BIOTECHNOLOGY IGNITION GRANT (BIG) is flagship programme of BIRAC, which provides the right admixture of fuel and support to young startups and entrepreneurial individuals. BIG is the largest early stage biotech funding programme in india. Funding grant of upto INR 5 Million (USD 70,000 approx) to best in class innovative ideas to build and refine idea to proof-of-concept.
As a startup that was awarded the grant it made sense to share what worked for us and what we learnt. We had applied twice for BIG, got selected the second time and got 50L from BIRAC (lesson: don't give up because the first application was rejected). We have decided to open our application so it might help others improve their application and bring their ideas to life.
To note: This might not be the best application, this is old and what we wrote when we started (this can be improved a lot). You might face more competition these days given BIG has gained more popularity. However wishing all innovators all the best.
Title Of The Proposal
Mobile Oral cancer screening device using Artificial Intelligence.
Skyfire Applied Intelligence Pvt Ltd
Centre for Cellular and Molecular Platform (C-CAMP), Bangalore.
Proposal Summary [Provide a brief one paragraph overview of the proposal, i.e. the idea and the problem it may solve and brief project plan.]
Oral Cancer is the 6th most common cancer in the world  and claims more than 10 lives every hour in India alone . The inaccessibility high cost of health care has resulted in patients visiting the hospital with advanced stages of cancer thereby associated with dismal prognosis. It is very important to screen oral cancer early as the tumor volume doubling time ranges between 15 and 40 days . The 2-year cumulative survival probability is 17 even after radiotherapy for advanced inoperable oral cancer . Currently, the most effective way to control oral cancer is to combine early detection and appropriate treatment. Because more than 90 of all oral cancers are squamous cell carcinomas, the vast majority of oral cancers will be diagnosed from lesions on the mucosal surface .
Our aim is to develop an accessible, affordable alternative method for healthcare delivery to screen non-communicable disease such as cancer. We are developing an AI algorithm which will be used to identify potentially malignant disorders precancerous and cancerous oral lesions by analyzing images captured by a camera device attached to any smartphone.
The significance of the proposed research and relevance to public health: By equipping existing health workers with our smartphone-based AI screening tool, we could screen the community and get instantaneous results on the field to refer for further diagnosis, thereby avoiding the need for resource-intensive training or the physical presence of an oral cancer specialist. Through accessible early detection and treatment, it is possible to reduce treatment costs by at least 20x and treatment time from 3 months to approximately 15 days while saving more than 37,000 lives a year.
Briefly state the Objectives and Proposed Approach [Describe how the proposed project addresses the problem. Clarify the current status of the innovation.] The description should cover the following points: 1). Strategy and/or methodology of work. 2). Scope and boundaries of the work, including any issues that will not be covered. 3). Data analysis (sample size,data collection)
Our aim is to reduce deaths due to oral cancer by bringing patients early to the hospital.
To meet the shortage of oral cancer specialists, there is a need to equip community workers with the AI based mobile tool to screen the community and get an instantaneous result on the field to further facilitate referral or more frequent follow-up. In order to achieve these goals, we are developing a screening device with an AI algorithm which can accurately identify oral cavity lesions in the field using images captured using an external camera and smartphone. We intend to keep the costs low by making an affordable camera and running the AI algorithms on the smartphone. Our AI solution taps into the market of smartphone, which is projected to reach a subscription of 6.3 billion by 2021. Not only has the smartphone market reached the masses, it has also penetrated the rural regions of the developing countries making it possible to provide low cost and truly universal access to vital health care.
Novelty [Explain how your idea is innovative and how it is different from the existing products in the markets or current state-of-the-art. Tabular representation of the difference between your idea and the other products in market or competitive product which are under development will be appreciated. Concrete market data is encouraged.]
To reduce deaths due to cancer it is important to understand where current solutions fail. Our observations have been inaccessibility to screening, costs of screening devices and the skills/ training/ expertise required to use that device. Our solution addresses all these above-mentioned barriers. It is projected that 6.3 billion smartphone subscriptions will exist by the year 2021 . Not only has the smartphone reached the masses, but it has also penetrated rural regions of developing countries making it possible for us to provide low-cost and truly universal access to vital health care. By keeping in mind the BIG 13 Reviewers feedback on the variability of smartphone camera parameters and taking advantage of the computational capabilities and accessibilities of the smartphone we intend to develop an inexpensive external camera that could be attached to the smartphone using USB. Our solution is more accessible, less invasive, less dangerous, less expensive, less time-consuming, and less physically and psychologically discomforting for participants. To the best of our knowledge, there is no public literature available in evaluating the AI based mobile application for oral cancer screening using images. However, AI-based mobile phone application for screening skin cancer , oral cancer prediction based on knowledge of risk habits  and Google has launched an Artificial Intelligence program in Thailand following India to screen for a diabetic eye disease which causes permanent blindness.
Opportunity [What is the potential societal and market impact? Provide details of the problem you propose to solve.]
Current devices in the field are very expensive and have very little penetration in rural India. We present a novel method of delivery using AI which does not require the physical presence of an expert oral oncologist for screening and is affordable.
However, digitized health data for training AI algorithms to classify oral cavity lesions are scarce globally. With India being the oral cancer capital and oral cancer being the 6th most common cancer globally we have a unique opportunity to develop an AI-based screening solution from India for the world.
With reduction of costs to more than 20x Insurance companies will be the most benefited as treatment costs reduces from Lakhs to thousands. Healthcare delivery using insurance companies also presents a unique opportunity and differentiation on how healthcare is currently being delivered globally. Our solution could potentially turn into a self-administered yearly screening solution for the 300 million people insured in the US. It also opens a large market only considering the number of people insured in developed nations. We intend to head progress towards truly accessible and affordable preventive healthcare solutions.
Challenges or risk factors associated with the project [What are the challenges and risk factors that you envision which may affect this project?] What are the critical success factors/potential barriers?
Apart from variation in the subject such as skin color, age, and oral stains the smartphone images also exhibit variability in factors such as zoom, angle, camera parameters, and lighting.
Technology risk mitigation:
The generalization of AI algorithms to account these photographic variations is the reason why we need sophisticated AI algorithms at the core of the product. As per BIG13 expert panel discussion and suggestion, we are developing a smartphone-based camera device to standardize the photographic variations such as zoom, angle, camera parameters and lighting.
Data collection risk:
Risk of the collection of more image-based oral cancer data to improve current accuracies.
Data collection risk mitigation:
Already engaged with Kidwai Cancer Insititute for collecting images with pathological evidence and have obtained clearance from the hospitals Scientific and Ethics board. These images which have pathological evidence will be used for testing and evaluation of algorithm. We are also currently collaborating with 10 doctors across the country to collect images.
Has any preliminary work been carried out? Give status of work done If no, please provide the background details.
Yes. Proof of concept for AI algorithm to accurately predict specific smartphone images is completed. Android application for data collection is available on Playstore. We have created a web-based platform for oral cancer specialists to remotely annotate/ screen the images clicked by anyone using our Android App. These annotated images are used for training the AI algorithm.
Pilot testing for the app was done, and we screened the high-risk population for oral cancer in the streets of Bengaluru. Amongst 34 screened people, 13 were identified from the images of their oral cavity as potential cancerous cases by an expert oral oncologist from Kidwai Cancer Institute, Bengaluru. The patients were asked to report to the hospital immediately. We independently organized a district-wide oral cancer screening camp in Mandya, Karnataka. We are collaborating with NGOs such as Karnataka Cancer Society, NOCPI and VChangeU for organizing oral cancer screening camps. We have initiated the process to engage with Kidwai Cancer Institute for getting clinically validated datasets for validating our AI algorithm. And we are also in talks with AIIMS Bubaneshwar, AIIMS Delhi and TMH, Mumbai for the same. We are also experimenting with alternative methods to collect datasets across India. Currently, we have 3000 unique patients oral cavity images
i. Does the applicant or the applicant company own any IP related to this project. If yes, give details.(Please mention Patent Number, Patent Title and Patent Assignee)
We have a Provisional patent filed bearing application number: 201841035013. We also have unique data on oral cancers from camps we have participated in.
ii. List Of Patents That Appear To Cover Any Part Of The Technology Of Interest Or Similar (And Possibly Overlapping) Technologies And Thereby Restrict The Freedom-To-Operate In The Envisaged Area. (Please mention Patent Number, Patent Title and Patent Assignee)
Expired Patents for Intraoral camera without AI : US6276934B1/ Dental camera/ Current Assignee: Miratech Dental Imaging Ltd US6459920B1/ Method for detecting and diagnosing epithelial cancer/ Current Assignee CYGNUS TECHNOLOGIES LLC ZILA BIOTECHNOLGY Inc
iii. If there are patents that are overlapping and may restrict FTO, does the applicant have the required license/s to practise these inventions for the purposes of the proposed project? Please provide license agreement details if any or provide information of the proposed next steps to obtain said license/s.
The above-mentioned patents are for the intraoral camera and require a clinician to diagnose. These patents have also expired as well. On a basic google patent search, we are confident that our innovation AI to screen oral cavity lesions from images is unique and there is no conflicting innovation happened on the same field.
BIG is an amazing initiative that helps idea stage startups bring their ideas to life. There are multiple steps what is listed here is just step 1. Post this you will have interviews in the centre chosen then one more in Delhi. Your BIG partner should be able to guide you once you clear step 1. Money is just one part of it. What BIG also does is expose you to the ecosystem of startups, incubators, mentors which will make a huge difference.
All the best once again for your application. Hope this helped :).
 NIH oral cancer fact sheet https://report.nih.gov/NIHfactsheets/Pdfs/OralCancer NIDCR .pdf
 B. Gupta, A. Ariyawardana, and N. W. Johnson, Ã¢â‚¬Å“Oral cancer in india continues in epidemic proportions: evidence base and policy initiatives,Ã¢â‚¬Â• International dental journal, vol. 63, no. 1, pp. 12Ã¢â‚¬â€œ25, 2013.
 A. R. Jensen, H. M. Nellemann, and J. Overgaard, Ã¢â‚¬Å“Tumor progression in waiting time for radiotherapy in head and neck cancer,Ã¢â‚¬Â• Radiotherapy and oncology, vol. 84, no. 1, pp. 5Ã¢â‚¬â€œ10, 2007.
 I. Chitapanarux, P. Traisathit, N. Komolmalai, S. Chuachamsai, P. Sit- titrai, T. Pattarasakulchai, R. Tananuwat, D. Boonlert, P. Sripan, and A. Iamaroon, Ã¢â‚¬Å“Ten-year outcome of different treatment modalities for squamous cell carcinoma of oral cavity,Ã¢â‚¬Â• Asian Pacific journal of cancer prevention: APJCP, vol. 18, no. 7, p. 1919, 2017.
 A. Esteva, B. Kuprel, R. A. Novoa, J. Ko, S. M. Swetter, H. M. Blau, and S. Thrun, Ã¢â‚¬Å“Dermatologist-level classification of skin cancer with deep neural networks,Ã¢â‚¬Â• Nature, vol. 542, no. 7639, p. 115, 2017.
 Esteva, Andre, Kuprel, Brett, Novoa, Roberto A, Ko Justin, Swetter Susan M, Blau Helen M, Thrun Sebastian, Dermatologist-level classification of skin cancer with deep neural networks. Nature volume542, pages115Ã¢â‚¬â€œ118 02 February 2017 .
 Speight PM, Elliott AE, Jullien JA, Downer MC, Zakzrewska JM. The use of artificial intelligence to identify people at risk of oral cancer and precancer. Br Dent J. 1995 Nov 25 179 10 :382-7. PubMed PMID: 8519561