Gita Manjunath

A mammogram missed her cousin's breast cancer. By the time it was found it was too late, and she died.

About 6 months later a second woman in the same family was diagnosed. She died as well.

The woman watching this happen was an artificial intelligence researcher.

Her name is Geetha Manjunath.

She grew up in Bengaluru. She took her bachelor's degree in computer science at UVCE, then a master's and a doctorate at the Indian Institute of Science, where she was a gold medallist.

Her first job was at the Centre for Development of Advanced Computing, where she was part of the team that built India's first commercial supercomputer in the 1990s.

She spent 17 years at Hewlett Packard Labs as a principal research scientist. She then became a lab director at Xerox Research in India, heading their data analytics research.

Then the 2 deaths in her family happened, and she went and read about the disease.

What she found is worth knowing.

In India, roughly half the women diagnosed with breast cancer die of it. Her own words are blunt. Every alternate lady detected with breast cancer is dying.

The reason is almost never the treatment. It is that the disease is found too late.

Mammograms are the standard test. They use x-rays, they are expensive, they are uncomfortable, they need a large fixed machine, and there is a specific medical problem with them. In younger women the breast tissue is denser, and dense tissue and tumours both show up white on the scan. So the tumour can hide.

That is what happened to her cousin.

She started working on it in her spare time, on weekends, with a small team, while still employed.

The idea was to use thermal imaging instead. A cancer uses more energy than normal tissue and the blood flow around it increases, so the area is slightly warmer. A thermal camera can see that heat pattern from a distance, without touching the body and without any radiation.

The hard part is reading it. Heat patterns are messy. That is where her actual expertise came in. She trained machine learning software to interpret them.

Their first test was with a medical college in Manipal, using thermal images of 72 cancer patients.

It worked.

In 2016 she left her job and started a company with a former colleague, Nidhi Mathur. They called it Niramai, which stands for Non Invasive Risk Assessment with Machine Intelligence, and which in Sanskrit also means free from illness.

The test needs no radiation, no touch, no pain, and no one else in the room. It works on women of any age and any breast density. The equipment is portable, so it can be taken to villages.

More than 400,000 women have now been screened across over 250 locations in India. More than 2,000 cancers have been found.

The technology has regulatory clearance in the United States and in Europe. She holds 14 American patents.

She has one line she repeats.

Nobody needs to die of breast cancer, if you can catch it early enough.