1. Due to this, another method will have to be used to further compare the methods.
4.2.3 Assignment criteria
In order to theoretically compare the measurement methods that could perform all the must features examined, the assignment criteria methods were used. Further below in Table 3 the scores given to different ideas following the method explained in the method chapter under the
“assignment criteria” title is shown.
4.2.3.1 Explanation for scoring
The technical possibility score represented how easy it would be to develop and prototype a device following the specific measurement method. Concerning both the non-contact and the weighting method there exists simple Do It Yourself (DIY) Arduino projects showing how a rudimentary prototype could be made. The same was not the case for both the ultrasonic and the electromagnetic flowmeters where most examined DIY projects did not indirectly measure the volume but instead measured it in contact with the fluid. As such, higher points were given to the non-contact and load cell methods as they seemed easier to work with on a technical level. The same argument was used in scoring the fabrication score where both the load cell method and the non-contact liquid level method did not require many parts to prototype. Moreover, was one of the most important parts for prototyping the two measurement methods already in place at the lab.
As it is of utmost importance to avoid the blood getting contaminated, only methods measuring the volume indirectly was moved onto this part of the double diamond approach.
Because of this, all the different measurement methods were given the highest clinical possibility score.
In terms of usability, would the flow meter systems be ideal if they operate in a clamp on fashion.
This because the usage is intuitive, and the normal workflow would not need to change to accommodate measurements. The only course of action required would be to clamp the meter on to the tube and have them give feedback to the users when the measurement is complete. This is not true for both the capacitance level sensor and the load cell method where both measurement
methods, will require some finesse in handling to avoid errors while measuring. This is especially true in the case of the load cell where any form of pressure or weight added to the measurement apparatus would make measurements unreliable. Due to this, the flow meter sensors get the highest possible usability score while the outer two are given less points.
As all the measurement methods are intended to create a multiple-use and long-lasting product, they should not add on to the large amount of plastic waste normally associated with the hospital sector. As such, the different systems scored the same amount of points in terms of environmental impact.
Initially, when researching, it looked like all the methods would yield cheap solutions. This, however, was found out not to be true as flow meter sensors usable for very small tubes were found to be expensive. Because of this, the flow meter and the electromagnetic sensors were given low budget scores.
Table 3: Assignment list for deciding what method would yield the best solution.
Criteria
Non-Contact Capacitance Liquid Level Sensor
Weighting Method / Load Cell.
Ultrasonic Flow Meter
Electromagnetic Flow Meter
Technical possibility 5 5 3 3
Fabrication possibility 4 3 2 2
Clinical possibility 5 5 5 5
Usability 4 2 5 5
Environmental impact
5 5 5 5
Budget possibility
5 5 1 1
Total
28 25 21 21
Shown in table 3, the in theory best solution found was the non-contact liquid level sensor. As such, this idea was moved on to the next step of the double diamond approach “develop”.
4.3 Develop
4.3.1 Theory in connection to prototypes
In order to understand the prototypes created during the development phase, there is a need for some basic theory in connection to non-contact capacitance liquid level sensors.
4.3.2 Capacitive liquid level sensor
Capacitive liquid level sensors take advantage of the fact that many liquids such as blood and water are conductive and can operate as the dielectric for a capacitor setup. Normally this has been done with the sensor placed in the liquid but can be done in a non-contact manner by placing a sensor and potential shielding in a parallel finger set up following Figure 5 below (33).
Figure 5: Parallel finger setup with shielding. Blue arrows show the so called fringing electrical field flowing from Cinx to the liquid inside the bottle and then to ground.
As can be seen in Figure 5 when a current is applied to the electrodes, an electrical field (in this case called a fringing field) will be created and travel from Cinx to the inside of the tank containing a conducting liquid acting as a dielectric and then move over to ground. Capacitance (C) is defined as the ability to store an electrical charge (Q) in a body when a voltage V is applied (34).
C =
𝑄𝑉
(1)
For a specific conducting electrode setup, the conduction will be given dependent on the geometrical placement of electrodes and the properties of the dielectric. As it is out of the scope of this work, no general explanation will be given concerning how to derive the conduction
equation for any specific setup. Instead, all that needs to be known is that for any material with a dielectric k being sensed by the electrodes placed in a parallel finger setup will give rise to the capacitance Cd
following
(35).C
d= L*k*f(d,w,s)
(2)There L is the length of the electrodes, d is the distance between them, w is their thickness and s is the thickness of the dielectric material in front of the electrodes. If the setup follows Figure 5 the capacitance for the hole setup sensed by the electrodes can be given by the following equation (35).
C
total= C
air1+ C
tank+ C
liq+ C
air2(3)
Where Ctotal is the total capacitance, Cair is the capacitance of the air outside the vessel being measured, Ctank is the capacitance of the vessel itself and Cliq is the capacitance of the liquid inside the vessel.
Using (2) to expand (3) and approximating the thickness seen by the electrodes as infinity concerning Cair, Cliq and Cair2 the following equation is received (35).
C
total= L*k
air*f(d,w,∞)+L*k
tank*f(d,w,s
tank)+H*k
liq*f(d,w,∞)+(L-H)*k
air*f(d,w,∞) (4)
All terms in (3) will be constant outside of the variable H (height of the liquid column) as such Ctotal can be rewritten as.
C
total= C
e+H*(k
liq* f(d,w, ∞) – K
air* f(d,w, ∞)) = CE + NH (5)
Where CE and N are constant values that can be found by empirical calibration. CE is the capacitance when the vessel is empty and can be found by measuring the vessel when empty. Next step in empirical calibration is to measure the vessel when full. This will give the value CF and if all things are in order, the liquid level H should be equal to L. By using these constants, the constant value of N can be found in the following way (35).
C
F= C
E+ N
L→ N =
(𝐶𝑡−𝐶𝐸)L
(6)
With all these constants gathered, if the capacitance seen by the electrodes measured at a time t is equal to Ct then the height of the liquid column can be found by the following equation (35).
H(C
t) =
(Ct−CE)∗L(CF−CE )
(7)
The setup described above was only used for the first sensor prototype. The prototypes after the first one, used another two electrodes added to the setup to perform reference measurements and
four more electrodes to perform a method known as the out of phase liquid level method (OOP).
This setup is shown in the Figure 6 below (29).
Figure 6: Electrode layout when using the out of phase liquid level method. Following Figure 6, the longer electrodes measuring the liquid are called level sensors and the
smaller ones are called reference sensors. In connection to this setup the height of the liquid column will be given by the following equation (29).
H(t) = H
RL* (C
L- C
L0) / (C
RL- C
RL) (8)
H(t): Height of the liquid column at time t.CL: Capacitance measured by the level electrodes at time t.
CL0: Capacitance measured by the level electrodes when the vessel is empty.
CRL: Capacitance measured by the reference electrodes at time t.
CRL0: Capacitance measured by the reference electrodes when the vessel is empty.
HRL: Length of the reference electrodes.
After the height of the liquid column H has been found using either 7 or 8 the volume added to the vessel can be calculated by multiplying the found height with the area of the specific vessel.
In regards to this thesis work, as the vessel is a cylinder the volume added to the BC bottle can be found 2 in the following way.
V(H) = Pi * r2 *H (9)
Where the radius r of the BC bottle has been approximated to be 16,3mm.
4.3.3 Shielding and the out of phase liquid level method
It is off outmost importance, when operating a capacitance liquid level sensor to make sure that the sensor only senses the capacitance of the liquid dielectric inside the vessel. To make sure this happens, the capacitance liquid level sensor uses two different methods. The first method is the shielded backside added to the electrode layout. This shield is driven by the same excitation signal as the sensing electrodes placed on the front side and will because of this, have the same potential. Because the potential is the same no electric field will be created on the backside of the sensing electrodes. As there is no electric field on the backside of the sensing electrodes they can only sense in the direction off the vessel (29). Figure 7 below visualizes the effect of having a shielded backside for the capacitance liquid level sensor.
Figure 7: Effect of using active shielding when using a capacitance level sensor.
Measuring capacitance with active shielding works well in a calm environment without disturbing sources. But in an environment with high potential for external parasitic capacitances, the system will struggle to operate well. As even, just a human hand or any conductive object could create this form of parasitic capacitance the sensor would be unable to operate in the hospital environment. To lower the effect of this problem the OOP liquid level method can be applied (29).
Normally, the sensing electrodes will be driven by an excitation signal causing a voltage potential to be formed between the sensing and the ground electrode. For the electrical system to work, this potential should remain constant. However, if a disturbing source comes close to the sensors or the liquid inside the vessel, the potential of the theoretical node will change and as such the result found by the sensor. The out of phase liquid level method counteracts this by having the sensing electrodes measure the capacitance from one driven electrode to another driven electrode.
As both the electrodes are driven by the same signal and are 180o OOP with one another, the potential of the theoretical node will remain the same even when a disturbing source starts interacting with the system. This only works because the FDC1004 can perform differential measurements, something that many capacitances to digital converters (CDC) cannot do.
Moreover, when using the OOP method, it is of great importance the entire electrode and sensor layout is symmetrical. Any deviations from a symmetrical setup will cause changes to the potential of the node when measuring and as such will cause disturbances (29).
4.3.4 Calibration process
Before a prototype could be tested it needed to be calibrated in two ways to yield accurate results.
The first calibration process is setting initial values for the system. The second calibration process is to create a linear correction algorithm.
4.3.5 Setting parameter values for the system
The initial values required to operate the system depends on what kind of set up the system uses.
The process itself is straight forward fill up, and then empty, a BC bottle used for testing and place it in the view of the electrodes. Read the capacitance/capacitances values found these will correspond to either CE in equation 7 used by the first prototype or CL (0) and CRL (0) in equation 8 for the later prototypes. While still in the view of the sensing electrodes fill the bottle up to the max this will give the value CF in equation 7. After this has been performed all parameters needed, to use the different prototypes has been found. To use the level sensor for a different liquid type, these parameters will have to be set again following the same procedure (36).
4.3.6 Linear correction algorithm
After required parameter values have been found a wet test is performed to create a linear correction algorithm for the system. This test is performed in the following way. First, empty out the vessel to be filled and place the vessel so the sensor and a more accurate system can measure the volume added to the vessel at the same time. The more accurate measurement system, used to correct the prototypes, was a syringe filled with a specific volume of liquid. After the setup was completed, the vessel was filled with the syringe in 10 ml increments. The volume value found by the capacitance liquid level system for every increment was written down. Figure 8 shows, the result of a wet test performed on one of the prototypes created (36).
Figure 8: real vs measured volume from a wet test performed.
As can be seen in Figure 8, the real measurement line will be fully straight following Y = X. The second line is created by the volume datapoints gathered at every 10 ml increment from the capacitance liquid level sensor being tested. To create a linear correction after a wet test has been performed, first find the linear approximation estimating the gathered datapoints. This can easily be done using a linear plotting tool such as the one available in Microsoft Excel. After the linear approximation is found the following equations will give out constants required to create a linear correction algorithm.
If the linear approximation follows
Y = K*X + N (10)
Then linear correction will be given by
Measured
0 10 20 30 40 50 60
0 10 20 30 40 50 60
Real volume
Measured volume
Wet test
Real volume
Y(true) = B*X + M
(11)
Where
B = (1/K) and M = -(N/K) (12)
If the reader is interested in the proof for these equations a derivation can be found in the following citation (36). To showcase the effectivity of the correction algorithm before and after linear correction results are shown in Figure 9 below.
Figure 9: Real volume vs measured volume for the different prototypes with and without the linear correction algorithm.
4.3.7 Capacitive Liquid level sensor version 1
Following the parallel finger setup talked about in the capacitive sensing basics chapter a prototype was created. To convert the electrical signal created between the electrodes in the setup a
capacitance to digital converter (CDC) by the name Protocentral FDC1004 breakout board was used. In order to read signals and operate the breakout board the Arduino Uno microcontroller was used. To run and compile code for the microcontroller the Arduino Uno can be connected to any form of computer with a USB port. Connections between the Arduino Uno, the FDC1004 and the copper electrodes were as follows; from the Arduino to the CDC A4 to connection pin 1, A5 to connection pin 2, 5 Vdc to connection pin 5 and GND (ground) to connection pin 6. From the CDC to the electrodes pin Cin1 to the sensing electrode and the other electrode connects to the ground port of the CDC. A summary of the different connections is shown in Figure 10, meanwhile, Figure 11 shows the full physical system.
Figure 10: Schematic image of the connections between the UNO the CDC and the electrodes.
Figure 11: Full system with Arduino Uno copper electrodes FDC 1004 breakout board and BC bottle.
The code running the microcontroller was a modified open source code generated from the Protocentral project and is compiled and run by the Arduino IDE. The open source license of usage can be seen in the following citation (37). The full code used by the last prototype can be seen in Appendix 1. After calibration had been performed, initial testing showed an acceptable ability of the system to find a volume of water added to the system. However, the system also displayed poor robustness qualities and measurements could easily be made unstable by having a conductive object such as a mobile phone or a hand come close to the sensor. Moreover, even things such as the position of people in the same room as the sensor could significantly change the found volume value.
4.3.8 Capacitance level sensor version 2
The second prototype created used the OOP liquid level method describe earlier. As can be seen in Figure 12, the same FDC1004 breakout board was connected in the same way to the earlier used Arduino Uno. To perform the OOP liquid level method a different sensor layout was used. This sensor layout was made up of three layers. Where layer one closest to the bottle had four copper electrodes, layer two is entirely made of electrical isolation tape and layer three is made up of four more copper electrodes. The setup can be thought of as two symmetrical setups one with longer copper electrodes in the first layer and the second one with shorter. In both setups one sensing electrode is measuring capacitance towards a shield connected electrode in layer one with two additional shield electrodes behind them in layer three. Electrical isolation tape was placed around pin connections and wires to lower the effect of external parasitic capacitances. Below in Figures 12 and 13, the symmetrical setup with pin connections to the CDC are shown. In Figure 14 below, the physical system is shown.
Figure 12: Setup and pin connection for the sensor setup with longer copper strips.
Figure 13: Setup and pin connections for the sensor setup with shorter copper strips.
Figure 14: Capacitance liquid level sensor version 2.
The shield electrodes are the same length as the longer sensing electrodes and where 44 mm long. The two shorter copper electrodes where 15 mm long. The gap between different electrodes
in the same symmetrical setup was about 1 mm meanwhile, the distance between the long electrode and the short electrode setup was 5 mm long. The copper adhesive thickness and width are the same as in the former prototype. After calibration initial testing showed, that the system performed a lot better than the first prototype as it could handle external parasitic capacitances without losing measurement ability. However, other problems with the system were found. The open container did not create a tight fit between copper electrodes and the BC bottle. This lowered the prototypes reliability and accuracy, as volume values found changed depending on the proximity of the bottle towards the sensor.
4.3.9 Capacitive liquid level sensor version 3
The next step was to create a sensor capable of working as a real product, operating with the standard blood sample volume kit talked about under the “choosing a sample” title on page 10 in the result chapter. Both the sensor layout and setup were the same as in the earlier prototype. All connections between the Arduino Uno, the CDC and the copper electrodes were the same as the second prototype. The container holding on to the copper electrodes was replaced with a 3D printed horseshoe shaped structure where electrodes could be held towards the BC Bottle. This structure was connected to a 3D printed compartment that contained space for both the Arduino Uno and the CDC. The entire 3D printed structure was not created by the author, but instead by the design expert member of the working team at clinical microbiology Karolinska University Hospital. In Figure 15 below, the full physical system is showed.
Figure 15: Third liquid level sensor version 3 full physical system.
To see if the sensor could operate as normal with the standard kit used, when collecting blood samples volume, the sensor was tested using the cap adapter attachment normally connected to the tubing coming from the butterfly needle. Testing showed that the plastic attachment caused problems similar to the ones the second prototype had. Where, Accuracy and reliability of volume measurements were negatively impacted by the bad fitting of electrodes towards the BC bottle.
This because, the cap adapter attachment changed the distance between the electrodes and the
bottle during different measurements. This meant that the prototype could only work optimally without the cap adapter attachment.
4.3.10 Capacitance liquid level sensor version 4
The fourth prototype was very similar to the third prototype, only difference being that the system used a different 3D printed electrode holder. This was done, to see how much better the system would work if the copper electrodes were better fitted to the bottle. The 3D printed attachments were created by the same member of the team and the full physical system is shown in Figure 16 below.
Figure 16: Fourth prototype with better fitting sensor model.
Testing showed that this system had the best-found accuracy so far and that the reproducibility problems present in the second and third model were gone. However, was a different problem found in this sensor where, measurement results drifted to a significant degree over longer periods of time. As the blood volume sampling process is quick, this should not impede greatly on measurements. However, some software changes were made to the code operating the system to lower the effect of the problem.
A second problem was found when testing the prototype, connecting to the measurement ability of the sensor. When the liquid level inside the BC bottle reached the maximum height of the level electrode the sensitivity of the sensor increased. This caused accuracy to drop a significant degree around that liquid level high. This might, cause issues for the device dependent on how high the liquid level is expected to go while performing normal blood sample volume collection.
4.3.11 Programming / code
Most of the code used came from an opensource project that can be seen in the following citation (21). However, to generate measurement values, using the opensource project code a lot of changes were made. In appendix 1, the last iteration of the code is shown. To get deeper knowledge
