Breaking News
February 21, 2018 - Early childhood immune signature predicts risk of developing asthma later on
February 21, 2018 - Stanford researchers explore how gut bacteria respond to common changes in habitat
February 21, 2018 - Household Products May Pollute the Air as Much as Your Car Does: Study
February 21, 2018 - Combo Bests Targeted Agent in mRCC
February 21, 2018 - Researchers discover brain pathway that dissociates opioid addiction from analgesia
February 21, 2018 - Scientists uncover how newly discovered gene helps grow blood vessels
February 21, 2018 - Brain’s quality control process holds clues to obesity’s roots
February 21, 2018 - Researchers to study whether menstrual cups can help prevent vaginal infections
February 21, 2018 - MS patients who feel stigmatized more likely to suffer from depression
February 21, 2018 - Vitamin D supplementation during pregnancy could protect against childhood obesity
February 21, 2018 - Lower-Quality Medical Tx Might Have Skewed Key PCI vs CABG Trials
February 21, 2018 - Love and fear are visible across the brain instead of being restricted to any brain region
February 21, 2018 - Adults with congenital heart disease have increased risk for dementia, study finds
February 21, 2018 - Clinical trial studying type 1 diabetes reaches full enrollment
February 21, 2018 - Father’s stress affects the brain development of offspring, mice study shows
February 21, 2018 - ESRD Death Declines in Vasculitis Patients
February 21, 2018 - Taking ibuprofen for long periods found to alter human testicular physiology
February 21, 2018 - Google AI device could predict a person’s risk of a heart attack
February 20, 2018 - FDA Approves Domestic Source for Tc-99m Isotopes
February 20, 2018 - Sanofi rejects refund demand faces Philippine suit over dengue vaccine (Update)
February 20, 2018 - Researchers discover that activation of specific enzyme may help suppress tumor metastasis
February 20, 2018 - Blood or marrow transplantation survivors have higher risk of cognitive impairment
February 20, 2018 - Booze Beats Pot at Being Unhealthy: Oregon Poll
February 20, 2018 - Morning Break: ’20 Years Late’; Drugs in the Dirt; Catching Flu in the Dorm
February 20, 2018 - Another piece to the puzzle in naked mole rats’ long, cancer-free life
February 20, 2018 - Scientists identify four viruses that can produce insulin-like hormones
February 20, 2018 - New e-Health solution developed to prevent cardiovascular disease, dementia in senior citizens
February 20, 2018 - New genetic risk score could help guide screening decisions for prostate cancer
February 20, 2018 - Study finds higher risk of stroke among blacks with atrial fibrillation than whites
February 20, 2018 - Physical activity could be used as strategy for diabetes prevention
February 20, 2018 - Researchers develop sensing method for early detection of cancer and diabetes
February 20, 2018 - New wearable electronics could be game-changer for stroke rehabilitation
February 20, 2018 - Immune history influences person’s response to flu vaccine
February 20, 2018 - Research findings could help develop new drugs to prevent, treat dry eye disease
February 20, 2018 - Serenity Now! Learn to Have Patience with Patients
February 20, 2018 - Computer simulation addresses the problem of blood clotting
February 20, 2018 - Women with type 1 diabetes not protected against coronary artery disease
February 20, 2018 - Persistent bloating can be a sign of ovarian cancer, warns charity
February 20, 2018 - Trump administration proposes rule to loosen curbs on short-term health plans
February 20, 2018 - Key protein involved in epigenetic regulation of gene expression guides skin cell renewal
February 20, 2018 - Heart attack symptoms often missed in women
February 20, 2018 - Diagnosis of celiac disease takes 3.5 years for patients who do not report GI symptoms
February 20, 2018 - Study reveals functional dynamics of ion channels
February 20, 2018 - Study explores link between mortality risk and combustible tobacco use
February 20, 2018 - ‘She Trusted Me, and I’d Turned Her Away’
February 20, 2018 - AbbVie and Voyager Therapeutics collaborate to develop new treatments for tauopathies
February 20, 2018 - Fast food makes the immune system more aggressive in the long term
February 20, 2018 - Therapeutic target for glaucoma could have treatment ramifications for Alzheimer’s and Parkinson’s
February 20, 2018 - Overcoming Negative Reviews | Medpage Today
February 20, 2018 - MyD88—villain of allergies and asthma
February 20, 2018 - Food scientists develop rapid screening technique to detect pesticide residue in vegetables
February 20, 2018 - Lab-grown cerebellar cells may help explain how ASD develops at molecular level
February 20, 2018 - Scientists explore connection between bad sleep habits and stiff blood vessels
February 20, 2018 - New Treatment Apalutamide (Erleada) Approved for Prostate Cancer That Resists Hormone Therapy
February 20, 2018 - Do You Really Need My Signature on That?
February 20, 2018 - HIV-1 genetic diversity is higher in vaginal tract than in blood during early infection
February 20, 2018 - Diabetes does not increase work-loss years due to early retirement
February 20, 2018 - Researchers aim to find out how PTSD affects decisions of police
February 20, 2018 - UH Cleveland Medical Center explores novel treatments for uterine fibroids
February 20, 2018 - Flu Vax Efficacy 25% Against Predominant H3N2 Strain So Far
February 20, 2018 - HIV screening most optimal at 25 years of age if no risk factors
February 20, 2018 - Loyola Medicine primary care physician offers advice to minimize risk of flu
February 20, 2018 - Safe sleep recommendations for parents that may help reduce child’s risk of SUID
February 20, 2018 - Why Do So Few Docs Have Buprenorphine Waivers?
February 20, 2018 - Low levels of alcohol good for the brain
February 20, 2018 - Experimental treatment improves invisible symptoms of a man with spinal cord injury
February 20, 2018 - Myriad’s EndoPredict offers better prediction of breast cancer recurrence, analysis shows
February 20, 2018 - Researchers identify fifteen genes that determine our facial features
February 20, 2018 - Morning Break: New Health IT Player; Luxturna No Bargain; Nuclear Freakout
February 20, 2018 - How does it compare? Hospice care at home, at assisted living facility, at nursing home
February 19, 2018 - Scientists develop water-soluble warped nanographene for bioimaging
February 19, 2018 - It’s Not Your Imagination: You’re Hungrier After Losing Weight
February 19, 2018 - Antihypertensive Use At Delivery Rising in Preeclampsia
February 19, 2018 - A centuries-old math equation used to solve a modern-day genetics challenge
February 19, 2018 - Liquid biopsies could be used as new predictive marker for metastatic TNBC
February 19, 2018 - Russian researchers develop new multi-layered biodegradable scaffolds
February 19, 2018 - Are ‘Vaccine Skeptics’ Responsible for Flu Deaths?
February 19, 2018 - Hidden genetic effects behind immune diseases may be missed, study suggests
February 19, 2018 - Emergency nurses experience regular verbal and physical abuse
February 19, 2018 - Study sheds light on biology that guides behavior across different stages of life
The Artificial Brain as Doctor

The Artificial Brain as Doctor

image_pdfDownload PDFimage_print

Computers are learning to distinguish whether skin lesion images are benign or malignant and require further treatment. It appears they are on par with, if not more accurate than, many dermatologists, a study shows.

Stanford University researchers compared a computer-driven deep neural network with dermatologists’ ability to visually classify possible melanoma and basal cell or squamous cell carcinoma lesions.

“We recruited 21 board-certified dermatologists and showed them 300 images, where they had to classify the lesions as benign or malignant, and whether they would biopsy or reassure the patient,” said the study’s lead author, Roberto A. Novoa, MD, of Stanford University in California. “From there, the algorithm performed about as well as the dermatologists, if not better. There were dermatologists who did perform better than the algorithm but, in general, this was a proof of concept study, so … we were demonstrating the efficacy of these algorithms for making the diagnosis.”

Computers That Think Like Humans

Deep neural networks are a type of computer algorithm within the field of artificial intelligence, which uses computers to mimic brain function, including reasoning.

While it might sound like a new concept, it’s not. Researchers reported on the potential for computers to diagnose facial tumors in 1986.

“The idea for neural networks has been around since the 1960s,” Novoa said. “But it was only in the last 5 years that the computing power and technological capabilities caught up to the math.”

Today’s neural networks take in vast amounts of information. They then learn the rules that lie behind the data to derive patterns and, eventually, correct answers, Novoa explained.

“Essentially, deep neural networks are mathematical equations that are stacked in layers. They start at the most basic level by learning the edges of all the objects in an image. Then, they move on to telling you this is a triangle or a square. The next layer might indicate whether an image is a cat or a dog. Finally, there’s a layer that says this is a Belgian Malinois or a German Shepherd … If it gets the answer wrong, it goes back through the equation and changes the values and weights of that equation until it gets the most answers correct for the most number of images. By doing so, it learns, over time, what’s important and what’s not.”

From Dogs to Skin Cancer

The genesis of the research, published in Nature last year, began 3 years ago, when Novoa said he saw the advances in the field of deep learning, where these computer programs could differentiate on their own between Belgian Shepherds and German Shepherds.

“I thought if we can do this for dogs, we can do this for skin cancer,” he said.

Novoa and his colleagues at Stanford gathered a dataset of nearly 130,000 images from the internet, including open source images and images from Stanford databases. The researchers went through the data and created a visual taxonomy for the more than 2,000 disease categories, whittling those down to 10 different general categories. The team then used the dataset in the deep neural network’s classifier.

The researchers measured the algorithm’s performance by creating a sensitivity-specificity curve. Sensitivity represented the algorithm’s ability to correctly identify malignant lesions, and specificity represented its ability to accurately identify benign lesions. Assessing the algorithm through the diagnostic tasks of keratinocyte carcinoma classification, melanoma classification, and melanoma classification when viewed using dermoscopy, the researchers found that the algorithm matched the dermatologists’ performance in all three tasks, with the area under the sensitivity-specificity curve amounting to at least 91% of the total area of the graph, according to a Stanford news release.

The deep neural network, in this case, had been trained to classify skin cancer, but the study was done on images of skin cancer that the network had not yet seen. “If you test it on images it has already seen, it already knows the answer,” Novoa said.

The images in the study already had biopsy-proven results. This allowed the researchers to know the true answers when testing the deep neural network and dermatologists.

The next step is to test the algorithm in the real world, Novoa said. “We’re currently putting together a clinical trial to see how well it will perform in the real world. We want to see how well it performs with real patients, with a smartphone camera and a variety of lighting.”

There have been other studies using artificial intelligence and deep neural networks to diagnose melanoma, but this one is different in a few ways, Novoa explained. “Most of the other studies were performed using dermoscopic images. Dermoscopy kind of limits the number of variables. It’s always taken from the same distance and with pretty similar lighting.

“This was looking at clinical images, but they were from a variety of different distances and angles. This also was looking at melanoma, squamous cell carcinoma, and basal cell carcinomas. So, there was a wider group of images. We used a pretty large dataset — it’s among the larger datasets we’ve seen to date.”

Implications

The basic idea of neural networks is that they do pattern recognition, just as a dermatologist would learn patterns and visual data, Novoa said.

Deep neural networks and artificial intelligence may have a growing role in dermatology to help dermatologists provide better care, but not to replace the need for human expertise. While the computer algorithms might be used to triage or screen patients, ultimately a healthcare professional will take responsibility (and liability) for the care.

Algorithms, though, still need improvement.

“If there are biases in the dataset, these can be introduced into the algorithm results,” he continued. “So, for a long time there is going to be a need for supervision of these results in order to have things work optimally. For example, if the algorithm has a ruler in the image, it’s more likely to call it cancer. Why? Because on average, images in our dataset that have rulers are more likely to be malignant. That’s just a small example of the kinds of biases that can be introduced into the data.”

The near future of the technology has far-reaching implications, including the potential development of such things as a smartphone-compatible algorithm, where consumers could capture a lesion in an image and the phone would screen it for skin cancer probability.

“Technology has been changing medicine and all of human endeavors for hundreds of years, but it hasn’t eliminated the need for doctors,” Novoa said. “I don’t think we’re going anywhere. I think dermatologists will be able to adapt and take advantage of these technologies to take care of patients.”

This article originally appeared on the website of our partner Dermatology Times, which is a part of UBM Medica. (Free registration is required.)

2018-01-15T13:00:00-0500

Tagged with:

About author

Related Articles