Researchers develop automated melanoma detector for skin cancer screening

December 25, 2016
Malignant or benign?: An image of a skin lesion is processed by a new technology to extract quantitative data, such as irregularities in the shape of pigmented skin, which could help doctors determine if the growth is cancerous. Credit: Rockefeller University

Even experts can be fooled by melanoma. People with this type of skin cancer often have mole-looking growths on their skin that tend to be irregular in shape and color, and can be hard to tell apart from benign ones, making the disease difficult to diagnose.

Now, researchers at The Rockefeller University have developed an automated technology that combines imaging with digital analysis and machine learning to help physicians detect at its early stages.

"There is a real need for standardization across the field of dermatology in how melanomas are evaluated," says James Krueger, D. Martin Carter Professor in Clinical Investigation and head of the Laboratory of Investigative Dermatology. "Detection through screening saves lives but is very challenging visually, and even when a suspicious lesion is extracted and biopsied, it is confirmed to be melanoma in only about 10 percent of cases."

In the new approach, images of lesions are processed by a series of computer programs that extract information about the number of colors present in a growth, and other quantitative data. The analysis generates an overall risk score, called a Q-score, which indicates the likelihood that the growth is cancerous.

Published in Experimental Dermatology, a recent study evaluating the tool's usefulness indicates that the Q-score yields 98 percent sensitivity, meaning it is very likely to correctly identify early melanomas on the skin. The ability of the test to correctly diagnose normal moles was 36 percent, approaching the levels achieved by expert dermatologists performing visual examinations of suspect moles under the microscope.

"The success of the Q-score in predicting melanoma is a marked improvement over competing technologies," says Daniel Gareau, first author of the report and instructor in in the Krueger laboratory.

The researchers developed this tool by feeding 60 photos of cancerous melanomas and an equivalent batch of pictures of benign growths into image processing programs. They developed imaging biomarkers to precisely quantify visual features of the growths. Using computational methods, they generated a set of quantitative metrics that differed between the two groups of images—essentially identifying what visual aspects of the lesion mattered most in terms of malignancy—and gave each biomarker a malignancy rating.

By combining the data from each biomarker, they calculated the overall Q-score for each image, a value between zero and one in which a higher number indicates a higher probability of a lesion being a cancerous.

As previous studies have shown, the number of colors in a lesion turned out to be the most significant biomarker for determining malignancy. And some biomarkers were significant only if looked at in specific color channels—a finding the researchers say could potentially be exploited to identify additional biomarkers and further improve accuracy.

"I think this technology could help detect the disease earlier, which could save lives, and avoid unnecessary biopsies too," says Gareau. "Our next steps are to evaluate this method in larger studies, and take a closer look at how we can use specific color wavelengths to reveal aspects of the lesions that may be invisible to the human eye, but could still be useful in diagnosis."

Explore further: Computer program using skin images could improve early detection of melanoma

More information: Daniel S. Gareau et al. Digital imaging biomarkers feed machine learning for melanoma screening, Experimental Dermatology (2016). DOI: 10.1111/exd.13250

Related Stories

Computer program using skin images could improve early detection of melanoma

October 1, 2015
A computer program easily run on a tablet or a smartphone that compares a spot on a patient's skin to images of both benign and malignant lesions may be a useful, noninvasive tool for physicians when deciding when to do a ...

Expression of specific gene differentiates moles from melanoma

November 21, 2016
Most melanomas are driven by mutations that spur out-of-control cell replication, while nevi (moles composed of non-cancerous cells at the skin surface) harboring the same mutations do not grow wildly. However, changes in ...

Using clinical features to identify patients at high risk for melanoma

November 9, 2016
Can an individual's risk factors for melanoma be used to tailor skin self-examinations and surveillance programs? A new study published online by JAMA Dermatology suggests they could by identifying those patients at higher ...

Finding skin cancer in a flash

July 29, 2016
The typical nude skin cancer checks with long photo sessions at your dermatologist's office to track any suspicious skin marks just got a lot more accurate, while reducing the chance of unnecessary biopsies.

Recommended for you

Shooting the achilles heel of nervous system cancers

July 20, 2017
Virtually all cancer treatments used today also damage normal cells, causing the toxic side effects associated with cancer treatment. A cooperative research team led by researchers at Dartmouth's Norris Cotton Cancer Center ...

Molecular changes with age in normal breast tissue are linked to cancer-related changes

July 20, 2017
Several known factors are associated with a higher risk of breast cancer including increasing age, being overweight after menopause, alcohol intake, and family history. However, the underlying biologic mechanisms through ...

Immune-cell numbers predict response to combination immunotherapy in melanoma

July 20, 2017
Whether a melanoma patient will better respond to a single immunotherapy drug or two in combination depends on the abundance of certain white blood cells within their tumors, according to a new study conducted by UC San Francisco ...

Discovery could lead to better results for patients undergoing radiation

July 19, 2017
More than half of cancer patients undergo radiotherapy, in which high doses of radiation are aimed at diseased tissue to kill cancer cells. But due to a phenomenon known as radiation-induced bystander effect (RIBE), in which ...

Definitive genomic study reveals alterations driving most medulloblastoma brain tumors

July 19, 2017
The most comprehensive analysis yet of medulloblastoma has identified genomic changes responsible for more than 75 percent of the brain tumors, including two new suspected cancer genes that were found exclusively in the least ...

Novel CRISPR-Cas9 screening enables discovery of new targets to aid cancer immunotherapy

July 19, 2017
A novel screening method developed by a team at Dana-Farber/Boston Children's Cancer and Blood Disorders Center—using CRISPR-Cas9 genome editing technology to test the function of thousands of tumor genes in mice—has ...

0 comments

Please sign in to add a comment. Registration is free, and takes less than a minute. Read more

Click here to reset your password.
Sign in to get notified via email when new comments are made.