A senior US government advisory group has been saved from closure following a last-minute intervention from an agency within the Department of Energy (DOE). JASON — a group of often anonymous scientists that has advised the government on defence, security, and other issues for six decades – has been given a short-term contract by the DOE’s National Nuclear Security Agency (NNSA) after its work failed to be renewed by the Department of Defense last month. The group’s new contract will run until the end of next January allowing it to find and negotiate a fresh source of support.
JASON originated as a group of physicists funded by the defence department to spend the summer of 1960 studying scientific and technical issues arising from the struggle with the Soviet Union. Over the years, its membership has expanded to around 60, including many non-physicists too. The group has continued to spend summers advising on and suggesting remedies for problems relevant to government policy on military, intelligence, and national security issues.
The group’s recent problem-solving has gone beyond military- and nuclear-related matters. According to JASON’s chair — materials scientist Russell Hemley from George Washington University — it has, for example, advised the Department of Agriculture on using data related to crop production and the Census Bureau on its procedures. Indeed, in March the National Science Foundation contacted the group about a possible contract to examine concerns that overseas researchers funded by the foundation might present security risks.
[The move] appears to be part of a larger trend by federal agencies to limit independent scientific and technical advice
Steven Aftergood
Yet that future work was put at threat after the defence department announced on 28 March that it would discontinue its contract with the group at the end of that month. That decision left JASON without funding beyond the end of April – and desperately seeking alternative sponsors. Exactly why the defence department decided to cancel JASON’s contract remains unclear. The original agreement specified that JASON undertake an unlimited number of studies over the five years that ended on 31 March.
But in a statement on the cancellation, defence department spokesperson Heather Babb asserted that the department’s requirements for the group have changed. The department “will require only one study, rather than multiple studies,” she said. The cancellation, the statement continued, makes “the most economic sense for the department, and is in line with our efforts to gain full value from every taxpayer dollar spent on defence.”
New offer
Critics of the decision, however, take a sceptical view. “[The move] appears to be part of a larger trend by federal agencies to limit independent scientific and technical advice,” says Steven Aftergood, a senior research analyst at the Federation of American Scientists. He speculates that the group’s disagreement with government policies played a role in the decision, which he describes as “not good for the nation”.
Cancellation of the contract meant that JASON would lack the financial backing to carry out studies for other government departments. Those include the NNSA, which was considering agreements with the group on three issues relevant to the US nuclear stockpile. But when NNSA administrator Lisa Gordon-Hagerty examined the impact of cancellation on her agency, she decided to offer a temporary contract to give JASON time to find a new sponsor. The offer, which was made on 25 April, is similar to the defence department’s cancelled contract in all but length. It will start on 1 June and run for eight months and the JASON group has until 11 May to agree to it.
JASON had faced closure once before. In 2002 it refused an effort by the Defense Advanced Research Projects Agency, which oversaw its operations at the time, to decide on new group members. The affair was settled by a change in JASON’s administration. The non-profit MITRE Corporation took over its management, answering directly to the defence department’s undersecretary of research and engineering.
Following the release of the first image of a black hole from the Event Horizon Telescope (EHT) earlier this month, astronomers are now hoping the project can achieve another of its core aims: to produce a similar portrait of the environment near the black hole at the heart of our galaxy, dubbed Sagittarius A* (Sgr A*).
While we wait for those elusive images, some scientists – including several members of the EHT team – have been devising a new approach for capturing views of this enigmatic region. Their idea is to exploit a constellation of satellites to produce images that can reveal much more detail than can be recorded by the EHT’s current network of ground-based observatories.
A group led by Freek Roelofs of Radboud University in the Netherlands have now simulated the images that could be created of Sgr A* and its immediate surroundings by the proposed mission, called the “Event Horizon Imager”. While the EHT exploits a coalition of millimetre and sub-millimetre telescopes scattered across the planet, the space-based Event Horizon Imager would consist of multiple satellites – each one fitted with a dish capable of collecting sub-millimetre radiation – in orbit around the Earth.
Just like its pioneering ground-based forebear, the Event Horizon Imager would use interferometry to create detailed views of Sgr A*’s shadow and the plasma swirling near it. The technique would need two, or perhaps three, spacecraft – separated by as much as 25,000 km – to observe their target simultaneously. By carefully analysing and combining the data gathered by the individual telescopes, the collection of relatively small spacecraft could produce images with a similar resolution as could be achieved from one, vast, orbiting dish.
Since the Event Horizon Imager would be free from the distorting effects of atmospheric water vapour, it could make observations at frequencies as high as 690 GHz – much higher than the Event Horizon Telescope’s typical observing frequency of 230 GHz. “The angular resolution of a telescope increases linearly with observing frequency. Combined with the long baselines [the projected distances between the telescopes], this gives the Event Horizon Imager a resolution that is more than five times better than the attainable resolution from the ground at 230 GHz,” explains Roelofs.
Crucially, observing at such a high frequency also gets around the issue of interstellar scattering – one of the main hindrances to studying Sgr A* at lower frequencies. “Interstellar scattering is caused by material that resides in the interstellar medium [the space between the stars] between us and the Galactic Centre,” explains Roelofs. “The radio waves coming from near the black hole are distorted by electrons in the interstellar medium, causing the observed image to be blurred, as if we are looking through frosted glass. Above 500 GHz the effect is negligible.”
Having detailed views of the supermassive black hole right on our doorstep should offer astronomers important insights into some of the key ideas underpinning modern physics. “Sgr A* is particularly interesting for high-resolution imaging because its mass and distance are well known,” explains Roelofs. “By measuring the angular size of the black hole shadow, we can test general relativity and other theories of gravity to high precision.”
Not only that, but if the Event Horizon Imager were turned toward supermassive black holes with jets in other galaxies, researchers may be able to glean more clues about how those streams of material are ejected.
“Computer models and general relativity make predictions about how plasma flows around, into and away from the black-hole event horizon, and how this emission varies over time,” says Andrew Young, an expert in high-energy astrophysics at the University of Bristol, UK, who was not involved in the new study. “The finer details of these predictions vary depending on things like how fast the black hole is spinning, the configuration and strength of magnetic fields, and how the gas is flowing towards the black hole. But to see these details, and really put the theories to the test, requires much higher resolution than even the Event Horizon Telescope.”
Researchers have developed a hybrid predictive model to improve the identification of metastatic lymph nodes in individuals with head-and-neck cancer. Classifying nodes in PET/CT images, the model combines radiomics and deep learning and outperformed two existing techniques. By tailoring treatments more accurately to the individual, it has the potential to improve disease control and prevent over-treatment (Phys. Med. Biol. 10.1088/1361-6560/ab083a).
“Critical decisions often hinge on whether a patient has involved lymphadenopathy, such as treating with primary surgery or radiation, and whether to add chemotherapy to radiation treatment,” says David Sher, study co-author and radiation oncologist at University of Texas Southwestern Medical Center. “Radiation planning requires accurate diagnostics, since the optimum radiation dose to the neck is intimately related to the presence of lymph node metastases.”
Currently, clinicians typically assess images of lymph nodes by eye. Malignant nodes that are large and exhibit high PET tracer uptake are easy to spot, but smaller, less reactive ones are harder to pick out. “The delineation of questionable or suspicious lymph nodes strongly depends on the physicians’ clinical experience,” says first author, Liyuan Chen, also based at the University of Texas. “Machine-learning algorithms can potentially make clinical practice in malignant lymph node identification more accurate and consistent across physicians and facilities.”
The hybrid model classifies lymph nodes as normal, suspicious or involved. It is the first of its kind to be applied to the problem. In the current study, the group trained and tested the model using PET/CT images from 59 patients, participants in the separate INFIELD radiotherapy clinical trial at the Dallas centre. The images contained 236 lymph nodes (22 involved, 27 suspicious and 17 normal); 170 were used for training and 66 were used to test each of the two model components.
The researchers’ many-objective radiomics model extracts 257 handcrafted features — pre-defined metrics characterizing aspects including geometry, intensity and texture — from every image. It comprises a support vector machine to generate a predictive model, coupled with an overarching optimization algorithm. The algorithm uses six objective functions, two per type of lymph node, to test model parameters and subsets of features, identifying the combination of the two that classifies the nodes most accurately.
The deep-learning component is a 3D convolutional neural network (CNN). Here, the most optimal, abstract features for node classification are identified by the network automatically. Outputs from the CNN and radiomics models are then fused using evidential reasoning.
By fusing information from different imaging modalities and classifiers, the approach is, in principle, more stable than techniques using a single modality or classifier. “We expect that the hybrid model can perform more consistently when applied to different imaging data sets,” says senior author Jing Wang.
The hybrid model’s performance was compared against XmasNet, a recently developed CNN, and a conventional radiomics model using only one objective function. It was more accurate when analysing both PET and CT data (0.88 versus 0.81 for XmasNet and 0.75 for conventional radiomics). This was matched, however, by the performance of the CNN component alone. Slightly higher accuracy values were obtained with the hybrid approach when CT or PET images were analysed separately. The authors anticipate future studies will demonstrate the technique’s stability over non-hybrid approaches.
Prediction results obtained by the five different techniques. The hybrid model graph shows clear separation of the three lymph node groups, demonstrating accurate node classification. (Courtesy: Phys. Med. Biol. 10.1088/1361-6560/ab083a)
The group is currently working on further validations, as well as investigating more advanced neural network architectures for the model. These include an analysis of head-and-neck patients imaged prior to surgery, whose preliminary results are “very promising,” according to Wang. A clinical trial is also set to start this year.
The researchers envisage their technique could be integrated into the treatment planning process in radiotherapy clinics, with the clinician-contoured lymph nodes providing the input for the model. “Our model will generate malignancy probability for each lymph node, which will aid in physician’s decision whether to include the lymph node as part of clinical target volume,” Wang tells Physics World.
Magnetic robots could be used to kill, degrade and remove biofilms from a variety of dental and laboratory surfaces. The devices, developed by an interdisciplinary team of researchers at the University of Pennsylvania, may help in the fight against persistent antibiotic-resistant biofilm infections and biofouling of medical equipment.
Biofilms are made up of communities of bacterial cells encapsulated in a protective matrix of polymeric substances. They adhere to a variety of surfaces both within the body – where they can cause infections and other complications – or on non-biological structures, such as catheters and implants. Biofilms are complex and mechanically stable scaffolds that can be resistant to antibacterial treatments so removing them is an important challenge – especially as we approach a post-antibiotic era, says Hyun Koo of the School of Dental Medicine, who led this research effort together with Edward Steager of the School of Engineering and Applied Science.
The problem is that current antimicrobial techniques are proving to be largely ineffective for two reasons. The first is that they do not address the structural and biological properties of biofilms associated with drug resistance, say the researchers. The second is the fact that biofilms can rapidly re-establish themselves on a surface and retake a hold.
“Kill-degrade-and-remove” approach
The new “kill-degrade-and-remove” approach developed by Koo and Steager’s teams tackles both these issues thanks to catalytic antimicrobial robots (CARs). These biohybrid platforms are made from magnetic iron oxide nanoparticles suspended in a solution of hydrogen peroxide and enzymes and they generate free radicals on-site that break down the biofilm’s matrix and so kill the bacteria more effectively. Once destroyed, the bacterial biomass debris is “ploughed” away by applying an external magnetic field so the biofilm can’t regrow.
The researchers say they can sweep their CARs over surface areas as large as 150-mm2 or direct them over well-defined paths to locally remove bacteria with microscale precision while leaving surrounding domains intact. This might come in useful, for example, “to remove biofilms without damaging nearby host tissues or to sample biofilms at specific pathological sites, to identify microbial composition,” they explain.
Reaching hard-to-access areas
The Penn team also designed a second type of CAR consisting of a 3D moulded soft agar containing magnetic nanoparticles that could remove biofilms from more complex structures, like the curved walls of cylindrical tubes or from hard-to-access areas on teeth. Indeed, the devices are even able to target the most-difficult-to-reach parts of a tooth – the isthmus (a narrow channel between root canals where biofilms – dental plaque – often take hold). In these situations, the CARs “drill” though biofilm clogs while killing the bacteria within them at the same time.
“Existing treatments for biofilms are ineffective because they are incapable of simultaneously degrading the protective matrix, killing the embedded bacteria, and physically removing the biodegraded products,” says Koo. “These robots can do all three at once very effectively, leaving no trace of biofilm whatsoever.”
What is the ideal length of an oar used in a rowing race? You might think that the answer is simple and would be based on the size and strength of an individual rower. But unlike most other sports, where the size of equipment tends to scale with the athlete, oars used in specific rowing sports tend to be fairly uniform in size.
Now, Michael Benzaquen and colleagues at France’s Ecole Polytechnique have used physics to work-out what size oars are best for specific events and individual rowers. You can find out more in their preprint on arXiv, “Physics of rowing oars”.
The announcement of the first-ever image of a black hole by the Event Horizon Telescope (EHT) on 10 April created a lot of excitement in physics but it seems to have touched the LEGO community too. At least three black-hole-inspired LEGO designs have been submitted to the LEGO Ideas website, which lets fans share blueprints of their own models.
Luis Peña, who previously made LEGO models of the Hubble Space Telescope and the Mars Curiosity rover, has built a LEGO model of a single antenna belonging to the Atacama Large Millimeter/submillimeter Array, which played a crucial part in the observations. Peña’s design also features a small mosaic of 18×18 studs to illustrate the black hole image.
Accretion disc and jets
LEGO enthusiast “tm.bricks” has created a design consisting of all eight EHT telescopes together with a mosaic of bricks to signify the black hole while “douglasfx” has made a model of the M87* black hole featuring accretion disc and jets.
Peña’s design already has 340 supporters but it will need another 9660 before LEGO conducts a review of it. If any of the models reach the magic 10,000 votes and are also lucky enough to receive the green light by LEGO, then they go into the “development phase” where LEGO designers refine the product and develop it for commercial release.
Make sure you vote for your favourite.
In the random birthday file, Tuesday marked the 156th anniversary of the birth of Max Planck. Canada’s Perimeter Institute has celebrated the birth of the great quantum physicist by publishing a fact sheet about his remarkable life. Did you know, for example, that Planck climbed a 3000 m peak in the Alps at age 85? Or that his son Erwin was involved in the planning of the failed attempt to assassinate Adolf Hitler in 1944?
A soft dielectric material that could create a relatively high voltage when bent has been created by physicists in China. Qian Deng and colleagues at Xi’an Jiaotong University describe their material as the first-ever “flexoelectret”. It was made by embedding a charged polymer layer in the middle of a dielectric silicone rubber material. With some improvements, the new material could find a wide range of applications including wearable electronics.
When some materials are deformed non-uniformly, a strain gradient drives positive and negative ions apart to create a voltage across the material. Known as flexoelectricity, this effect is observed in many dielectric materials, including crystals, polymers, and semiconductors.
The effect is usually strongest in brittle ceramic materials, which are unsuitable for practical applications such as stretchable electronics. While far higher strain gradients can be achieved in softer dielectric materials, flexoelectric voltages generated in these materials are typically several orders of magnitude smaller than in ceramics.
Flexible electret
Deng and colleagues have now come up with a way of generating much higher voltages in a deformable dielectric material by adding a layer of permanent negative charges (called an electret) within the material. To test the idea, the physicists embedded a charged polymer layer in the central plane of a 10 cm-long bar of silicone rubber, forming a flexible electret – or flexoelectret.
In its undeformed state, the dielectric material is polarized equally and in opposite directions on either side of the charge layer – which means that there is no voltage between the top and bottom of the bar. However, if the bar is fixed at both ends and a force pushes down on its middle, the regions above and below the charged layer are deformed in different ways. The polarizations are no longer equal and opposite – leading to a voltage across the top and bottom of the bar. Deng and colleagues measured a “flexoelectric coefficient” that is 100 times greater than that of silicone rubber alone. The flexoelectric coefficient is a measure of the electric polarization that occurs when a material develops a strain gradient.
The team says that several improvements will be needed before the flexoelectret becomes a commercial possibility. Perhaps their most significant challenge will be to maintain the charge of the bar’s embedded layer, which tends to leak over time. However, Deng and colleagues hope that this and other challenges could be overcome within the next five years.
With these improvements, the physicists believe the flexoelectret will prove to be biocompatible, environmentally friendly, and suitable for a wide variety of commercial applications. In particular, it could lead to a new generation of flexible sensors, actuators, and energy harvesters that could be used in stretchable and wearable electronics.
“Atomically thin magnets were discovered barely two years ago and already form a vibrant field of research, which has shown a remarkable evolution in this short time,” says Patrick Maletinsky, principal investigator of the Quantum Sensing Lab at Basel University in Switzerland. Despite some groundbreaking results over the past few years, as Maletinsky points out, “Addressing nanoscale properties of these magnets remained an open challenge and one that only a few existing techniques could in principle address.” Among these techniques was “single-spin magnetometry”, a method he had developed for scanning probe experiments at Harvard University a few years previously. Alongside colleagues at the University of Basel, University of Geneva and École Polytechnique Fédérale de Lausanne (EPFL) in Switzerland, Maletinsky has now reported in Science experiments that establish the technique as “a unique tool to address nanoscale magnetism in van der Waals crystals, down to the limit of a single atomic layer”.
Key to the interest in 2D materials, as Maletinsky explains, is the ability to customize their properties by stacking layers of various 2D materials in different ways. For example, by twisting the angle between layers researchers have shown they can tune the properties of pristine bilayer graphene between opposing extremes from a Mott insulator to a superconductor, a discovery that Physics World awarded the Breakthrough of the year in 2018. When these stacked 2D materials are magnetic, Maletinsky describes the research as “experimentally uncharted territory”, rich with exciting proposals including predicted novel “spin textures” such as skyrmions.
The technological prospects are equally promising. “These are the thinnest possible magnets,” says Maletinsky. “They thus offer opportunities for miniaturization of devices, which no other approach or material can. In addition, the tunability yields new functionalities such as fully electrical manipulation and readout of magnetization, which is relevant for data storage.”
Researchers have used various approaches to study the magnetism in 2D materials, including measuring how the field from 2D materials affects the electronic transport properties, fluorescence and the magneto-optical Kerr effect. The advantage single-spin magnetometry offers over these approaches is that it can provide direct quantitative data with nanoscale resolution. “I would now consider single-spin magnetometry as a close-to-ideal tool to study the physics of atomically thin 2D magnets,” says Maletinsky.
Decorating diamonds
To measure the magnetism in their 2D samples Maletinsky and colleagues measured the spin levels of a nitrogen-vacancy (NV) defect in diamond, where a nitrogen atom substitutes a carbon atom in the crystal lattice next to a vacant crystal lattice site. The optical responses of NV defects, such as photoluminescence, have distinctive peaks in their spectra that result from the energy levels of the electron spin. Applying a magnetic field shifts the spin energy levels, a response described as the Zeeman effect. In the range of magnetic field values produced by samples of 2D magnets this shift is proportional to the field, providing a convenient, quantitative magnetometer.
“The technique we employ – DC [direct current] NV magnetometry – is conceptually rather straightforward and mastered by many labs worldwide,” says Maletinsky. “The real difficulty comes about when one wants to perform nanoscale magnetic imaging using scanning NV centres and this approach is required to achieve the results we report on in our paper.”
Our solution for this are advanced all-diamond scanning probes
He highlights how important it was to have access to high quality samples, and his good fortune to collaborate with the group of Alberto Morpurgo at the University of Geneva who had mastered the non-trivial task of preparing excellent samples in a robust and controlled way. In addition, he describes two difficulties nanoscale studies raise: combining scanning probe microscopy (SPM) with an approach that allows efficient measurement of optical responses of the NV spins; and a robust method to add an NV centre to the tip of the SPM probe.
“I would consider this last part as the key crux for our experiments,” he adds. “Our solution for this are advanced all-diamond scanning probes, which we fabricate with my group in Basel and whose first generation I originally developed while still a postdoc in the lab of Amir Yacoby in Harvard.” The result is a technique that can quantitatively determine key magnetic properties of 2D crystals and directly image magnetic domains with spatial resolutions of just a few tens of nanometres.
Lucky break
Maletinsky says that a back of the envelope calculation was enough to suggest that their approach should work. So although experiments do not always go to plan the successful imaging was not a total surprise. However, the study was not without its white-knuckle moments, and one in particular proved to be a blessing in disguise.
Lead researchers. Patrick Maletinsky (left) and Lucas Thiel (right), two of the researchers behind the demonstrated nanoscale NV magnetometry of 2D crystals. Credit: Quantum Sensing Group, University of Basel
“About halfway through our planned measurement campaign, my student Lucas Thiel, who is the main author of the paper, came to my office all depressed to report to me that he had destroyed one of our precious samples by an operator error in our setup (SPMs are not very forgiving in this regard) and that none of the data he took after this crash made any sense to him anymore,” says Maletinsky. “This was especially problematic since we had already acquired a rather extensive dataset on the sample in question and its loss would have meant a setback of several weeks.”
It took us several weeks to convince ourselves that this accident had indeed been a very fortunate event
After initially mourning the accident, Maletinsky says they went back and analysed the post-crash data in more detail and found that what had initially made no sense, in fact explained an effect that had long puzzled researchers studying CrI3. In the bulk CrI3 is ferromagnetic, that is, all the spins align to give strong magnetic properties that are stronger for larger samples. In contrast studies of samples of few-layer 2D CrI3 all have either zero magnetism or the same magnetism as a monolayer, suggesting that the spins of adjacent layers are antiparallel. However, the magnetism of the punctured sample was roughly nine times the magnetism of a monolayer. Crucially the punctured sample was also nine layers thick, prompting the researchers to compare Raman measurements of the damaged sample with a pristine flake to determine whether there were differences in the crystalline structure.
“Even though the story and explanation might appear obvious in hindsight, it took us several weeks with additional control experiments and extensive discussions with our collaborators from Geneva to convince ourselves that this accident had indeed been a very fortunate event,” says Maletinsky. “In fact, it had offered us an explanation for one of the bigger questions in the field, namely why CrI3 showed the puzzling “even-odd effect” that we could now clearly link to a structural distortion in the material and one that we could “undo” using our puncture.”
NV-magnetometry for all
So will this become the go-to tool for 2D magnet experiments worldwide? Unfortunately the availability of the all-diamond NV decorated tips Maletinsky and collaborators used remains a key obstacle as so few groups worldwide fabricate and employ them: Maletinsky’s group at Basel University, Yacoby’s group at Harvard and the group at University of California at Santa Barbara run by Ania Bleszinsky Jayich. The group run by Christian Degen at ETH Zürich also designs, builds and operates scanning probe microscopes that incorporate NV centers at the tip apex.
“Given my strong belief that this technique could be broadly applicable and useful to various areas of science, we recently spun out the startup company Qnami to overcome this shortcoming,” Maletinsky tells Physics World. “Qnami’s mission is thereby to provide commercial solutions for scanning NV magnetometry and in particular our hard-to-fabricate diamond tips.” Qnami already sells tips to research groups including those run by Jörg Wrachtrup at Stuttgart University in Germany and Vincent Jacques at Montpellier University in France and will reach out to first beta testers for a complete magnetometer-based quantum microscope later this year.
This article was updated 2 May 2019 to include the Degen Lab among those using diamond NV decorated tips.
Members of the Hassanpour Lab envision a world where, in less than 30 seconds, a pathologist will know the diagnoses and prognoses for one of the deadliest cancers in the western world.
In collaboration with pathologists from Dartmouth-Hitchcock Medical Center (DHMC), Saeed Hassanpour’s research team has developed and validated a deep-learning model that quickly assesses histopathology slides from lobectomy procedures, during which a lobe of the lung is surgically removed. Their tool promises to reduce pathologists’ workload and the subjectivity of visual analysis of heterogeneous histologic patterns. The findings were published in Nature Scientific Reports (10.1038/s41598-019-40041-7), and their code is publicly released on GitHub.
Deep learning for diagnosis and prognosis
Lung adenocarcinoma is the most common type of lung cancer and one of the most lethal. Pathologists classify and grade histopathology slides from lobectomy procedures under a microscope according to their histologic patterns, or subtypes. Accurate interpretation of predominant and subtle minor patterns is important because these can determine a patient’s diagnosis, prognosis and health outcomes.
Determining at what point lung adenocarcinoma becomes invasive is a key challenge for pathologists’ visual analyses of histopathology slides.
“Some [histologic] patterns are more aggressive,” explains Laura Tafe, one of three pathologists involved with the study. “Even a small component of micropapillary pattern, as little as 5–10%, is important to identify because these patients are at increased risk for having disease spread to the regional lymph nodes.”
Intermingling invasive and non-invasive patterns, as well as inflammation or scarring, can also hinder accurate quantification of patterns and contribute to variation in pathologists’ analyses.
Deep learning, a subset of machine learning, “is not currently in any position to replace pathologists or radiologists,” says Hassanpour. Still, deep-learning neural networks may be able to assist pathologists by automating time-intensive, difficult tasks. Last year, Hassanpour’s team set about using such a deep-learning neural network to assist pathologists’ analyses of lobectomy slides.
Collect, scan, train, validate
The researchers collected and scanned 422 histopathology slides procured from lobectomy procedures at DHMC. Three pathologists digitally annotated five histologic patterns and benign tissue on 245 whole-slide images.
The researchers trained a convolutional neural network to identify major and minor histologic patterns on these 245 slides. To improve the model’s performance, they broke up pathologists’ annotations into square, fixed-size patches, oversampled patches from less prevalent patterns, neutralized colour differences between slides, and performed image augmentation to generate a larger number of samples.
After training, the researchers evaluated the model against pathologists’ performance using 143 whole-slide images. Results suggested that the deep-learning neural network performs on par with the pathologists, while both pathologists and the model found detection of minor patterns challenging.
Toward clinical implementation
Hassanpour and Tafe see great promise in their deep-learning model for pathologists and patients alike. For example, it could pre-screen lung adenocarcinoma cases, aiding in pathologists’ visual inspection and interpretation by identifying aggressive minor histologic patterns. The team also plans to conduct a clinical trial incorporating education and trust-building to determine whether the deep-learning model improves outcomes in patients.
“Collaborations such as this between pathologists and data scientists are essential to advancing the field of oncology. Our expertise is disparate, which perhaps is exactly why it can be so powerfully complementary,” Tafe says.
One of the world’s largest and most sensitive cosmic-ray facilities has begun operation with its first set of detectors. Located about 4410 m above sea level in the Haizi Mountain in Sichuan Province in southwest China, the 1.2 billion yuan ($180m) Large High Altitude Air Shower Observatory (LHAASO) will attempt to understand the origins of high-energy cosmic rays.
Cosmic rays are particles that originate in outer space and are accelerated to energies higher than those that can be achieved in even the largest man-made particle accelerators. Composed mainly of high-energy protons and atomic nuclei, cosmic rays create an air shower of particles such as photons and muons when they hit the atmosphere. Where cosmic rays come from, however, has remained a mystery since they were first spotted some 100 years ago.
LHAASO aims to detect cosmic rays over a range of energies from 1011-1018 eV. The first phase of the project, which is now complete, involves 900 Cherenkov water detectors that are each equipped with a 20cm- and 4 cm-diameter photomultiplier tube. The 900 units are located in a single “pool” containing 100 000 tonnes of water that is 4.5 m deep and has an area of 22 500 m2.
Being above ground, the detectors will spot the Cherenkov radiation emitted when a charged particle travels through a medium faster than light can travel through that medium. The first phase also includes two wide-field Cherenkov telescopes as well as 80 muon detectors and 180 scintillation detectors, which will study electrons and photons in the air showers.
With the first phase complete, the LHAASO will now begin studying cosmic rays as well as identifying the high-energy radiation produced by gamma-ray bursts and detect and observe extragalactic blazars. Astrophysicist Zhen Cao from the Institute of High Energy Physics (IHEP), Chinese Academy of Sciences, who is LHAASO’s chief scientist, expects that LHASSO will make breakthroughs in elucidating the origins of cosmic rays.
International collaboration
LHAASO’s water Cherenkov detector array will rival a similar experiment in Mexico — the High Altitude Water Cherenkov Experiment (HAWC), which is built mainly by the US and Mexico. HAWC detects Cherenkov radiation using 190 000 litres of extra-pure water that is contained in 300 circular water tanks each 5 m high and 7.3 m in diameter.
“LHASSO will be one of the most sensitive gamma-ray and cosmic-ray detectors in the world, not only because of its large size, but also because it uses different particle detection techniques” physicist Ibrahim Daniel Torres Aguilar, who is HAWC’s site manager, told Physics World. “Once complete it will have a better sensitivity than the HAWC observatory, however, because both observatories have a large field of view and they are on opposite longitudes, it opens the possibility that the two observatories can work together to constantly map the cosmos”.
LHAASO is an international collaboration that includes scientists from China, France, Russia, Switzerland and Thailand. First mooted in 2008, the facility won approval from the National Development and Reform Commission of China in December 2015. Construction of LHAASO began in 2016 and it is set to be fully operational in 2021. When complete it will consist of a water Cherenkov detector array covering a total area of 78 000 m2, 12 wide-field Cherenkov telescopes distributed over 1.3 km2 as well as 5195 scintillation detectors and 1171 muon detectors.
IHEP director Yifang Wang says that astronomical observations have now “entered the era of combining multi-wavelength observations and multi-messenger studies,” adding that LHAASO will enhance worldwide astronomical observations.
Global warming has increased global economic inequality. Some countries have profited from climate change while the same rise in average planetary temperatures has dragged down economic growth in the warmer countries.
The gap between those groups of nations with the highest and lowest economic output per person is now around 25% larger than it would have been had there been no climate change.
“Our results show that most of the poorest countries on Earth are considerably poorer than they would have been without global warming,” said Noah Diffenbaugh, a climate scientist at Stanford University in California. “At the same time the majority of rich countries are richer than they would have been.”
They found that during warmer than average years growth was accelerated in those nations with normally cool climates – such as Norway and Sweden – but was slowed significantly in those countries with tropical or subtropical climates such as India or Nigeria.
And between 1961 and 2010, they found that global warming depressed the wealth per person in the poorest nations by between 17% and 30%.
“The historical data clearly show that crops are more productive, people are healthier and we are more productive at work when temperatures are neither too hot nor too cold,” said Burke. “This means that in cold countries, a little bit of warming can help. The opposite is true in places that are already hot.”
The two scientists put the message of climate injustice bluntly in their paper: “Our results show that, in addition to not sharing equally in the direct benefits of fossil fuel use, many poor countries have been significantly harmed by the warming arising from wealthy countries’ energy consumption.”
What if … ?
All such research is tortured by uncertainties, and none greater than what historians call counter-factual comparison: that is, what would have happened if global average temperatures had not risen by around 1 °C in the last century.
To make their case, the researchers calculated 20,000 versions of what each separate country’s economic growth rate would have been without global warming, and based their estimates on the range of outcomes. So, they concede, there are uncertainties.
In effect, the latest research provides a kind of national climate audit. If greenhouse emissions are a measure of economic output, then the richest 10% produce atmospheric carbon dioxide almost as much as the bottom 90% together.
The Stanford study offers an estimate of the costs and benefits the richest and poorest have borne as a consequence of emissions. It also makes it clear that the poorer nations would benefit more from investment in renewable energy: that is, they could create more wealth in ways that did not intensify costly climate change.
“Our study makes the first accounting of exactly how much each country has been impacted economically by global warming, relative to historical greenhouse gas emissions,” said Diffenbaugh.
“Historically, rapid economic development has been powered by fossil fuels. Our finding that global warming has exacerbated economic inequality suggests that there is an added economic benefit of energy sources that don’t contribute to further warming.”
