Butterfly Network's Single-Chip Ultrasound Technology Attracts $100M Investment from Jonathan Rothberg

Can Single-Chip Ultrasound Technology Reduce Medical Imaging Costs and Make It Affordable for Everyone?

Hold a scanner no larger than an iPhone to your chest, and you can view vivid 3D images of the human body’s interior.

This is a new type of medical imaging device sponsored and developed by entrepreneur Jonathan Rothberg. Rothberg claims that he raised $100 million to create this innovative product, which is “as affordable as a stethoscope” and can “increase doctors’ work efficiency by 100 times.”

As described in the patent documentation, this technology relies on a novel ultrasonic chip that may ultimately become a new method for destroying cancer cells using thermal energy or delivering information to brain cells.

Will this technology guide the future direction of medical imaging?

For Rothberg, applying semiconductor technology to solve biological problems is second nature. Earlier, he sold his two DNA sequencing companies, 454 and Ion Torrent Systems, for more than $500 million. With the substantial profits from these sales, Rothberg has been able to enjoy life aboard his luxury yacht, named Gene Machine, while devoting himself to researching DNA sequencing, an area of personal interest.

This new imaging system was developed by Butterfly Network, a company founded nearly three years ago. The company is part of 4Combinator, a startup incubator established by Rothberg, and is among its most advanced ventures. 4Combinator aims to integrate medical sensors with “deep learning” in artificial intelligence.

Rothberg did not disclose exactly how the Butterfly device works or what it looks like. “For details, please wait for the official launch of the device. We will roll out this product within the next 18 months,” he said. However, Rothberg assured that the device would be compact in size and priced at only a few hundred dollars; it can connect to a smartphone and is capable of diagnosing breast cancer or displaying images of a fetus.

Butterfly’s patent applications reveal its aim to design a compact, multifunctional new ultrasound scanner capable of generating 3D images in real time. By holding the device against the chest, users can gain insight into the human body through a “window.”

Conceptual diagrams submitted by Butterfly Network to the patent office show that the product is a compact 3D ultrasound imaging device.

With $100 million in funding from Rothberg and other investors, including Stanford University and Germany’s Aeris Capital, Butterfly appears to have placed the largest bet in history on this emerging technology. The technology involves directly etching ultrasound transmitters onto semiconductor wafers alongside circuits and processors; these devices are known as “capacitive micromachined ultrasonic transducers,” or CMUTs for short.

Most ultrasound machines use small piezoelectric crystals or ceramics to generate and receive sound waves. However, these components must be carefully interconnected and then linked via cables to a separate box for signal processing. If ultrasound elements could be directly integrated onto computer chips, they could be mass-produced at low cost, and it would be easier to create array configurations capable of generating 3D images.

“The concept for this product has been long in the making, awaiting someone to turn the idea into reality.”

In imaging tests, doctors more frequently use ultrasound, including using it to view images of babies during pregnancy in pregnant women, to find soft tissue tumors in the liver, and more recently, utilizing sound waves to heat cells for treating prostate cancer.

The idea of manufacturing miniature ultrasound chips dates back to 1994. At that time, under the suggestion of Stanford University professor Butrus Khuri-Yakub, Rothberg’s company produced the first-ever miniature ultrasound chip in history. Unfortunately, despite sustained interest from General Electric and Philips, the chip failed to achieve commercial success. This was because reliable operational processes had not yet been established, and mass production indeed posed significant challenges.

“The concept for this product has been long in the making, waiting to see if anyone could turn it from an idea into reality,” said Richard Przybyla, Director of Circuit Design at Chirp Microsystems, a Berkeley-based startup dedicated to using ultrasonic systems to enable computers to recognize human gestures. “Perhaps what it had been waiting for all along was sufficient funding and a sufficiently focused team.”

Rothberg said his interest in ultrasound technology stemmed from his eldest daughter, who is currently in college and has tuberous sclerosis—a condition that can cause seizures and kidney cysts. In 2011, he funded a medical trial in Cincinnati to investigate whether kidney tumors could be eliminated by heating them with high-intensity focused ultrasound pulses.

The trial results convinced Rothberg that there was still ample room for improvement in this technology. “The trial used an MRI machine to visualize the tumor and an ultrasound probe to heat it; the machine cost millions of dollars, yet its speed was far from impressive—more akin to a laser printer that takes eight days to produce a single printout, with image quality resembling crayon drawings made by my children,” said Rothberg. “Since then, I have been determined to develop a product that is 1,000 times cheaper than this $6 million machine, 1,000 times faster, and 100 times more accurate.”

Rothberg claims that Butterfly’s technology has a “secret sauce,” though he has not disclosed what it is. Speculation suggests that it may involve intelligent device and circuit designs capable of overcoming the physical limits and manufacturing challenges currently facing CMUT technology. One reason for this speculation is that one of the company’s co-founders, Nevada Sánchez, previously helped cosmologists design a significantly more affordable radio telescope. The signal-processing technique employed in that device, known as the “butterfly network,” gave the startup its name. Greg Charvat of Butterfly, who comes from MIT Lincoln Laboratory, developed radar systems there capable of seeing through thick stone walls to visualize the human body.

Sánchez and Charvat once demonstrated to the media a coin-sized image capable of rendering letters and numbers in fine detail. This image was captured using a prototype chip they developed earlier this spring. “The ultrasound industry has essentially remained stuck in the 1970s. Companies like General Electric and Siemens are also products of that outdated mindset,” said Charvat. He noted that with the emergence of new concepts such as chip manufacturing and radar technology, “we can achieve faster imaging, wider fields of view, and upgrade resolution from the millimeter scale to the micrometer scale.”

Ultrasound devices operate by emitting sound waves and then capturing the echoes. They can also generate focused energy beams; leveraging this capability, this chip-based device may eventually evolve into a novel system capable of killing tumor cells. According to recent findings, neurons can be activated by ultrasound, suggesting that such miniature devices could also be used to deliver feedback information to the brain.

“After I have accumulated tens of thousands of these images, I believe they will be able to make a difference for patients with Down syndrome or cleft lip; when people feel pressed for time, tasks often become unmanageable.”

Rothberg stated that his primary objective would be to launch an affordable imaging system on the market, making it accessible even in the world’s poorest regions. He noted that the system would rely heavily on software, including technologies developed by artificial intelligence researchers, to perform automated diagnoses by analyzing large volumes of images and extracting key features.

“We hope it can function like the ‘Panorama’ feature in the iPhone.” He was referring to a smartphone photography feature that allows users to pan the device to automatically generate a stitched panoramic image.

Rothberg also pointed out that, in addition to identifying objects and helping users locate them—such as fetal anatomical structures in prenatal screening—the system can also generate preliminary diagnostic conclusions based on pattern-recognition software.

“Once I have accumulated tens of thousands of these images, I believe they will be able to make a difference for patients with Down syndrome or cleft lip; when people feel pressed for time, tasks often become insurmountable,” said Rothberg. “But I will empower every technician to achieve this.”

In addition to Butterfly, Rothberg’s incubator has begun investing in three other companies, each of which secured $5 million to $20 million in seed funding. They are Lam Therapeutics, a biotechnology company developing treatments for tuberous sclerosis complex; Hyperfine Research, which maintains strict confidentiality regarding its research activities; and another company that has chosen to remain anonymous.

Compiled by Chen Xin | Edited by Mo Renying