AI Achieves Near-Human Detection of Breast Cancer


Pathologists still do the bulk of their diagnosis of metastatic cancer cells in tissue and lymph nodes by hand, putting slides under a microscope and looking for signature irregularities they’re trained to see. Recent advances in computer technology, however, particularly in artificial intelligence (AI), have begun to teach machines to do this kind of detection with growing rates of improvement.

Now, a research team from Beth Israel Deaconess Medical Center (BIDMC) and Harvard Medical School have developed a form of AI that can interpret these pathology images with accuracy levels of 92.5 percent. That's not too far below the human detection rate of 97 percent. Moreover, when the two are used in combination, the detection rate approaches 100 percent (approximately 99.5 percent).

Their AI method is a form of deep learning, in which the system attempts to replicate the activity of the human neocortex through artificial neural networks. The goal was to teach the machine to interpret patterns and structuresAndrew Beck, director of bioinformatics at the Cancer Research Institute at BIDMC and an associate professor at Harvard Medical School, is co-author of the technical report describing these findings, recently uploaded to [PDF], an open access archive. He tells mental_floss, “We use a subset of AI where you are trying to train the computer to do something in a data-driven way to learn model parameters and to make predictions on new examples.”

To teach and test the AI, they input 400 whole slide images—270 for teaching and 130 for testing. Some of the slides contained metastatic breast cancer lymph node tissue, and some healthy tissue. The team was able to identify which slides the computer was more prone to making mistakes about—primarily by flagging false positives—and used those examples to re-train the computer to improve its performance.

They submitted their system to the International Symposium of Biomedical Imaging (ISBI), where they placed first in two categories in the ISBI's Camelyon Grand Challenge 2016up against private companies and academic research institutions from around the world. According to ISBI’s website, the goal of this challenge is "to evaluate new and existing algorithms for automated detection of metastases … in stained whole-slide images of lymph node sections."

Beck was surprised at how efficient the system turned out to be. “I was impressed at how well the computer did, because it is really a complicated visual task," he says. "The cancer can take a bunch of different appearances and the normal lymph nodes, too. To think that a single model in a purely data-driven way could accurately make this classification was surprising.”

It did a much more accurate job of detecting cancer than a recent study that reported that pigeons had an 85 percent accuracy rate of detecting breast cancer individually; when the scores of a flock of four were combined, they had a 99 percent accuracy rate. Beck feels that associating the two studies is like comparing apples to oranges because his study wasn’t diagnosing breast cancer, but breast cancer in lymph nodes, he explains. "It wasn’t trying to separate normal breasts from pre-invasive breast lesions and breast cancers."

Moreover, he says, "I think you can imagine computers being worked into the workflow much more simply than pigeons.”

One especially positive application of this kind of AI is its ability to remove some of the burden of detection from the pathologist, who can then focus more on treatment plans and patient health. “You can imagine that in the future the computer will keep getting better. I can see things evolving where pathologists move away from the more tedious, lower-level tasks because there are higher-level, more integrated things that humans are way better at than computers,” says Beck. For instance, the computer could count all of the individual cells.

It may also help solve diagnostic errors by improving accuracy in combination with the manual method. Further research by his team will continue to test the system by expanding the types of cancer used, and increasing the number of slides. “This could be integrated into existing workflows to make the process faster, more accurate and hopefully more cost effective, ranging from the clinic, to research in pharmaceutical companies, to global health,” says Beck.

Beck has since formed the start-up company PathAI with Aditya Khosla of the MIT Computer Science and Artificial Intelligence Laboratory. It aims to develop and apply AI technology to pathology.

New Cancer-Fighting Nanobots Can Track Down Tumors and Cut Off Their Blood Supply

Scientists have developed a new way to cut off the blood flow to cancerous tumors, causing them to eventually shrivel up and die. As Business Insider reports, the new treatment uses a design inspired by origami to infiltrate crucial blood vessels while leaving the rest of the body unharmed.

A team of molecular chemists from Arizona State University and the Chinese Academy of Sciences describe their method in the journal Nature Biotechnology. First, they constructed robots that are 1000 times smaller than a human hair from strands of DNA. These tiny devices contain enzymes called thrombin that encourage blood clotting, and they're rolled up tightly enough to keep the substance contained.

Next, researchers injected the robots into the bloodstreams of mice and small pigs sick with different types of cancer. The DNA sought the tumor in the body while leaving healthy cells alone. The robot knew when it reached the tumor and responded by unfurling and releasing the thrombin into the blood vessel that fed it. A clot started to form, eventually blocking off the tumor's blood supply and causing the cancerous tissues to die.

The treatment has been tested on dozen of animals with breast, lung, skin, and ovarian cancers. In mice, the average life expectancy doubled, and in three of the skin cancer cases tumors regressed completely.

Researchers are optimistic about the therapy's effectiveness on cancers throughout the body. There's not much variation between the blood vessels that supply tumors, whether they're in an ovary in or a prostate. So if triggering a blood clot causes one type of tumor to waste away, the same method holds promise for other cancers.

But before the scientists think too far ahead, they'll need to test the treatments on human patients. Nanobots have been an appealing cancer-fighting option to researchers for years. If effective, the machines can target cancer at the microscopic level without causing harm to healthy cells. But if something goes wrong, the bots could end up attacking the wrong tissue and leave the patient worse off. Study co-author Hao Yan believes this latest method may be the one that gets it right. He said in a statement, "I think we are much closer to real, practical medical applications of the technology."

[h/t Business Insider]

Photo by Fox Photos/Getty Images
Essential Science
How Are Vaccines Made?
Quality checks on the Salk polio vaccine at Glaxo's virus research laboratory in Buckinghamshire, UK, in January 1956.
Quality checks on the Salk polio vaccine at Glaxo's virus research laboratory in Buckinghamshire, UK, in January 1956.
Photo by Fox Photos/Getty Images

Vaccines have long been hailed as one of our greatest public health achievements. They can be made to protect us from infections with either viral or bacterial microbes. Measles and smallpox, for example, are viruses; Streptococcus pneumoniae is a bacterium that causes a range of diseases, including pneumonia, ear and sinus infections, and meningitis. Hundreds of millions of illnesses and deaths have been prevented due to vaccines that eradicated smallpox and significantly reduced polio and measles infections. However, some misunderstanding remains regarding how vaccines are made, and why some scary-sounding ingredients [PDF] are included in the manufacturing process.

The production of our vaccines has greatly evolved since the early days, when vaccination was potentially dangerous. Inoculating an individual with ground-up smallpox scabs usually led to a mild infection (called "variolation"), and protected them from acquiring the disease the "regular" way (via the air). But there was always a chance the infection could still be severe. When Edward Jenner introduced the first true vaccination with cowpox, protection from smallpox became safer, but there were still issues: The cowpox material could be contaminated with other germs, and sometimes was transmitted from one vaccinated person to another, leading to the inadvertent spread of blood-borne pathogens. We’ve come far in the last 200 years.

There are different kinds of vaccines, and each requires different processes to move from the laboratory to your physician's office. The key to all of them is production of one or more antigens—the portion of the microbe that triggers a host immune response.


There are several methods to produce antigens. One common technique is to grow a virus in what's called a cell culture. Typically grown in large vats called bioreactors, living cells are inoculated with a virus and placed in a liquid growth medium that contains nutrients—proteins, amino acids, carbohydrates, essential minerals—that help the virus grow in the cells, producing thousands of copies of itself in each infected cell. At this stage the virus is also getting its own dose of protective medicine: antibiotics like neomycin or polymyxin B, which prevent bacterial and fungal contamination that could kill the cells serving as hosts for the virus.

Once a virus completes its life cycle in the host cell, the viruses are purified by separating them from the host cells and growth media, which are discarded. This is often done using several different types of filters; the viruses are small and can pass through holes in the filter that trap larger host cells and cell debris.

This is how "live attenuated vaccines" are created. These vaccines contain viruses that have been modified so that they are no longer harmful to humans. Some of them are grown for many generations in cells that aren't human, such as chicken cells, so that they have mutated to no longer cause harm to humans. Others, like the influenza nasal mist, were grown at low temperatures until they lost the ability to replicate in the warmer temperatures of the lungs. Many of these vaccines you were probably given as a child: measles, mumps, rubella ("German measles"), and chickenpox.

Live attenuated vaccines replicate briefly in the body, triggering a strong—and long-lasting—response from your immune system. Because your immune system kicks into high gear at what it perceives to be a major threat, you need fewer doses of the vaccine for protection against these diseases. And unlike the harmful form of the virus, it is extremely unlikely (because they only replicate at low levels) that these vaccines will cause the host to develop the actual disease, or to spread it to other contacts. One exception is the live polio vaccine, which could spread to others and, extremely rarely, caused polio disease (approximately one case of polio from 3 million doses of the virus). For this reason, the live polio virus was discontinued in the United States in 2000.

Scientists use the same growth technique for what are known as "killed" or "inactivated" vaccines, but they add an extra step: viral death. Inactivated viruses are killed, typically via heat treatment or use of a chemical such as formaldehyde, which modifies the virus's proteins and nucleic acids and renders the virus unable to replicate. Inactivated vaccines include Hepatitis A, the injected polio virus, and the flu shot.

A dead virus can't replicate in your body, obviously. This means that the immune response to inactivated vaccines isn't as robust as it is with live attenuated vaccines; replication by the live viruses alerts many different types of your immune cells of a potential invader, while killed vaccines primarily alert only one part of your immune system (your B cells, which produce antibodies). That's why you need more doses to achieve and maintain immunity.

While live attenuated vaccines were the primary way to make vaccines until the 1960s, concerns about potential safety issues, and the difficulty of making them, mean that few are attempting to develop new live attenuated vaccines today.


Other vaccines aren't made of whole organisms at all, but rather bits and pieces of a microbe. The combination vaccine that protects against diphtheria, pertussis, and tetanus—all at once—is one example. This vaccine is called the DTaP for children, and Tdap for adults. It contains toxins (the proteins that cause disease) from diphtheria, pertussis, and tetanus bacteria that have been inactivated by chemicals. (The toxins are called "toxoids" once inactivated.) This protects the host—a.k.a. you, potentially—from developing clinical diphtheria and tetanus disease, even if you are exposed to the microorganisms. (Some viruses have toxins—Ebola appears to, for example—but they're not the key antigens, so they're not used for our current vaccines.)

As they do when developing live attenuated or inactivated vaccines, scientists who create these bacterial vaccines need some target bacteria to culture. But because the bacteria don't need a host cell to grow, they can be produced in simple nutrient broths by vaccine manufacturers. The toxins are then separated from the rest of the bacteria and growth media and inactivated for use as vaccines.

Similarly, some vaccines contain just a few antigens from a bacterial species. Vaccines for Streptococcus pneumoniae, Haemophilus influenzae type B, and Neisseria meningitidis all use sugars that are found on the outer part of the bacteria as antigens. These sugars are purified from the bacteria and then bound to another protein to enhance the immune response. The protein helps to recruit T cells in addition to B cells and create a more robust reaction.

Finally, we can also use genetic engineering to produce vaccines. We do this for Hepatitis B, a virus that can cause severe liver disease and liver cancer. The vaccine for it consists of a single antigen: the hepatitis B surface antigen, which is a protein on the outside of the virus. The gene that makes this antigen is inserted into yeast cells; these cells can then be grown in a medium similar to bacteria and without the need for cell culture. The hepatitis B surface antigen is then separated from the yeast and serves as the primary vaccine component.


Once you have the live or killed viruses, or purified antigens, sometimes chemicals need to be added to protect the vaccine or to make it work better. Adjuvants, such as aluminum salts, are a common additive; they help enhance the immune response to some antigens by keeping the antigen in contact with the cells of the immune system for a longer period of time. Vaccines for DTaP/Tdap, meningitis, pneumococcus, and hepatitis B all use aluminum salts as an adjuvant.

Other chemicals may be added as stabilizers, to help keep the vaccine working effectively even in extreme conditions (such as hot temperatures). Stabilizers can include sugars or monosodium glutamate (MSG). Preservatives can be added to prevent microbial growth in the finished product.

For many years, the most common preservative was a compound called thimerosal, which is 50 percent ethylmercury by weight. Ethylmercury doesn't stick around; your body quickly eliminates it via the gut and feces. (This is different from methylmercury, which accumulates in fish and can, at high doses, cause long-lasting damage in humans.) In 2001, thimerosal was removed from the vaccines given in childhood due to consumer concerns, but many studies have demonstrated its safety.

Finally, the vaccine is divided into vials for shipping to physicians, hospitals, public health departments, and some pharmacies. These can be single-dose or multi-dose vials, which can be used for multiple patients as long as they're prepared and stored away from patient treatment areas. Preservatives are important for multi-dose vials: bacteria and fungi are very opportunistic, and multiple uses increase the potential for contamination of the vaccine. This is why thimerosal is still used in some multi-dose influenza vaccines.

Though some of the vaccine ingredients sound worrisome, most of these chemicals are removed during multiple purification steps, and those that remain (such as adjuvants) are necessary for the vaccine's effectiveness, are present in very low levels, and have an excellent track record of safety.


More from mental floss studios