Weekly Reports
Weekly Report #2
Sep.17- Sep.22
During the week of September 17th, our group worked polishing our project scope and determining which modality we wanted to use for our device.
Our project scope and need statement was written with the intent to develop a test for determining mice pregnancies within the first 96 hours of conception to help aid biology researchers in their work. The project specifications were oriented to reducing the time and cost of using the device.
After that, our group used the information from class on September 20th to evaluate several types of device modalities we could use. Initially, we were planning on developing a device that can detect a biomarker inside mice urine or stool that is indicative of pregnancy, but this may take longer that the duration of the project will allow. Other possibilities concerning the examination of mouse circadian activity or uterine pressure were considered and compared. Before we can decide which type of device to build, we would like to schedule a meeting with both our mentor and Professor Yin/Klaesner. We have gone ahead and scheduled this meeting.
Weekly Report #3
Sep.25- Sep.30
(1)Met with Dr. Yin to discuss project (9/25)
a. We were very much focused on biomarker solution.
b. Dr. Yin encouraged us to look at many avenues to start with when constructing a chart No need to pinpoint a specific strategy right now and look for many different avenues
(2) Met with Ron (Lab Technician) for the England Lab (9/28)
a. Discussed the England Lab’s work with examining SrY gene as a possibility for pregnancy detection
i. Did not seem to work well i.e. a mouse that is pregnant does not have the SrY gene in its fecal matter 1. Why not? Potentially because DNA from fetus cannot get into the mother’s intestinal lining b. Discussed importance of early detection of pregnancy i. Need further clarification – it actually seems that their accuracy percentage is already very high (95% for overnight breeding) ii. Will discuss with Ron very soon – critical for our project c. Discussed more biomarkers we would need to test i. Progesterone, estrogen, prolactin
Weekly Report #4
Oct.1 - Oct.6
During the week of October 1st, our group worked on creating our preliminary report and the accompanying presentation.
We revisited our design specs to make them more quantitative in nature. Accuracy was recalculated keeping the earliness of detection in mind. Price effectiveness was added as a design spec to make the device a less expensive solution to the current standard of purchasing more mice. The need statement was revisited when creating an introduction for the preliminary report. Further emphasis was placed on the market environment and help researchers would receive from this device.
We also further explored the pre-existing solutions this week for our report. We discussed ultrasonography, core body temperature, progesterone, and copulation plugs as solutions for early pregnancy detection in more detail and by doing so obtained further understanding of the need for an early and accurate detection of pregnancy.
We seek to finish finalizing our report and paper, and meet with our clients again to further discuss solutions of early pregnancy detection involved similar hormones such as Luteinizing Hormone and Estrogen.
Weekly Report #5
Oct.8 - Oct.13
This week in class, Daniel was selected to give the preliminary presentation. Therefore, much of this weekend was dedicated to helping Daniel with developing the material in the presentation, and providing feedback on his delivery and presentation skills. Daniel gave his presentation on Monday.
After the presentation, the team worked on their individual tasks: Abhi researched different biomarkers and the current detection capabilities of these biomarkers, Rohan examined past solutions and tried to relate them to innovative ways to detect pregnancies, and Daniel researched more physiological changes in mice to develop a novel solution not dependent on biomarkers. Also, Daniel worked on developing a new CAD design per the client’s request to aid with the urine collection, as the first design failed to be printed due to the supports being embedded in the design.
In the following week, the team will wrap up the literature search and begin brainstorming devices to begin designing. Therefore, we hope to have a pugh chart put together soon. This will require the synthesis of all the literature searches, a discussion of each device’s ability to satisfy the design specifications with our client, and possibly, some advice from different experts about the feasibility of each device.
Weekly Report #6
Oct.16- Oct.20
For the first half of this week, the team was on fall break, with most of the members out of town and visiting family. Therefore, not much work was accomplished during this period.
For the latter half of the week, the senior design team focused on establishing the website. Rohan Khopkar has past website design experience, so he began using the service “wix” to create a webpage. The team also organized all the past weekly reports and uploaded them on the webpage.
In addition to the webpage, the senior design team went over the feedback from the preliminary report. The comments regarding the difficulty to examine several design specifications and the preconceptions we already had about how our device would look like were discussed. As the feedback was read, the team came up with a list of questions to discuss with the mentor.
In the next week, the team will meet with the mentor and discuss the questions formulated this week. Also, the team hopes to wrap up the literature search and begin forming a pugh chart distinguishing the benefits and drawbacks of different solutions that can satisfy the need and remain within the project scope.
Weekly Report #7
Oct.23- Oct.27
A quick summary of the progress done this week was (1) discuss the progress report due in early
December, along with who should present, (2) discuss the meeting with Dr. England and
schedule meeting with her in the future, and (3) discuss ideas we came up with for our method to
detect pregnancy. Possible ideas include:
EMG electrodes to examine pregnancy-related activity changes
o Pressure of uterus rising during pregnancy
o Fluid in uterus detection while develops during pregnancy
o Biomarker examination which fluctuate because of pregnancy
o Temperature assessment using temperature probes with high resolution
o Palpation device to detect the pups physically
Rohan, Daniel, and Abhishek will continue researching these next week. On Friday, November
3
rd, the team will meet with Dr. England to discuss these six ideas, along with the feasibilities of
these ideas using the research developed over the past few weeks.
Weekly Report #8
Oct.30- Nov.3
This week our team focused on researching various solutions for early pregnancy determination in mice.
Considering new feedback from our mentor that SRY expression was unable to be shown in adult Male
mice via PCR, the group decided to look at other potential biomarkers that change during early
pregnancy in mice, such as cortisol and estrogen. The group also looked at Sox9, a critically downstream
transcription factor to SRY.
In addition to biomarkers the group also researched other methods of ascertaining pregnancy status.
Electrodes being placed on the surface of the mouse skin could be used with an EEG to determine sleep-
wake cycles due to the mice activity. This could be used to observe a possible chance in sleep-wake
cycles during pregnancy. Using uterine fluid detection methods via ultrasound were also studied.
The group will meet with the client Sarah England on Friday, November 3, 2017 to discuss future
direction and the best methods of early pregnancy detection.
Weekly Report #9
Nov.5 - Nov.10
This week, our team reflected on the meeting with Professor England last week. We decided that
biomarkers were a very ‘hit or miss’ detection method for mice pregnancies, so we decided to look at
other options.
The main ideas that were discussed were using a motion sensor device to examine changes in activity
post conception, and the increase in fluid accumulation in the uterus post conception. The motion
sensor would need to have wifi capabilities to avoid mice gnawing at wires, and be located on the tail to
prevent mice chewing at the electrode. The uterine fluid would need to be detected using an ultrasound
probe with a single piezoelectric element, and a calibration curve for different compositions of mouse
uteri. A meeting was scheduled with Professor Widder to discuss these ideas further.
In addition to discussing these ideas, we also designed and 3D printed a collection funnel for the client.
This would aid in their ability to obtain urine and fecal matter, and help speed up the assessment of
biomarkers. This was printed at STS.
Weekly Report #10
Nov.13- Nov. 17
Met with Patricia Widder and Noah Ledbetter regarding some ideas & advice regarding
the project on Tuesday (11/14)
o Patricia suggested discussing with the Raman Lab
o Try to make a solution that is minimally invasive
o Try to mark the mice and wirelessly record a signal that represents mouse
movement over time
o Raman Lab uses this technique for locusts
o We contacted Debajit Saha to gain his advice regarding this potential marking and
video recording technique
- Met Dr. England to deliver prototype for urine collection device. This is not the final
device that we are constructing, just a device that we 3-D printed to help our mentors
with biomarker analysis.
- Ron sent us feedback; Rohan and Abhi worked together on Thursday (11/16) to find
potential improvements in design. Mouse urine may be viscous, which prevents it from
flowing through the collection device, and the urine may dry up.
- Rohan and Abhi will meet with Ron on Friday (11/17) to discuss improvements together
and start constructing a smaller model for the next urine collecting device
- Progress Report: Rohan, Daniel, and Abhi discussed briefly on Thursday (11/16) to
discuss our aims and general ideas regarding the progress report. Our goal is to finish the
report by the time we come back from Thanksgiving break (11/27) and schedule an
appointment with Kristina or Dr. Yin to discuss potential improvements. We may also
contact the Writing Center to get advice regarding improvements in technical writing.
Weekly Report #12
Nov.27 - Dec.1
This week the group worked on compiling the research and design ideas to formally outline design
alternatives for the progress report. Abhishek worked on the skin conductivity model and the volume
pressure recording solution. Rohan emailed the clients for more information concerning the SrY and
biomarker response and formulated that solution, as well as the motion tracking solutions for PIR
sensing and RFID motion tracking. Daniel worked on the ultrasound using uterine fluid solution and the
video tracking solution.
To get more information about video tracking fluorescent dyes on an organism, the group met with
Dr.Raman for a brief discussion regarding the video tracking their lab conducts on locust models.
The group came together on several occasions for the latter half of the week to work on the progress
report and finalize preparations for the progress presentation coming up.
Weekly Report #13
Dec. 4 - Dec. 9
This week, we followed up with the design alternative we selected for our progress report with our client. We rewrote sections of the progress report applicable to our particular design selection and tailored it to an audience of biologists who understand a great deal about the biology of mice pregnancies but not as much about the signal processing of our design.
After that, we began to research how we can implement this design as soon as we got back from winter break. We already knew what we needed to perform our experiments due to our progress report, but had yet to physically obtain these items. Therefore, we needed to put in orders through online vendors and request mice and time to obtain these items from our client. We spent the majority of this week getting into contact with our clients and ensuring not only will we have the mice we need, but also the time with the mice required to inject RFID chips and take measurements of circadian rhythm.
Semester 2
Weekly Report #1
Earlier last week, we had a meeting with our client discussing the logistical considerations for
running our product on the client’s mice. We prepared a room where we can test and
troubleshoot our product, and agreed on the number of mice we will use for the initial testing.
After developing the experimental outline, we went over the instruments needed to prepare our
device. We placed several inquires about decoder boards, antenna, and transponders to several
different companies. Currently, we are waiting to hear back from many companies.
Unfortunately, our initial design will not be viable with the experimental design as the location
where we are testing our product will require more antennas than we expected. Therefore, we
will need many more antennas connected to a decoder board rather than having a single reader
with a decoder-antenna pair. We are looking at the most economic way to manage this problem –
which will consist of purchasing the fewest antennas and placing them in high traffic areas that
can detect the activity of a mouse quite accurately.
Jan.22 - Jan.26
Weekly Report #2
Jan.29 - Feb. 2
This week, we focused on purchasing our equipment for developing our device. Earlier last
week, we reached out to Wayne Sacco from Microchip ID, who stopped returning our calls.
Furthermore, many other retailers have been dragging their feet in returning our calls for quotes
on various devices.
We did learn that the frequency of the transponder and reader need to match, which restricts the
types of transducers and readers we can use. Furthermore, our client asked us to purchase a very
specific type of transducer that can be injected subcutaneously into mice using a cannula.
Purchase the transducer that is used for this purpose and the corresponding reader will cost more
than our budget has allocated for us, so we are looking at other, cheaper, more viable options.
Lastly, we spent some time this week looking for other folk at WashU who use RFID in their
research. They may have excess RFID readers and transducers, and can answer some of the
questions we have about data processing. We talked to Minquan Yang, Aaron Norris, and
Shamus Keeler about RFID tags, and scheduled a meeting with Dr. Norris next week and Dr.
Yang today (Friday February 2 nd , 2018). We hope to jump to a quick start with our equipment
arrives with their help.
Weekly Report #3
Feb.5 - Feb. 10
This week, we brought up our concerns with using RFID to a third party, Dr. Shanatu
Charabratty, in the ESE department. This professor has extensive experience in using RFID for
cutting edge research, and was able to tell us exactly what type of RFID device we wanted for
our project. The cheapest UHF reader and tag combination will be at least $260, but we only
have an approved budget of $185.
After learning about the cost issue with RFID, we decided to talk to our client about the
purchasing price. The client agreed to help us with the cost. However, we also decided to take
initiative and attempt to flesh out a video tracking solution in case we could not obtain additional
funding for the RFID. Our client was very impressed with our initial data for video tracking
solutions, and asked us to look into pursuing which one shows more promise.
Currently, we are scheduling a meeting with Dr. Yin to discuss the feasibility of working on two
projects concurrently, then selecting the modality that shows the most promise in two weeks
when we need to start testing. The potential for RFID seems much higher, but the feasibility of
completing the video tracking solution seems much more realistic, and we are torn about which
direction to proceed. We hope that after our meeting with Yin, we can find which direction to
move forward.
Weekly Report #4
Feb. 12 - Feb. 16
This week, we set up a meeting with Dr. Yin, our course professor. We discussed our concerns
with using RFID. As a group, we decided, that given the time constraints and our lack of
experience with the technology, we were going to change project ideas and track mouse
movement using paint and a camera. This was an option that we detailed in our progress report
but pushed aside for RFID.
We met as a group and discussed the preliminary code that we have written for finger tracking.
We have decided to do some independent research in the image processing tool in MatLab. We
also met with members with Dr. Raman’s lab to see how we can use paint as a contrast agent to
image mice during the nighttime. We were initially confused how we could use a camera that
can image objects in the dark. We recognized this was not feasible, so we could use an infrared
camera.
After talking to Dr. Raman’s lab, we learned that mice cannot see red light; we will be
performing further research regarding mice optics. Therefore, we can use a red lightbulb, or a
filter that will selectively filter out only red light. If we put red paint on the mouse, then the red
paint will reflect the red light and will be detectable using a regular every-day camera.
Weekly Report #5
Feb. 19 - Feb. 23
This week, we realized that the red light would not be a viable alternative to tracking mice at
night, as it would still interrupt their sleep cycles despite being invisible to mice. Therefore, we
would have to use infrared cameras and a corresponding contrast agent to image the mice.
We worked on the code to transfer the processing language from Matlab to Python because the
code kept crashing in Matlab. This was due to an incorrect array size in Matlab when an object
would fall off the video screen. However, Python was able to accommodate this error much
better and we were able to proceed with our code. We have developed the code to successfully
track and create a velocity trajectory for two objects of different colors simulatenously. We still
need to determine a method of changing the tracking algorthym from day to night, but first need
to find contrast agent before we begin tackling this problem.
After the code made significant headway, we turned to find a contrast agent to image the mice
using an infrared camera. This contrast agent would preferably reflect all light, as it is easier to
detect a bright spot rather than a black spot. We examined three different types of contrast
agents: paint; aluminum foil; and metal ear tags. After discussing the feasibility of these topics
with our clients, it appears that the metal ear tags are most convenient for our clients. They have
set up an experiment to view the mice tonight.
Weekly Report #6
Feb. 30 - Mar. 4
This week, our team focused on writing the Verification and Validation report and proceeding
with the verification of our project. We also had three meetings with our client to perform
verification steps, and a meeting with professor Yin to discuss our V&V report.
The verification steps we took this week included object tracking with an infrared lens and
multiple object tracking with multiple colors. Expanding our software to accommodate infrared
reflective objects is necessary to visualize mice at night. Tracking multiple objects with multiple
colors is necessary to track the mouse when the researcher switches from a regular camcorder to
an infrared camcorder without having to restart the program. Lastly, we tried to visualize
aluminum spray paint and “mirror” paint with an infrared camera because we currently do not
have specific infrared-reflective material ready but wanted to show object tracking with an
infrared camera for our V&V report. This spray paint did not work very well.
We met with our clients to use their camera for verification of infrared visualization and setting
up the cameras for the validation. We also talked to Yin to understand the difference between
verification and validation. Afterwards, we put together our verification and validation paper.
Weekly Report #7
Mar. 5 - Mar. 9
This week, we focused on ordering the infrared paint to put on mice. We compiled several
different infrared reflective solutions, and sent it to our client to determine which option is most
reasonable for mice. We have also decided on which option is most likely to provide good data,
and sent that option to our client as our best selection. We are still waiting to hear back from her
about which paint would be best for her purposes.
Our client’s lab specialist, Ron, was out of town this week. Therefore, we were unable to set up
the cameras and gather data from the mice. However, we were able to schedule a time to meet
with him next week over spring break to begin data collection. We will follow up with him next
week.
The next steps for our project is to order the paint and apply it to mice. Rohan and Abhishek will
work together on the code over spring break to target the specific paint colors we plan on using,
and create a simple GUI that will help the user select the parameters needed for our program to
work. Daniel will work on the ground to ensure the cameras are properly lined up, the glare is
minimized, and the paint is properly applied before baseline data collection begins.
Weekly Report #9
Mar. 19 - Mar. 23
During spring break, we focused on figuring out an ideal paint option for the mice. We
determined that paint with titanium dioxide would produce the strongest contrast and would be
reasonable to apply to mice because it is non-toxic. We also ordered IR reflective paint that we
discussed with our client. This way, we had multiple options to test this week.
On Tuesday, we went to meet our client and finished up the penultimate step of the verification
process. We applied three paint samples (two mentioned above and one as a mixture) to black
pieces of paper and then tested to see if the paint was visible in a dark room under infrared light.
We noticed that all three options were visible in IR light; there was strong contrast seen with the
paint.
On Thursday, we met again with our client but were ready to apply the paint on the mouse ears.
We used two samples and eliminated the darker colored paint, as it would be difficult to track
this during visible light. When we applied the paint to the mouse’s ear, the mouse licked it off
with its hind legs. We did see contrast when placing the mouse in a cage in a dark room using an
IR light camera. However, this is an unfeasible approach because the quantity of paint on the
mouse’s ear would probably not stay constant over time and may irritate the mouse. Therefore,
we decided to apply paint to the ear tag and then apply the ear tag to the mouse’s ear. We tried
this and saw decent contrast with the camera. The concern we have is that if our code can
accurately detect the ear tag, which is a small object and may not have sufficient IR light.
This weekend, we will work to see the size of objects the code can track and try to further
develop the code to track the mouse’s position in two directions (along both sides of the cage). If
this goes well, next week, we should be able to initiate preliminary tracking of mouse movement
by painting the ear tag and collecting videos. We will also ask our client to track controls to start
our validation part of the project.
Weekly Report #10
Mar. 26 - Mar. 30
This week, we worked to set up the cameras for visualizing the mice. The cameras had to
be set up to limit the amount of artifact from the background, yet still provide a large amount of
signal to visualize the mouse. The two important aspects that needed to be considered was
eliminating the glare from the infrared source and eliminating the noise from the bedding. The
glare was removed by placing tape over the lightbulb of the infrared source. The bedding noise
was removed by aligning the camera with the same z-height as the bedding, allowing us to filter
out the bedding from the camera field of view though post-recording process.
After that, we examined applying the paint to mice. Unfortunately, all types of paint peel
off once applied to the ear tags because the ear tag puncher scrapes off the paint as the tag is
placed on the mouse. However, if the ear tag is painted after being applied to the mouse, then the
mouse will lick off the paint before the paint is dried. Therefore, we could not create an effective
way to apply the paint to the mouse’s ear tag. Moving forward, we will try to monitor the mouse
movement without any contrast agent.
Using some trial and error this past week, we successfully circumvented the paint
problem. Typically, the camera uses visible light during daytime but switches to IR light during
nighttime. In our current setup, we found it would be best to use IR light during both the daytime
and the night time (disabled the visible light setting). This way, we are creating a frame where
the object that has the highest intensity of black color is the mouse. Therefore, with our program,
we can track the black object successfully using filters to remove the top and bottom portions.
Weekly Report #11
Apr. 2 - Apr. 6
This week, we went back to check up on our camera placement that we set last week.
Because the cameras were sandwiched between the railings but lacked any rotational support, the
wires slowly pulled the cameras to different angles, making it impossible to use them to visualize
the cages. Therefore, we decided to remove the cameras from between the rails, place them on a
level surface, and simply stack boxes to raise the mice cages up to the elevation to be visualized
with the cameras. After checking throughout the week, it appears that this camera setup is stable
and will not move like the previous setup did.
After setting up the cameras, we analyzed five different mice for two days to collect a
baseline level of activity. Then, Four out of the five mice were sent through the breeding process.
Breeding takes about 3-4 days, and it is unsure whether or not a pregnancy occurs after breeding.
So, next week we plan on placing the mice back into their cages and observing if there is a
significant shift in their sleep cycle. The fifth mouse was not sent through the breeding process
as a control for any sleep cycle shifts that may occur due to events in the containment room.
Lastly, the code for the program was further enhanced. To remove noise created from the
bedding and surroundings outside the cage, we implemented a tool that the user can select a
rectangular region of the camera encompassing the cage. Because it was difficult to detect small
levels of paint on the mouse, we decided to simply track the entire mouse itself – and optimized
the color and object size to succeed in this endeavor. We recently managed to get a small sample
of data and have started analyzing our code’s ability to detect movement patterns using this data.
We will then optimize our code using this sample.
Weekly Report #12
Apr. 9 - Apr. 13
This week, we have finalized the mice that we will get data for to use in our final report.
We have seen 4 mice that have mated (as detected by a copulation plug). These 4 mice are also
being recorded with a camera for 4 days post-breeding. There is also one mouse that we did not
mate, and this will be used as a control mouse for the remainder of data collection. We will
analyze data for all five mice and generate position vs time graphs to hopefully detect a change
in circadian rhythm during pregnancy (i.e. a change in sleep-wake cycle).
The challenge we have faced is time pressure. While we developed the motion tracing
software and successfully completed V&V in early March, our project is validation-heavy. It
takes a long time and numerous attempts for mice to mate. While we used four mice as our
controls, only one of those mice have mated. The other three had to be removed and replaced
with new mice so that we could obtain sufficient data. We were fortunate to have one mouse that
is always used as a control; this way, we can compare pregnant mouse data with a non-pregnant
control. However, each mouse has individual variations, so it is not as ideal as comparing each
mouse’s sleep-wake cycle with itself prior to pregnancy.
The files that the software tracks the mice uploads dat files, whereas for our code, we
need to use an avi file format. We have found an online converter, which has allowed us to
successfully process a few frames this week of our video. However, we are currently looking for
a software that can convert an entire repository of dat files into avi fles. This makes it easier for
the client than having to convert each of the dat files (which is around 5 minutes long) to
generate a respective avi file to record mouse data.
Looking forward, we 1) will hopefully find the software by this weekend that can convert
our file format, 2) be able to run the data we have generated so far to obtain graphs, and 3)
analyze the graphs to find the respective shift in sleep-wake cycle. By the end of this week, we
hope to get an accuracy measurement of our method that we will be able to publish in our final
report. On Monday April 23, our group will also give a presentation on our work thus far.
Weekly Report #13
Apr. 16- Apr. 20
This week, we have made good progress to finish our project and write a compelling
report. We found the software that could convert our dat files to avi files that could be inputted
into our program. We used D-Link, which is a software used in the England Lab. The researchers
and technicians use D-Link to change recording schedule and record mouse movement.
Additionally, D-Link records mouse movement and uploads dat files. We had to convert every
dat file, that is, each camera’s recordings for each mouse, separately to retrieve avi files.
We had two options regarding the conversion. We could either convert the entire file into
an avi, which was very time consuming, or extract a certain time span for each file. We started
with the control mouse and extracted an avi file spanning from 4-8PM, which helps us obtain
data two hours before lights out and two hours after lights out. We also knew from our clients’
study that the mouse’s sleep-wake cycle shifts earlier in time; we hypothesized that the mice
would show higher bursts of activity between 4-6PM. After graphing velocity time graphs from
4-8PM, we noticed that the control mouse has limited activity before 6PM but is highly active at
and after 6PM. Therefore, we were confident to select a 4-6PM timespan for the rest of the mice,
as we wish to not necessarily examine overall activity throughout the day but a shift in sleep-
wake cycle.
Additionally, we heard exciting news from our client. Initially, we set up four
experimental mice which would be bred. Our client only noticed successful breeding in one
mouse. Therefore, he exchanged the three mice with three other mice that he tried to breed in the
mouse house. This way, we were sure that we would have enough data, as presence of a
copulation plug does not correlate well with pregnancy. But, it turned out that two of the three
mice took longer for a copulation plug to form. Therefore, we now have three out of these four
mice where we have baseline data (i.e. pre-breeding data) and can compare it to post-breeding
data. For the last mouse, where we only have post-breeding data, we can use the control mouse
that was never bred as a comparison.
This week, we also plugged in data from the four experimental mice into our program.
We talked to our clients and used a threshold of 15 minutes as an indication of pregnancy. That
is, if a mouse is active for 15 more minutes in the 4-6PM timeframe compared to its baseline
data, then we would consider the mouse pregnant. We took the velocity vs time graphs and
converted them to an inactive/active graph. Basically, we defined a low threshold for inactivity,
which corresponds to small vibrations during sleep. Therefore, we measured the time at which
the mouse was active or inactive. Based on the shifts in data, we predicted that three of the four
mice were pregnant, while the last was nonpregnant. Our client sacrificed the mice and discussed
that all four mice were pregnant. Therefore, our method is 75% accurate, although more data
must be tested given the small sample size (n=4).