An artificially intelligent pathologist bags India’s biggest funding in healthcare AI

Photo credit: Pixabay.

The World Health Organization requires a pathologist to spend 20 minutes examining a blood smear on a slide under a microscope before ruling out malaria if no parasites are seen. You can imagine how fatigue, an urge to get home, or some other distraction can disrupt that requirement – and the consequences of the disruption. Wouldn’t it be better for everyone concerned if the spotting of parasites could be automated? That’s the sort of thing healthcare AI startup SigTuple does.

Today SigTuple, which uses computer vision and artificial intelligence for diagnosis, announced series A funding of US$5.8 million. This is the largest investment in a healthcare AI startup out of India. It will help SigTuple scale up and go global.

Accel Partners led the round. Other investors include IDG Ventures India, Pi Ventures, Axilor Ventures, and Endiya Partners. Solid backing for the Bangalore-based startup from marquee investors signals the arrival of Indian AI on the world stage. Ex-Googler Amit Singhal, who is now senior VP with Uber, and VH Capital advised by WhatsApp’s Neeraj Arora also invested. Flipkart co-founders Sachin Bansal and Binny Bansal, who joined Accel in SigTuple’s initial funding of US$740,000, participated in this round too.

The prolonged diligence process was a ripple effect of negative reports about “nano-prick” blood-testing firm Theranos in Silicon Valley.

It took several months of discussions before the series A round could be closed, SigTuple co-founder and CEO Rohit Kumar Pandey tells Tech in Asia.

Partly, the prolonged diligence process was a ripple effect of negative reports about “nano-prick” blood-testing firm Theranos in Silicon Valley, after its faulty results to patients were exposed. As SigTuple’s first product Shonit was for blood tests, the Theranos question would immediately pop up.

Unlike Theranos, which came up with new testing models with smaller blood samples, SigTuple uses the same tests and slides for examination and diagnosis as pathologists do, explains Rohit. It only automates the process for more consistent and faster results.

See: Half the work people do can be automated: McKinsey

The pieces of the puzzle

SigTuple microscope with smartphone attachment. Photo credit: SigTuple.

The first piece of the puzzle is to digitize the slides. For this, SigTuple attached mechanical components and a smartphone to a regular microscope. The smartphone does an auto-scan of the slides.

SigTuple’s AI engine then takes over, learning to classify and tag the visual data. The fact that it can not only churn out results, but also support it with visual evidence makes it easily verifiable, points out Rohit. For example, it doesn’t just tell you there are 200 neutrophils in the blood sample – it can also show the neutrophils.

Even a layman can see the self-learning AI engine at work, as it learns to classify and tag various elements in the images.

Pilot programs in hospitals