FAQ

XraySetu solution is a cloud-hosted AI-based service that analyzes Chest X-Rays using machine learning algorithms and generates a patient report showing suspicious abnormal regions in the lungs and detects if the person is likely positive for COVID, pneumonia, or other lung abnormalities.

Tested and validated on 125000 X-Ray images from standard Open Source NIH X-Ray dataset with the following results.

Sensitivity : 98.86%

Specificity: 74.74%

AUC: 0.9316

Using XraySetu, a rural doctor can get an interpretation of chest XRay within minutes.

1. A rural doctor takes a phone picture of the Chest X-Ray of a suspected patient

2. The doctor sends the X-Ray pic Over to a Whatsapp number XraySetu (+91 80 4616 3838) via chatbot

3. At XraySetu end, Image review is done for anonymization and sends it to the XraySetu AI service

4. XraySetu AI Service Analyses the X-ray image using specialized machine learning and deep learning algorithms built for detecting unique Covid features in the lungs.

5. Full 2-page report with a probability of COVID, pneumonia, markings on the lung image in minutes, with explainable AI with annotations, is automatically generated.

6.The doctor receives the report from the same WhatsApp number on his phone

7.Doctor reads the report and appropriately advises the patient

• Whatsapp Interface makes it a simple and accessible service
• Rapid automated interpretation enables early intervention for covid patients
• Supplemental screening modality to RTPCR addressing delays on RTPCR reporting
• Works for both analog and digital XRay machine
• No IT integration needed with existing systems. Easily pluggable for any medical unit.

Technical Differentiators:

• Works on low-resolution images sent over the mobile messaging system
• Explainable AI: Semantic annotations of affected areas for review by doctors
• Framework suitable for several lung abnormalities and variants of infection
• Localized heatmap for easy review

The solution is developed in collaboration with IISc, Niramai along ARTPARK.

ARTPARK - a unique Not-for-Profit Foundation has been set up by the Indian Institute of Science, Bengaluru with support from AI foundry in a public-private model with $32 mn seed grant from DST(Department of Science & Technology (DST), Govt. of India) and Govt. of Karnataka. ARTPARK is a unique experiment to promote the creation of a new foundation for innovation and commercialization of global cutting-edge research for India. ARTPARK will promote technology innovations in AI & Robotics leading to societal impact, by identifying and executing ambitious mission mode moonshots in Healthcare, Education, Mobility, Infrastructure, Agriculture, Retail, and Cyber Security focusing on problems unique to India.

www.artpark.in

300 plus doctors have been connected with XraySetu in the last 10 months. Most of the Doctors are from rural parts of India. We are getting queries mostly from Medical officers at PHC (Primary Health Centre). XraySetu has been specifically designed to identify covid positive patients even from low-resolution Chest X-Ray images sent over Whatsapp. It also has semantic annotations of affected areas for review and localized heatmap by doctors to help them verify it easily with other modalities as well. Since its launch, it has already served close to 1200+ reports so far from the interior parts of India.

Dr. Padmanabh Kamath, Prof & HOD Cardiology, KMC, Mangalore, who has been an early advisor and user of XraySetu, remarked “ I have been the biggest proponent of taking healthcare and technology to the underprivileged and rural areas.” Another early user of the service, Dr. Anil Kumar A D, Medical Officer of Health, Shimoga in Karnataka remarked, “I am very happy to witness how technology is helping to get the quick diagnosis of patients. XraySetu has been helping us to get the analyses of the patient in a very quick time. It is a very easy, reliable, and accessible solution for doctors. ”

ARTPARK is an autonomous section-8, not-for-profit company, promoted by IISc and supported by AI-foundry, seed-funded by DST (Department of Science and Technology ) and GoK (Government of Karnataka) to the tune of $32 mn (230 cr).

ARTPARK’s goal is to promote technology innovations in artificial intelligence (AI) and Robotics to lead societal impact by executing mission mode R&D projects of strategic health in healthcare, education, mobility, infrastructure, agriculture, retail, and cybersecurity, focusing on problems that are unique to India.

RTPCR test is the primary Covid-19 test but during the second wave of COVID-19 cases rising across the nation, large-scale testing and early intervention are absolutely critical to saving lives. Due to the sheer load on the system, reports of RT-PCR are being delayed whereas CT scans are only available in bigger cities. When we look at the situation in rural parts of the country, there are limited RT-PCR test centers, and even for a sample to reach a test center, it takes one to two days. Hence, quick screening results are important to ensure early intervention and better treatment efficacy.

XraySetu is an accessible WhatsApp-based screening service that enables rapid identification of suspected COVID patients using AI algorithms over Chest X-Rays. This allows doctors to identify Covid patients at an early stage and save lives.

XraySetu performance :

Tested and validated with over 1,25,000 X-Ray images from the National Institute of Health, UK as well as over 1000 Indian patients from the first wave, XraySetu has shown excellent performance on open-source standard data sets as below:

Sensitivity 1 : 98.86% Specificity 2: 74.74% AUC3: 0.9316

Understanding the rural background, XraySetu has been designed to identify covid positive patients even from low-resolution Chest X-Ray images sent over Whatsapp.

1 Sensitivity (True Positive rate) measures the proportion of positives that are correctly identified (i.e. the proportion of those who have some condition (affected) who are correctly identified as having the condition).

2 (True Negative rate) measures the proportion of negatives that are correctly identified (i.e. the proportion of those who do not have the condition (unaffected) who are correctly identified as not having the condition).

3 In general, an AUC of 0.5 suggests no discrimination (i.e., ability to diagnose patients with and without the disease or condition based on the test), 0.7 to 0.8 is considered acceptable, 0.8 to 0.9 is considered excellent, and more than 0.9 is considered outstanding.

We took this idea from Italy where doctors were using Chest X-Ray to detect Covid-19 during the 1st wave. On the other hand, in India, the healthcare infrastructure is uneven and quite sparse in rural areas. We have only 1 radiologist for 10 lakh people in rural India. We realized that AI interpretation can be a big scaling advantage in the healthcare sector in India. For that, we initially started a WhatsApp group for rural doctors, who could send us the image to analyze the same. It quickly hit the ceiling of 255 members, mostly doctors, who were using it for their patients. We then decided to convert it into a simpler chatbot service, which now can be used by any doctor from all parts of India.

The automatically generated report gives annotated images marking 15 different lung abnormalities along with the probability of covid and pneumonia. With more images and data, we believe that we can screen more lung abnormalities accurately as well.

We have evaluated our models with the open Covid-19 data set of NIH, UK, and got an AuC of 0.93, which is a very good performance for any ML. Our model however is tuned for low-resolution phone pictures of X-Rays, which is suitable for WhatsApp-based delivery. While RTPCR tests remain the gold standard, in rural India, the major challenge is not only about the accuracy, but rather at what time the RTPCR tests become available. The core idea of XraySetu is to give doctors another screening option for Covid 19 and plan early intervention in case of corresponding direct observations so that lungs are not compromised and lives can be saved. Otherwise, we are observing the practice of people being given unnecessary medications and steroids, which are creating another set of problems. XraySetu is getting robust results over well-taken Chest Xray images sent over WhatsApp with high sensitivity.

ARTPARK (www.artpark.in) is an autonomous section-8, not-for-profit company, promoted by IISc.

This joint initiative (ArtPark & IISC) is to promote technology innovations in artificial intelligence (AI) and Robotics to lead societal impact by executing mission mode R&D projects in healthcare, education, mobility, infrastructure, agriculture, retail and cybersecurity, focusing on problems which are unique to India.

NIRAMAI Health Analytix is a Bangalore-based AI startup addressing critical healthcare problems through automated solutions. Winner of multiple international events, NIRAMAI is the only Indian company listed on cbInsights Top 100 AI startups in the world. A flagship product of Niramai is its patented Universal Cancer Screening Method using AI-driven thermal imaging.

Website: http://niramai.com/

XraySetu is a collaborative effort of ARTPARK , Niramai & IISC.

ARTPARK is an autonomous section-8, not-for-profit company, promoted by IISc and supported by AI-foundry, seed-funded by DST (Department of Science and Technology ) and GoK (Government of Karnataka) to the tune of $32 mn (230 cr) to help translate cutting edge research into solutions that can impact society.

Solutions like Xraysetu can be extended to cover other lung abnormalities with more data and research to empower healthcare infrastructure in rural India. India is planning to spend 64000 cr in the next 6 years to revamp the healthcare infrastructure. Instead of building just physical PHCs, we need to think ahead and plan for future Covid like challenges, which have overwhelmed the current infrastructure. We believe that leveraging recent advances in exponential technologies like AI & Robotics to provide state of art care to everyone is the model that India can embrace. Recent innovations applying exponential technologies like AI & Robotics in prevention, early diagnosis, and remote delivery, can be modeled in the form of next-generation digital PHCs to take effective care to the last mile. This would be the most effective and cost-efficient path for a vast and diverse country like India.

By bringing global talent, industry, and academia together at ARTPARK, we can create a model for the next generation of healthcare for India, which is truly inclusive for 1.36bn people.