IEEE Spectrum logo Continue to site ➔
ADVERTISEMENT

Tech Talk iconTech Talk

The Human Face Within Big Data

For three decades, Rick Smolan—a former Time, Life and National Geographic photographer—has taken global cultural snapshots through his Day in the Life series of coffee table books that explore a time capsule of activity involving a country, discipline, or issue.

The projects—produced by the New York-based Against All Odds Productions—which Smolan runs with wife and co-author, Jennifer Erwitt, and COO Katya Able, take around 18 months and involve upwards of 200 writers and photographers around the globe.

Their latest book, The Human Face of Big Data, out this week, takes a more encompassing approach to a topic than its predecessors. Tackling the idea of Big Data—mankind’s ability to collect, analyze, and act on an unprecedented amount of information in real time—the book uses photos, essays, and articles (including one by yours truly) to examine the phenomenon, and how individuals and companies are harnessing it for human benefit, while raising concerns about data ownership and privacy invasion.

“Having now spent a year looking closely at this emerging world of big data, I hope the book will spark a global conversation about both the tremendous potential good and the concerns we all need to have about who owns data that you and I generate,” says Smolan. "Right now it's primarily companies and governments who are thinking about the uses of Big Data. It's really important that each of us also thinks about how this is going to affect our lives."  

The project employs Big Data as a storytelling device and business platform. In the two months leading up to the book’s release, the Against All Odds team organized a three-city technology networking event and unveiled a website to gather digital behavior data from 300 000-plus anonymous volunteers. The results will be available free to researchers and academics next year. The project is self-published from sponsorship by several technology companies, primarily EMC2, along with Cisco Systems, VMWare, Tableau Software, Originate, plus FedEx. It’s the first coffee table book to use the Aurasma mobile app, which triggers related multimedia when readers hold smartphone and tablet cameras to yellow key graphics on its pages. There’s also an iPad app, with profits going to charity: water.

Hopeful But Wary

The book is a hopeful look at Big Data, highlighting its impact on agricultural efficiency, weather and earthquake prediction, fertility and genome mapping, space junk, crime solving, eradicating disease, and tracking endangered species, to name a few.

In one particularly striking example, The Artificial Retina—looking like something straight out of Star Trek—Weill Cornell Medical College’s Sheila Nirenberg used Big Data to circumvent certain types of blindness, such as that caused by damaged retinal photoreceptor cells. Her team employs high-speed, parallel processing to embed custom software into microprocessors and cameras to be built into eyeglasses. The camera images are translated into code (in the form of flashing lights) that can be transmitted by still healthy ganglion cells and understood by the brain.

But with hope comes concern. That's Smolan's take on J. Craig Venter’s Synthetic Genomics in La Jolla, Calif., which also relies on massive computer processing, to create genetic sequencing for new types of bacteria, algae, and plants to assist in industry and replace fossil fuels. “He is patenting new forms of life,” says Smolan. “While these new forms are being designed for human good, it does make you think of unintended consequences, like Frankenstein.”

 

Photo credits: Baby: ©Catherine Balet “Strangers in the light” (Steidl) 2012 / from The Human Face of Big Data; Artificial Retina: ©Joe McNally 2012 / from The Human Face of Big Data

A 15-Year-Old Polymer May Challenge Indium Tin Oxide In OLED Displays

OLED TVs, hitting store shelves in giant screen sizes, are likely to draw more consumer eyes than wallets in 2013, due to their cost of about US $10 000 for a 55-inch diagonal unit. That's a consumer problem. Manufacturers have another concern: potential shortages. Like LCD TVs, they rely on indium, a material in high demand for solar panels as well as displays. In OLEDs, indium is part of the indium tin oxide (ITO) used as the anode. The oxide is both conductive and transparent, making it perfect for displays.

But because of indium’s growing cost and limited availability—and because it is brittle, and therefore not suitable for future flexible displays—researchers have for years sought a viable substitute. Scientists at the U.S. Department of Energy’s Ames Laboratory and Iowa State University's Microelectronics Research Center recently announced that they may have found that particular holy grail—in the 15-year-old polymer poly (3,4-ethylene dioxythiophene):poly (styrene sulfonate), aka PEDOT:PSS. Previously, PEDOT:PSS wasn’t conductive enough or transparent enough to work in displays; it’s more commonly been used for coating photographic films to prevent static discharge. But Ames senior scientists Joseph Shinar and Ruth Shinar, and research scientist Min Cai, and their colleagues have developed a multilayer PEDOT:PSS fabrication technique that, with other manufacturing tricks, improved both conductivity and transparency. The research was published in the journal Advanced Materials.

University of Cincinnati associate professor Jason Heikenfeld, considering the researchers results, said that Ames’ version of PEDOT:PSS is not superior to the best ITO in performance as a basic transparent conductor. But, he indicated, the results are good for PEDOT:PSS. Joseph Shinar agrees, but notes that in spite of the lower transparency and lower conductivity of the multilayer PEDOT:PSS compared with indium tin oxide, the PEDOT:PSS-based devices are more efficient than the indium-tin-oxide-based devices because of a favorable micro optical cavity effect.

PEDOT:PSS isn’t the only technology taking aim at indium tin oxide’s dominance of OLED electrodes. Another leading competitor, Heikenfeld says, has been silver nanowire ink, being commercialized as ClearOhm. Heikenfeld pointed out that ClearOhm has shown amazing transparency, but the semiconductor community has a deep understanding of how to work with PEDOT:PSS, and PEDOT:PSS  has some unique current-injection properties that are desirable for OLEDs . But he’s not convinced that anything yet appears to be the ultimate ITO alternative. 

Paul O’Donovan, principal research analyst in Gartner’s Semiconductor Research Group, is more excited. “This is a significant advance in the commercial development of OLEDs,” he told Spectrum. “Not only does PEDOT:PSS offer an over 40 percent efficient improvement over indium tin oxide, but it’s flexible, opening up the real proposition of a range of flexible displays for pocket-sized to projection-screen-sized displays. And the added bonus is that PEDOT:PSS is lower cost and not in short supply. This development will help with reducing the manufacturing costs of large OLED TV panels, which is good news for consumers.”

Image: PEDOT:PSS. Source: Wikimedia Commons

New Northern Lights Camera Finds 'Ripples' in the Sky

NORUSCA II is the mother of all fisheye whole-sky cameras. The device can rapidly tune through the spectrum from 430 nm to 750 nm, using a 3.5 mm focal length, f/1.1 lens with a 180-degree field of view, to capture images on a high-resolution Princeton Instruments electron multiplying charge coupled device (EMCCD) camera.

Fred Sigernes of the University Centre on Svalbard (with collaborators from the Ukraine National Academy of Sciences, the Murmansk Region Polar Geophysical Institute, and Calgary’s Keo Scientific) built NORUSCA II to study auroras at the Kjell Henriksen Observatory. The observatory is located on the Norwegian island of Spitsbergen, at 78º N (where the sun sets for the winter before Halloween and doesn’t appear again until Valentine’s Day). The results are described in a new, open-access Optics Express paper.

NORUSCA II all-sky camera, with 12 optical elements, rather than the 19 of a standard auroral telescope: 1) focusing and collimation, 2) filter box, 3) camera lens, 4) camera body.

The camera’s hyperspectral heart is a filter box—an array of electronically controlled wave plates. These tunable Lyot filters—a series of successively thinner liquid crystal screens whose optical properties vary with the voltage across them—can switch from one wavelength to another in about one second.

Sigernes set NORUSCA II up under a clear observatory dome at the Spitsbergen observatory on 7 November 2011, to watch the sky during the nearly four-month-long night.

Over that period, the camera captured one full spectral scan a minute--cycling through 15 wavelengths (set to 13 key nitrogen, oxygen, hydrogen, and sodium transitions plus two background control settings) with time for a half-minute nap in between cycles--to produce stunning time-lapse videos of the Northern Lights...and the northern sky, generally.

Back on 24 January 2012, NORUSCA II caught the earthly aftereffects of a storm on the Sun, a strong coronal mass ejection of particles blasted out of an M8.7-force solar flare. While the flare was not big enough to make an historic Top-Ten list, it was impressive enough to generate NASA press releases and prompt space-weather storm warnings from IEEE Spectrum, among others..

The result was a mesmerizing video of the aurora borealis undulating in the solar wind—followed by a phenomenon that had never been documented before: glowing ripples running across the heavens, for all the world as though someone had dropped a pebble into the pond of the sky. (The red arrow in the photo at left indicates the ripples; the blue arrow shows faint emissions from the Milky Way.) Sigernes says he is still analyzing the ripple spectrum (which also appeared on the observatory’s near-infrared hydroxide imager), so we will have to wait to see what all of the (electron) excitement was about.

Images: Optics Express / Fred Sigernes

Stanford Names Seven Heroes of Engineering—Who Are Yours?

Since 2010, Stanford University has annually selected engineering heroes, engineers with some Stanford affiliation who have “advanced the course of human, social, and economic progress through engineering.”

This year’s class of engineering heroes numbers seven. (The first class included HP founders William Hewlett and David Packard.) They are:

—Computer scientist and entrepreneur James H. Clark: While an associate professor, Clark developed the Geometry Engine, a processor optimized for rendering computer graphics, and used that technology to kick off his first startup, Silicon Graphics. He went on to found Netscape with Marc Andreessen. And for years he a lot of time commissioning and sailing extremely high tech yachts, though he's gotten over that interest and this year put the yachts on the market.

—Yahoo Founders David Filo and Jerry Yang. The two hold masters degrees in electrical engineering from Stanford, and jointly founded Yahoo in 1995 to commercialize what had previously been called “Jerry and Dave’s guide to the World Wide Web.”

—Public key cryptography inventor Martin Hellman. Along with Whitfield Diffie and Ralph Merkle, Hellman developed public key cryptography. He received his M.S. and Ph.D. degrees from Stanford, and served on the faculty for 25 years. While at Stanford he was honored several times for his efforts to overcome ethnic tension at the university. Lately IEEE Fellow Hellman has been concerned about the threat of nuclear weapons.

—AI pioneer John McCarthy. Stanford computer science professor John McCarthy, who died late last year, coined the term “artificial intelligence,” developed the LISP language, and invented computer time-sharing.

—Former U.S. Secretary of Defense William J. Perry has B.S. and M.S. degrees from Stanford in mathematics, and is currently a professor emeritus in the school's Department of Management Science and Engineering. He served as U.S. secretary of defense from 1994 to 1997, and undersecretary of defense for research and engineering in the 1970s.

—“Father of Earthquake Engineering John Blume. Blume was a consulting professor of civil and environmental engineering at Stanford and a Stanford alum. His advances in seismic engineering contributed to the design of the Stanford Linear Accelerator and the California State Capitol; he also consulted for the U.S. Nuclear Regulatory Commission and on 70 nuclear plant projects. Blume died in 2002.

Who are your engineering heroes? (Your choices do not, of course, have to have a Stanford connection.) Tell us in the comments below.

Black Swan Cyclones Could Be Much Worse Than Sandy

One of the scariest aspects of Hurricane Sandy was that it wasn't really a freak event. The storm surges produced in Lower Manhattan actually were predictable based on historical data and climate modeling, so you could argue that maybe we should have been a bit better prepared for this type of storm.

But what about storms that are truly unprecedented?

This week, researchers discussed their work on so-called "black swan" cyclones at the American Geophysical Union meeting in San Francisco. Black swans, said Ning Lin, a professor of civil and environmental engineering at Princeton, are storms that cannot be predicted based on historical data, and they can have dire consequences. Such storms actually have "retrospective predictability," which means that we can only explain why they happened, after they happened.

You might wonder how it's even possible to study such storms. Lin, whose study at AGU this week was co-authored by hurricane research eminence Kerry Emanuel of MIT, said that predicting these anomalous storms just requires a different approach. While the bulk of tropical cyclone risk assessments are based on historical data, she relied on synthetic models instead. "We can generate large numbers of synthetic storms, and physically possible storms for different climate conditions, and then carry out storm surge simulations," she said during a press conference. So far, they have only modeled a few specific locations, largely because the computational requirements for each run are huge.

In the U.S., the team chose to study Tampa, Florida. Lin said the highest recorded storm surge in Tampa was due to a storm in the 19th century, when waters rose 4.6 meters. But that doesn't mean a new storm couldn't go higher. "Ten meters is possible," she said. She added that the record 3.5 meter storm surge Sandy sent to New York was several meters off from what a black swan would produce. "In terms of storm surge, Sandy was not a black swan." Lin's team also analyzed the possibility for black swan storms that could inundate Darwin, Australia, and, amazingly enough, parts of the Persian Gulf like Doha, Qatar (where, incidentally, somewhat punchless climate talks are ongoing this week).

One of the lessons that emerged out of Sandy's receding waters was the idea that this type of storm is likely to happen again as seas rise and weather patterns shift thanks to climate change. In that context, black swans are still a fringe concern: Lin said the 10-meter storm surge in Tampa, or a 5 or 6 meter surge in New York, are probably around a one-in-ten-thousand event. But the fact that Sandy and its spiraling-into-the-billions cost doesn't even qualify for black swan status is terrifying: there are super-rare superstorms out there that could double Sandy's devastation, but Sandy-like storms are probably on their way again before long.

Image via Maryland National Guard

Finding Mercury’s Water with Infrared Lasers and Neutron Counters

The satellite giveth, and the satellite taketh away. While some space-borne probes seem to confirm that earthly ice caps are shrinking, others indicate that billion-year-old ice deposits lurk in Mercury’s deep arctic shadows—which remain at a cool -173 °C despite the planet’s proximity to the sun, which can drive summer noon temperatures up to 627 °C.

Are Mercurians swiping Earth’s ice to chill their martinis? Hardly. For one thing, it’s too hard to get good vermouth up there. More important, it appears that the ice (and possibly some organic matter, which may serve as insulation, like sawdust in an old icehouse) was delivered by accommodating comets.

Two instrument packages on the Mercury Surface, Space Environment, Geochemistry and Ranging spacecraft (Messenger) provide complementary evidence confirming that the deep eternal shadows of the planet’s steep-walled polar craters harbor massive plates of ice.

Two Science papers--describing Messenger’s Mercury Laser Altimeter (MLA) and Neutron Spectrometer (NS) experiments—provide the up-close confirmation that, yes, those really are caves of ice in that sunniest of pleasure domes.

The MLA, built at NASA’s Goddard Space Flight Center, lights up the Mercurial surface with a 1064-nanometer (deep infrared) chromium/neodymium yttrium-aluminum garnet laser. Like orbiting laser altimeters everywhere, it flashes light off the ground below (in this case, lighting up spots about 50 meters in diameter at 400-meter intervals) and gauges distance by measuring the time until the reflected light returns. The MLA gathers additional information about the surface by alternating high- and low-power flashes. By comparing the strengths of the returning signals, it can provide more accurate estimates of the reflectance of the ground below, possibly revealing something about the surface's composition.

The result is a topographic map of Mercury based on more than 4 million individual elevation measurements—half of them including information on the nature of the ground.

The Messenger researchers were already primed to look for ice. Earth-based radar probes had previously indicated that the craters might hide deposits of frozen water—or some other captured volatile substance such as sulfur—in radar-bright areas hidden deep in the perpetual shadows of arctic craters.

The scientists found that these radar-bright areas fall into two categories. Those at the highest latitudes—where the crater-wall shadows are longest and the valleys coldest—reflect both radar and 1064-nm laser light strongly. These, the team says, are consistent with large areas of exposed ice, in layers at least several meters thick.

In slightly lower latitudes, the researchers found that the radar bright areas are often enveloped or overlaid by larger laser-altimeter-dark regions (areas that reflect little infrared). Indeed, “all craters with [radar bright] deposits and sufficient altimeter sampling show at least some [laser-altimeter dark] features in their poleward facing portions.” This suggests, say the researchers, that in these warmer craters, a thin layer of something—regolith or even comet-deposited organic compounds—may shield deeper strata of ice, protecting them from sublimation.

The altimeter findings are supported by the Johns Hopkins–built Neutron Spectrometer, which detects neutrons thrown off of atoms on Mercury’s surface as the atoms are struck by gamma rays. The emitted neutrons fall into three energy ranges: fast (energy greater than 0.5 megaelectron volt), epithermal (0.5 electron volt to 0.5 MeV), and thermal (less than 0.5 eV).  Because hydrogen atoms and neutrons have such similar masses, they transfer momentum very efficiently, and the hydrogen absorbs momentum from the epithermal neutrons.

The neutron spectrometer counts the number of incoming fast, epithermal, and thermal neutrons (correcting for changes in altitude) and extrapolates the data to infer the amount of hydrogen in the surface materials. The relative proportion of epithermal neutrons among the outbound particles reveals the amount of hydrogen present. A drop of even 4 percent in the rate of incoming fast and epithermal neutrons indicates that the spacecraft is passing over hydrogen rich water.  

The neutron data suggest that only half of the radar-bright regions are actually water at the surface. At the same time, the overall picture shows as much as 1000 cubic kilometers of water icebound at Mercury's poles, lying in layers “tens of centimeters thick.” Many of these layers are insulated below a superficial covering (much poorer in hydrogen compounds) 10 to 20 centimeters deep.

Image: NASA/Johns Hopkins University Applied Physics Laboratory/Carnegie Institution of Washington/National Astronomy and Ionosphere Center, Arecibo Observatory. 

Check Your Vitals on Your Smartphone

Walter De Brouwer wants to get inside your medicine cabinet. According to the founder and CEO of the medical startup Scanadu, there’s a big market opportunity in there.

“When you look at the numbers, the only medical tool for at-home use that really sells is the thermometer,” he told me in a recent phone interview. So Scanadu has designed what is essentially a super-smart thermometer. The Scanadu Scout, expected to hit market in late 2013 and sell for less than US $150, allows a user to quickly measure six physiological parameters—temperature, respiratory rate, blood oxygenation, blood pressure, heart rate, and electrical heart activity—and send all that info to a smartphone via Bluetooth.

“These parameters are called the vital signs, and they test them in every emergency room in the world,” explains De Brouwer. “You have a complete emergency room in your smart phone.” The Scout takes all these readings with one small sensor-studded device that the user holds against his temple for about 10 seconds. 

Scanadu announced the Scout on Thursday, and says it's now ready for manufacturing. The Scout is the latest device to seek customers in the “quantified self” movement: the growing tribe of fitness enthusiasts, dieters, and data geeks who use clip-on devices to track things like exercise and sleep. All these quantified self tools send info to a website or a smartphone, where the consumer can analyze the data over time and measure progress towards his or her goals.

De Brouwer says the Scout will be used in a similar way, and says the data produced will be “meaningful and actionable.” It can change consumers’ conversations with their doctors, he says. “With our device, it takes 10 seconds to diagnose yourself, and you have all these parameters,” he told me. “After a couple of months, you have all this data about yourself, and you can do analytics. Then you can take the data to your doctor and say, ‘This is my average blood pressure. But here I can see it going wrong, because I started taking this new medicine, or I started sleeping less.’”  

His use of the word “diagnose” there was a bit of a slip-up, because officially, the Scout is not a medical device, and it’s not meant to be used for diagnosis. That would require a type of FDA approval that Scanadu wants nothing to do with. So later in our conversation, De Brouwer stressed that the Scout isn’t actually a medical device; it’s an educational device. “We do not provide a diagnosis, we are just opening up that diagnostic space to consumers, so they can try it out themselves,” he says. In other words, the Scout won’t tell you if there’s something wrong with you, and it won’t suggest any remedies. It will simply tell you that your temperature is 103 degrees and your respiratory rate is high, and leave it to you to figure out if such readings are a sign of trouble.

The Scout is the company’s first step. Scanadu also plans to sell several add-ons for fluid analysis: One disposable cartridge will allow for urine analysis; another will test saliva for the flu virus. The company will also compete for the Qualcomm Tricorder X Prize, a $10 million award to whoever can build a device that resembles the Tricorder used in Star Trek’s medical clinic. To win the prize, the device must be able to measure key health metrics and diagnose a set of 15 diseases. De Brouwer thinks the Scout will be a good step in that direction. “We’re building this Tricorder from the bottom up,” he says.  

For more on Scanadu's vision of the future, check out their aspirational video below.

'Hyper-Ramsey' Method Refines Atomic Clock Research

The push to engineer ever-more-accurate clocks continues. A group at the Physikalisch-Technische Bundesanstalt (PTB, the German national standards institute) has developed a technique for more accurately measuring the frequencies of photons emitted by promising “clock transitions”—rare energy states durable enough to stick around for sustained probing by ham-fisted humans.

PTB researchers Nils Huntemann and Ekkehard Peik (with their collaborators at PTB and at the Institute of Laser Physics in Novosibirsk, Russia) are particularly interested in developing clocks based on ytterbium 171. The Yb ion boasts one particular transition (2S1/22F7/2), whose high-energy state has a natural lifetime of several years (desirable in a clock because random emissions pose less of a problem).  In experiments reported in Physics Review Letters (also available through arXiv) the researchers imprisoned a single ytterbium atom in an ion trap and probed it with a laser (picture a police suspect in an interrogation room being questioned under an impossibly bright light) to produce a precise experimental definition of the clock transition.

To do this, they refined a technique central to atomic clock technology: the separated oscillatory field (SOF) method, also called “Ramsey excitation" after physicist Norman Ramsey. (Ramsey pioneered the method in the 1950s and developed it into the robust and general metrological approach that won him a share of the 1989 Nobel Prize for Physics.)

The original Ramsey approach uses two spaced pulses of electromagnetic energy. The initial excitation pulse pumps up the system. An intentional pause between pulses allows the elements of the system to settle into “coherent superposition.” Finally, the probe pulse—at the same frequency as the excitation pulse, and precisely in phase with it—“reads” the system. If the captured ion fluoresces, it indicates that the energy was not locked in the clock transition (which doesn’t decay spontaneously), so it’s scored as a 0. If there is no fluorescence, however, then the target transition was excited at that wavelength, and it’s scored as a 1.

Since this is quantum world, the experiment is repeated hundreds of times; the 1s and 0s are added up and averaged, and the result is the excitation probability. The experimenters then vary the frequency by detuning the laser, and repeat the process for a new excitation frequency. Thus, step by incremental step, they build up a chart of excitement probability at each resonant wavelength. Over time, the accumulating data clearly show this allows them to precisely characterize the frequency of the target transition.

There’s a catch, though. Even at the right wavelength, it takes an intense jolt to drive an electron into a forbidden state—and walloping the ion that hard disturbs the whole system. This general excitement increases the apparent frequency of the resonant sweet spot. This change in frequency is called “light shift,” (Imagine a belfry-size bell. Rap it sharply with a wooden mallet, and you’ll be rewarded with a deep, mellow “bong.” Whack it too hard with a sledgehammer, though, and you get an unhappy “clang.” The whole frequency response has climbed up the register.)

Light shift is roughly proportional to the intensity of the stimulating laser pulse. In the past, researchers coped by repeating the experiment with a variety of laser intensities. They then plotted resonant frequency against beam intensity, and extrapolated the line back to zero-intensity to locate the “true” frequency of the target transition.

It’s a cumbersome approach, which requires significant data manipulation and introduces some errors of its own. The PTB researchers have tried to eliminate this backtracking with a method that automatically accounts for light shift and neutralizes it.

They call the approach “hyper-Ramsey spectroscopy” (HRS). This extension of the conventional Ramsey technique essentially “pre-pays” for the light shift, and then cancels it out in the instrument, eliminating the need for calculating light-shift corrections. HRS uses three (rather than two) sequential laser pulses. In conventional Ramsey excitation, the stimulating-frequency grid centers on the expected resonant frequency. In HRS, the central frequency is the expected resonant frequency plus the anticipated light shift.

In HRS, first pulse is the excitation pulse. After a pause comes a “noise-cancelling” pulse (as the researchers put it, the pulse “compensates the dephasing between the atomic coherence and the probe laser field caused by the Ramsey pulses”); it’s the same frequency as the first pulse, but its phase is inverted.  There is another, briefer, pause. And then comes the probe pulse, in phase with the initial excitation pulse.

The net effect is that researchers can measure the frequency of the target transition directly. They also reduced the light-shift by four orders of magnitude, pushing it down below the 10-17 level—results that, they say, have significant implications for the future of precision laser spectroscopy and time measurement.

Images: NPL, Teddington, Middlesex, UK. PTB, Braunschweig, Germany.

Medical Imaging Group Calls for Fewer CT Tests and More Research on Health Effects

I'm a CT refusenik. There hasn't been a whole lot of scientific support for my position, so a story in today's news—“Experts recommend closer scrutiny of radiation exposure from CT scans”—was welcome. Individual doctors, though, have had their suspicions all along, and they're the source of my resistance.

Back in 2005, as I was packing up my then-ten-year-old-and-small-for-her-age daughter to transfer from the pediatrician’s office to the emergency room with possible appendicitis, the pediatrician pulled me aside for a moment. “They’re going to want to do a CT,” she said. “Tell them no, that you want an ultrasound. They’ll say you’ll have to wait for an ultrasound, while you could get her a CT right away. Tell them you’ll wait. I’ve read some research on abdominal CTs in children and it concerns me.” Before that quick conversation, I had never given a second thought to the risks of CTs, but I followed the pediatrician’s advice, and the conversation in the emergency room went exactly as she had predicted. We waited, the ultrasound was conclusive, and my daughter’s appendix was removed a couple of hours later.

Four months later, I was back in the ER with my daughter after she’d gotten hit in the head with a ball. Again the ER doctor started writing up an order for a CT. Again I asked for an alternative. That doctor said that they could do an MRI if I “refused the CT” but they couldn’t sedate her for it and they typically sedate kids so they don’t freak out. He wrote “parent refuses CT on her chart.” They did the MRI; she didn’t freak out.

Just last year, my 20-year-old son was sent from the doctor’s office to the ER with suspected appendicitis, and I texted instructions to him to refuse the CT and hold out for the ultrasound. (He did wait for an ultrasound and the appendix turned out to be fine.)

I hope I don’t ever have to talk to an ER doctor about a CT again, but I’m not counting on it. So today’s story from the University of California at Davis is one I’m printing out and carrying in my wallet.

The gist of it:

- We don’t fully understand the biologic effects of medical imaging.

- The number of CT tests conducted in the U.S. is going up 10 percent a year.

- CT machines made by different manufacturers have vastly different control systems, increasing the chance of human error such as administering an incorrect dose of radiation.

- Kids and small adults can absorb two to three times the expected radiation because settings are standardized on "average-size" adults.

This analysis came out of the Radiation Dose Summit, a meeting of more than 100 medical physicists, radiologists, cardiologists, engineers, industry representatives and patient advocates in 2001. The findings were published in the November 2012 issue of Radiology, with UC Davis professor John Boone as lead author. Boone could have been talking to me when he said (as reported in the story), "In reaction to media coverage of radiation overexposure cases, some patients refuse to undergo medical imaging procedures. Yet for almost all patients, the risks of foregoing a needed medical procedure far outweigh any potential radiation-associated risks."

But Boone and his coauthors pointed out that what those radiation-related risks are is not exactly clear, and the risk/benefit tradeoff of using CT imaging needs to be better understood.

The summit also recommended some real changes that would both reduce the risks related to CTs and reassure patients. These include building CT machines more like cars—that is, standardizing controls so they are all “driven” the same way; and reducing “wasteful imaging,” that is, tests that have little impact on a diagnosis or outcome.

Meanwhile, Boone said, the University of California is funding efforts at all five of its medical schools to develop more accurate measures of radiation exposure from medical imaging.

Brain Model Connects Neurons to Behavior

Researchers have competed fiercely for years to build computational models of the brain with an ever-larger number of simulated neurons, but scientists from the University of Waterloo have taken a different approach: they have built a model to explain how brain activity generates complex behavior.

In a paper published today in Science, Chris Eliasmith and his colleagues describe "Spaun," a 2.5-million-neuron model of the brain they hope will help bridge the brain-behavior gap. Spaun, short for Semantic Pointer Architecture Unified Network, can recognize numbers, generate answers to simple numerical questions, and write them down using a physically modeled arm.  

In their experiments the researchers presented Spaun with images of handwritten or typed characters, mostly numbers. Spaun processes them in various ways, and writes responses. For example, if Spaun is presented the number six, it can identify it and recreate it. 

The incoming visual image is first compressed to extract the essential visual elements. Spaun's working memory system is composed of a high-dimensional neural integrator, taken from computational neuroscience, and convolution memories taken from mathematical psychology. The neural integrator allows Spaun to store information and the convolution memories provide a memory-efficient algorithm that allows Spaun to bind newly arriving information with a representation indicating its syntactic role. The computations to turn that into arm movements are based on optimal control theory.

Previous brain model projects, such as the Blue Brain Project (1 million neurons), IBM's SyNAPSE Cognitive Computing Project (1 billion neurons, or a bit larger than a cat brain), and a human-scale simulation of 100 billion neurons have been reported. These projects are impressive in scale, but simulating large numbers of neurons alone doesn't explain how complex brain activity generates complex behavior. Eliasmith's 2.5-million spiking neuron model attempts to address the brain-behavior gap, a central challenge in neuroscience.

Spaun can peform eight different tasks involving recognizing numbers and performing motor responses, but the model is hard-wired and cannot learn new tasks. The learning issues are a principle shortcoming, but perhaps a wise one to sidestep at this point, says Christian Machens at the Champalimaud Neuroscience Programme in Lisbon, in a related article also published today in Science. Machens says other than this shortcoming, Eliasmith's model provides a coherent theory on how the brain works. He writes: "To paraphrase the statistician George Box, their model is likely to be wrong, but is certainly useful."

Photo and video: Chris Eliasmith et. al.

Most Commented Posts

Tech Talk

IEEE Spectrum’s general technology blog, featuring news, analysis, and opinions about engineering, consumer electronics, and technology and society, from the editorial staff and freelance contributors.

Newsletter Sign Up

Sign up for the Tech Alert newsletter and receive ground-breaking technology and science news from IEEE Spectrum every Thursday.

Advertisement
Advertisement
Advertisement
Load More