Geetha Manjunatha (CEO) and Nidhi Mathur (COO), co-founders of NIRAMAI, a Bengaluru based deep tech startup that provides breast cancer screening solution using AI, had seen cancer in their family and felt deeply about solving this problem in the society. This led to the creation of NIRAMAI, which in Sanskrit means being without diseases. It also expands to “Non-Invasive Risk Assessment through MAchine Intelligence”.
Having developed a screening solution that uses Thermalytix, i.e. machine intelligence over thermography images, their cloud hosted analytics solutions uses big data analytics, artificial intelligence and machine learning algorithms for early and accurate breast cancer screening.
As we are celebrating women’s day for the month of March, we interacted with Geetha, one of the cofounders of NIRAMAI, to get an insight on her journey as a women entrepreneur in analytics space. She says “To have a career in analytics, a person should definitely be smart and a quick learner. The tools and API are evolving so quickly that the person has to be on a lookout for learning new things everyday. The person should have good patience – which females are usually better off”.
Breast cancer screening solution by NIRAMAI—ML and analytics at the core
The breast cancer screening solution by NIRAMAI is a non-contact, non-invasive, radiation-free method of detecting early stage breast cancer in women of all age groups. The core technology by the company called Thermalytix is a fusion of sophisticated machine learning algorithms over thermal images.
“Thermography is well known to sense earliest signs of cancer. However, traditional manual interpretation of a thermogram has not been accurate enough to become accepted as a standard of care. Interpreting 400000 colour values in thermograms and to diagnose breast abnormality is a huge cognitive overload to the radiologist. Use of machine learning enables automated analysis and helps in better interpretation of thermal images and considerably improves the overall accuracy of diagnosis”, shares Geetha, who confessed to have been in love with math since the time she can remember and enjoyed solving challenging problems. She was so interested in emerging technologies and innovative projects that she automatically drove herself towards analytics as a career option.
How does the screening solution work?
Geetha is quick to add that the process of screening using NIRAMAI solution is very simple. She further explains that the woman who wants to get screened is made to relax for about 10 minutes before taking the test. A high resolution thermal sensor is placed 3 feet from the lady to measure the temperature distribution on her chest and create thermal images. Then the NIRAMAI software analyses these thermal images to automatically generate a screening/ diagnostic report and sends a radiologist-certified report to the lady. The test is completely privacy aware, as no one touches or even sees the person during screening.
The patented algorithms automate the process of analyzing the 400,000 temperature values measured per person. Unique thermal patterns and image characteristics that are typically used by medical professionals while detecting cancer in other modalities like XRay or Ultrasound are also used in NIRAMAI algorithms to make the report accurate as well as understandable by the radiologist.
Since it’s radiation-free, the test is very safe to undergo and works on women of all age groups. “This is unlike mammography which is based on X-Ray and is recommended for women above 45 years only once in 2 years. It is also noncontact and does not require any breast compression hence not painful. “Since the equipment is very portable, it is amenable to be used in outreach programs being a rural camp or urban corporate screening”, she adds.
Currently, there are three hospitals in Bengaluru who are now providing NIRAMAI solution to public.
Overcoming Challenges on a personal and professional front
“Analytics and AI are the most controversial topics when it comes to healthcare. It takes a lot of effort, discussions and experimental trials before a doctor agrees an AI tool as an aid to his diagnostic workflow. Being a startup, we face further challenge due to the size of our company in comparison to companies like Philips or GE”, she shares.
Another challenge is that once an initial validation of the solution is shown, there will definitely be other players who would like to copy. However, she believes that though there are other AI solutions for cancer screening, very few startups or research organizations are using AI in thermal imaging. “Our 10 patents in this area are one way of protecting us. The data we have collected so far is also a great asset that our competitors don’t have”, she adds.
On a personal front she believes that life is a struggle. Her mantra is, if you enjoy doing what you are doing, then it doesn’t look like a struggle at all. “Having a good understanding and supporting husband and in-laws is very critical to have a peace of mind while at work. I have always shared highlights of my work and kept things transparent with them, which makes them feel like supporting and contributing to my work”, she shares.
Growth story so far
Geetha shares that as they started NIRAMAI, their plan was to use advanced image processing to identify the tumour location in a cancerous patient, so that they could help in the surgical procedure. “Later, we felt that it could have wider usage if applied as a screening methodology. We then collaborated with a diagnostic center to get data about healthy, benign and malignant women. This data was used to create machine learning models that has the ability to classify cancerous patients versus others”, she shares.
For the startup, it has been a very challenging problem due to the high levels of sensitivity and specificity that is expected for a critical healthcare problem. To add to that, the input data can have many errors, missing information, incorrect data capture by humans, etc. The result of automated analysis should be good despite the errors, which is a very hard task, and they are constantly working on it.
The personal growth story of Geetha is as interesting. After her masters from IISc, she joined government research institute, CDAC, where she learnt about distributed computing and tools for the same. She later joined as almost the first member of Hewlett Packard Labs in India, where she worked on cutting edge technologies ranging from embedded systems, web computing, grid computing, cloud and finally analytics frameworks. It was during this time that she went back to do doctorate from IISc and brushed up her math skills, enabling to lead analytics research for Xerox in India. “After a few years, I decided to do a startup with one of the ideas we had explored and founded NIRAMAI”, she shares. NIRAMAI is currently seed funded and would like to go for next round of funding in April this year. They plan to use it for getting regulatory clearances outside India.
The startup plans to apply AI in the overall NIRAMAI solution, and not just for classifying normal and abnormal subjects. Geetha believes that it can be used to ensure better quality in data collection, simpler usage as well as bring in explainability in results.
Also, since breast cancer is a huge problem in India, the NIRAMAI team believes that it is important to create awareness for the need of regular screening. One of their key areas of work is to encourage the ladies to know that their screening is safe and early detection will help them keep any cancer related risks at bay. Women can register for screening at http://niramai.com.
Key takeaway for women in analytics
Geetha says that it is only recently that many women are choosing math and engineering career. Parents and teachers should identify top talent and encourage them to study further and take up challenging opportunities without worrying if it can be done by a girl or not.
“We read that girls do better than boys in most board exams, why should they stop after engineering?One needs to do a thorough survey to understand the gap and provide the right support for advancing the career of girls. Scholarships for girls for further studies will also help, especially for those coming from families who may not prioritize girl’s advanced education”, she adds.
She believes that there is also a need for incorporating more women in technical career. “As a manager, I have always found women employees to be more dependable and extremely sincere at work. They also take full ownership of their activity and improve teamwork, creates an atmosphere of a mini family”, she shares.
“Having said that, I don’t believe in quota system for women. Corporates need to make sure that they don’t bend interviewing rules or reduce the requirements just to get in a woman for a job. She needs to feel that she deserved the seat. There are many capable ladies, corporates need to spend the time to search and find them”, she says in the concluding.
Try deep learning using MATLAB