BloodDrop

Malaria is a common disease which is preventable and curable, yet almost 228 million cases of malaria were found in 2018 and the estimated number of malaria deaths stood at 408,000 with the numbers increasing each year (WHO). The major problem is the detection of malaria due to lack of affordability, accuracy and trained micro-biologists in the developing and underdeveloped countries.

We're using Ball Lens as a microscope to take pictures of the blood specimen using PiCamera and with Raspberry Pi, we’ll be sending pictures to the Cloud where we have built our Deep Learning model using Rest-Net Architecture and Transfer Learning for detecting Red Blood Cells infected with the Malaria parasite.

Our solution will help in the detection of Malaria with the same consistency and accuracy all around the globe. This product will open a new way to detect diseases with blood sample and impact lives of people directly.

Malaria is a very common disease which is preventable and treatable yet according to a report by WHO, 228 million cases of malaria were found in 2018 and the estimated number of malaria deaths stood at 435,000 with the numbers increasing every year. Children aged under 5 years are the most vulnerable group affected by malaria, they accounted for 67% (272, 000) of all malaria deaths worldwide in year 2018. The WHO African Region carries a disproportionately high share of the global malaria burden. In 2018, the region was home to 93% of malaria cases and 94% of malaria deaths. One of the major problems is the detection of malaria due to lack of affordability and less accurate detection as they lack trained-microbiologist. Even though in some areas tests are conducted through Malaria RDTs by public sector but it requires control over test sensitivity, storage and transport which are frequently be more difficult to assess. On a large scale, Microscopy is considered the gold standard for the detection of disease due to high accuracy and cost effectiveness as compared to RDTs but due to the lack of trained micro-biologists in the most parts of the world it is not practiced.

Our solution aims to change the traditional method of detecting malaria disease. We designed the product which will replace the microscope and automate the whole process. Our device is using Spherical Ball Lens of 1mm diameter which is acting as a microscope which magnifies the blood sample at a magnification of 400X. We are further working on increasing the magnification up to 1400X. The image of magnified blood sample is captured with Raspberry PiCamera module and Raspberry Pi 3 Model B will be sending pictures to the Cloud where we have built our Deep Learning model using Rest-Net Architecture and Transfer Learning for the detection of Red Blood Cells infected by the Malaria Parasite. We build our product to increase the availability of the detection of malaria and giving them the result instantly and accurately. Our Model is made to be the tool to detect Malaria in a much efficient and in a cost-effective way. Detecting malaria is just the start, we aim to detect all kinds of diseases which can be detected from blood sample in future.

We believe that medical care should be available to everyone from an urban city of New York to a remote village of Africa irrespective of the technical drawbacks, connectivity, and lack of medical professionals or well infrastructure. Our solution for detecting of Malaria will cover the whole world, from developed countries to endemic countries with different market strategies and policies keeping in mind the standard of living of people so that it will be beneficial to all.Detection of Malaria is a start.Think of a world where a blood sample taken from a small village in Africa can be examined in the United States in real time which is been stored in the Cloud.In case the device detects an unknown parasite the images of the blood sample could be viewed by the researchers all around the globe and safety measures could be taken in a much faster way. With the increase in population, detection of diseases would become difficult. We could use this device to detect diseases in much faster and an efficient way and maintain a database of the diseases in the different parts of the world in real-time and focus on the regions which are most severely affected.

According to WHO, nearly half of the world’s population is still at risk of the malaria disease. As of now, many countries with high burden of malaria have a weak surveillance system or no system at all. Our solution will not only detect diseases in a much faster and efficient way but also provides enhanced disease surveillance system for the detection of diseases with great accuracy and cost effectiveness all around the globe. Our product is targeted to all types of market but will be more affecting to the people of sub-Saharan Africa as they are the most vulnerable place.

For detection of malaria, we are dependent on microscope from many decades and there are many drawbacks of using it like they are quite bulky, immobile, expensive and requires an expert micro-biologist to operate and spot the strain of the parasite from the given sample and it also takes time. Our product with the help of Artificial Intelligence revolutionizes the way diseases have been detected so far. We bring the replacement of microscope buy using spherical ball lens for magnification which is infinitesimal as compared to microscope and also much cheaper. Now, we are integrating ball lens with PiCamera for image capturing. With the help of Deep Learning, we automated the process of detecting disease (currently for Malaria and further expanding it for other diseases). Our goal is to provide most affordable and accurate detection of diseases from blood samples and make it available to everyone present on this planet.

Our prototype is using Spherical Ball Lens, Raspberry Pi 3 and PiCamera in the hardware part and in the software part, we are using Deep Learning with Rest-Net Architecture and Transfer Learning and python based application for Raspberry Pi.Currently we are using the ball lens of 1 mm in diameter which provides a magnification of 400x which is used to take the magnified images of the blood sample using Raspberry Pi Camera , the ball lens is fitted above the Pi camera and the blood sample is prepared over the glass slide using the Leishman Staining to identify the Red Blood Cells infected with the Malaria Parasite which is placed over the ball lens and illuminated using White light. The images are taken using the application build by us for the Raspberry pi while displaying the blood sample on the screen which is then processed , segmented and then sent to Cloud by the Raspberry Pi for detection using the Deep Learning model we have developed which is based on ResNet architecture and Transfer learning for the detection of the cells infected with Malaria. 

Here some images of the blood sample taken from our device including the device itself.

Our very first focus area is on increasing the magnification of the device by using more smaller lens and finding an efficient way to integrate it to the system as it is would be very small to handle with open-hands as compared to the lens we are currently using. Our next work will be to do more rigorous field testing as to test the potential of our system in the real world. Next, we will be designing our propriety single-board computer to replace Raspberry Pi as it has lot more computing power and features which we don’t require. By using our own board, we will be able to cut the cost. This is our goals within next year.

Our 5 years goal is to complete deployment of our product in public and private sector and generating sales and scaling it up to different market as much as possible. Also, we will be working on detecting different diseases with blood sample and covering more and more diseases.

Our biggest barriers we are currently facing to accomplish our goals in funding. As a student, we lack resources and financial support for working further on our product. Till now, we have completed the first prototype of our project completely from our personal savings.

Secondly, we are engineering student and have little to no knowledge of entrepreneurship and also lack connections and reach for the deployment of our product. We have completed our prototype without any guidance for mentor or any senior professors. It is our completely self-made project.

MIT Solve is a marketplace for social impact innovation which helps to bring extraordinary ideas to real world and support those ideas by providing right partnership and helping us to get involved in MIT’s innovation ecosystem and provides funds from community members to support entrepreneurs to drive lasting and transformational impact to the world.

Being selected as one of the Solver in MIT Solve, we will be able to reach different organizations or donors and will attract good amount of exposure to our product, also it will help us to raise fund which will help us to do further research and will be able implement our idea in the practical world at a much greater pace in the near future. Partnerships with organizations will expedite the process and we will be able to reach more and more remote parts of the world and creating good opportunity for us as well as for the people working with us.

Two

Our team comprises of two individual , we both are pursuing our bachelors in technology. I am in pursuing in Electronics and Communication Engineering and another teammate is pursuing in Computer Science Engineering. We have in depth knowledge in Embedded Systems , Advance Digital Design and Deep Learning. Our idea was also selected in Keysight IOT Innovation Challenge and we were among the top 3 finalists from all around the globe in our category. Above all we are passionate and have a strong believe that "We are innovators; we build things which will be a bridge connecting healthcare to the remotest parts of the world to the millions of people by detecting diseases and tackling them in a much efficient and faster way."

We plan to sell our product to public sectors like Not-for-profit Organizations, Government hospitals etc and in a diverse private for-profit sectors consisting of hospitals and clinics ,local pharmacies ,drug shops and itinerant drug seller.In this way we would be able to  generate revenue which would fund our research to detect other diseases using the blood sample. Most of the diseases would require similar magnification so detecting a new disease would require a new Deep Learning Model which would then can be implemented just by an software update in the actual product , which would come at a one time cost for a particular disease. And the organization would be able to select the update for the particular disease they require,thereby not  paying for any disease which are not present in that region.

Being a student the biggest barrier we have is funding for our project. Currently we have no funding or financial support for our project. Every thing we have done in our project was done from our own savings. So, being selected as a solver would be a help financially to continue our work at a greater pace. 

Secondly , MIT Solve would help us reach individual and organization who would help us to deploy our final product into various regions in the near future.

Coming from an Engineering background we have little to no knowledge in business administration. Partnership would help us to refine our business model and the revenue model and help in the product distribution in various affected regions around the globe.