For centuries, scientists have trained themselves to look through microscopes and carefully study their structural and molecular features. But those long hours bent over a microscope poring over microscopic images could be less necessary in the years ahead. The job of analyzing cellular features could one day belong to specially trained computers.
In a new study published in the journal Cell, researchers trained computers by feeding them paired sets of fluorescently labeled and unlabeled images of brain tissue millions of times in a row . This allowed the computers to discern patterns in the images, form rules, and apply them to viewing future images. Using this so-called deep learning approach, the researchers demonstrated that the computers not only learned to recognize individual cells, they also developed an almost superhuman ability to identify the cell type and whether a cell was alive or dead. Even more remarkable, the trained computers made all those calls without any need for harsh chemical labels, including fluorescent dyes or stains, which researchers normally require to study cells. In other words, the computers learned to “see” the invisible!
Posted In: News
Tags: Alzheimer's disease, brain, Brain Bot, cell biology, cells, computer learning, computers, deep learning, Google, machine learning, microscopy, neurology, neuroscience, Parkinson's disease, schizophrenia
Early detection usually offers the best chance to beat cancer. Unfortunately, many tumors aren’t caught until they’ve grown relatively large and spread to other parts of the body. That’s why researchers have worked so tirelessly to develop new and more effective ways of screening for cancer as early as possible. One innovative approach, called “liquid biopsy,” screens for specific molecules that tumors release into the bloodstream.
Recently, an NIH-funded research team reported some encouraging results using a “universal” liquid biopsy called CancerSEEK . By analyzing samples of a person’s blood for eight proteins and segments of 16 genes, CancerSEEK was able to detect most cases of eight different kinds of cancer, including some highly lethal forms—such as pancreatic, ovarian, and liver—that currently lack screening tests.
In a study of 1,005 people known to have one of eight early-stage tumor types, CancerSEEK detected the cancer in blood about 70 percent of the time, which is among the best performances to date for a blood test. Importantly, when CancerSEEK was performed on 812 healthy people without cancer, the test rarely delivered a false-positive result. The test can also be run relatively cheaply, at an estimated cost of less than $500.
Tags: blood test, breast cancer, cancer, cancer blood test, cancer detection, cancer diagnostics, CancerSEEK, clinical study, colorectal cancer, early detection, esophageal cancer, liquid biopsy, liver cancer, lung cancer, machine learning, ovarian cancer, pancreatic cancer, stomach cancer, universal liquid biopsy
Science has always fascinated Anshul Kundaje, whether it was biology, physics, or chemistry. When he left his home country of India to pursue graduate studies in electrical engineering at Columbia University, New York, his plan was to focus on telecommunications and computer networks. But a course in computational genomics during his first semester showed him he could follow his interest in computing without giving up his love for biology.
Now an assistant professor of genetics and computer science at Stanford University, Palo Alto, CA, Kundaje has received a 2016 NIH Director’s New Innovator Award to explore not just how the human genome sequence encodes function, but also why it functions in the way that it does. Kundaje even envisions a time when it might be possible to use sophisticated computational approaches to predict the genomic basis of many human diseases.
Tags: 2016 NIH Director’s New Innovator Award, Alzheimer’s disease, artificial neural networks, cancer, colorectal cancer, computational genomics, computer science, DNA, DNA elements, ENCODE, epigenomics, gene function, gene variants, genomics, heart disease, machine learning, MYC, noncoding DNA, Roadmap Epigenomics Project, transcription factor, yeast
Research shows that the roots of autism spectrum disorder (ASD) generally start early—most likely in the womb. That’s one more reason, on top of a large number of epidemiological studies, why current claims about the role of vaccines in causing autism can’t be right. But how early is ASD detectable? It’s a critical question, since early intervention has been shown to help limit the effects of autism. The problem is there’s currently no reliable way to detect ASD until around 18–24 months, when the social deficits and repetitive behaviors associated with the condition begin to appear.
Several months ago, an NIH-funded team offered promising evidence that it may be possible to detect ASD in high-risk 1-year-olds by shifting attention from how kids act to how their brains have grown . Now, new evidence from that same team suggests that neurological signs of ASD might be detectable even earlier.
Tags: ASD, autism, Autism Spectrum Disorder, brain, brain connectivity, brain connectivity maps, brain development, brain scans, childhood disease, childhood vaccinations, computer learning, diagnosis, early detection, fMRI, Functional magnetic resonance imaging, imaging, infants, machine learning, neuroimaging, neurology, vaccines