Comentários do leitor

How to spot skin cancer with your phone

por Jere Petterd (2019-09-14)

id="article-body" class="row" section="article-body"> Early detection of skin cancer could be the difference between a simple mole removal or several rounds of chemotherapy. 

SkinVision We are in the height of summer, which for most of us means spending long sunny days outside, and getting exposed to UV rays

The scary part of all that time in the sun is that it can lead to skin cancer, which accounts for more diagnoses each year than all other cancers. The good news is that early detection could be the difference between a simple mole removal or malignant cancer that spreads to other parts of the body. 

A handful of smartphone apps and devices claim to aid early detection and keep you on track with regular self-exams. You can capture photos of suspicious moles or marks and track them yourself, or send them off to a dermatologist for assessment. Either way, these apps can be helpful, but they do have limitations. Here's what you need to know about using your smartphone to detect skin cancer. 

Read more: The best sunscreens to protect your skin in 2019

Know the facts about skin cancer
Every year, doctors diagnose more than 4 million cases of nonmelanoma (including basal and squamous cell) skin cancers in the US, and it's estimated that nearly 200,000 people will receive a melanoma diagnosis in 2019. 

Basal and squamous cell skin cancers develop on the outer layers of the skin and are more common, though less harmful, than melanoma. 

Melanoma is the deadliest form of skin cancer. It forms in the cells responsible for skin pigmentation, called melanocytes. It's an aggressive form of cancer and accounts for nearly 10,000 deaths each year. Even with early detection, it can be fatal.

Symptoms of all types of skin cancers include:

Change in the size or color of a mole or other spot on the skin

A new growth on the skin

Odd skin sensations, such as persistent itchiness or tenderness

Spread of pigmentation outside the border of a mole
Skin cancer may develop due to a variety of factors, including genetics and exposure to toxic chemicals, but the clearest connection is that of skin cancer and UV exposure. 

Read more: I got my face scanned for wrinkles, sun damage and acne scars. The results were mind-blowing

Now playing: Watch this: Procter & Gamble's freckle-erasing makeup wand is pure... 1:14 How your phone can help you spot skin cancer
Telemedicine is a growing field, and skin care is not to be left out: Over the last several years, a handful of skin cancer detection apps popped up allowing you to analyze your skin with your smartphone and artificial intelligence algorithms. 

Some send photos to a dermatologist, some provide instant feedback and others offer helpful reminders about self-checking your skin and scheduling a doctor's appointment. 

Here are a few you can download on iOS and Android. 

Miiskin uses hi-res digital photography to capture magnified photos of moles on your skin. 

Miiskin Miiskin 
Miiskin uses mole mapping to analyze your skin. Dermatologists perform mole maps as part of a clinical full-body skin exam, using digital dermoscopy (magnified digital photography) to catch suspicious lesions they may not catch with their own eyes. 

Because they're so high-definition, dermoscopy photos provide much more information than normal digital photos. The developers behind Miiskin wanted to offer a version of this technology to consumers, so they built an app that takes magnified photos of large areas of your skin, for example, your entire leg. According to the website, anyone with an iPhone with iOS 10 and newer or a phone running Android 4.4 and newer can use Miiskin.

The app stores your photos separate from your smartphone library and allows you to compare moles over time, which is helpful in detecting changes. 

Find it: iOS | Android

This app comes from researchers at the University of Michigan (UM) school of medicine and allows you to complete a full-body skin cancer self-exam, as well as create and track a history of moles, growths and lesions.

The app guides you step-by-step on how to complete the  exam with graphics and written instructions. UMSkinCheck also comes with access to informational videos and articles, as well as a melanoma risk calculator. 

UMSkinCheck also sends push reminders to encourage people to follow-up on their self-exams and check on the lesions or moles they are tracking. You can decide how often you want to see those reminders in the app.

Find it: iOS | Android

With a clip-on camera, MoleScope uses the ABCD method to complete a risk assessment of your moles.

MoleScope Like Miiskin, MoleScope uses magnified images to help people determine whether they should see a dermatologist to get their skin checked. 

A product of MetaOptima (a supplier of clinical dermatology technology) MoleScope is a device that attaches to your smartphone and sends photos to a dermatologist for an online checkup.

Though MoleScope itself won't analyze or diagnose your moles, you can use the ABCD guide in the app to keep tabs on any suspicious moles: The app helps you document your moles with photos and sends them to a dermatologist, who can assess them using the ABCD method:  

Asymmetry: the shape of one half doesn't match the other

Border: edges are bumpy, ragged or blurred

Color: uneven shades of brown, black and tan; odd colors such as red or blue

Diameter: a change in size greater than 6 mm
Unlike Miiskin, you can only take photos of one mole or Folexin small areas with a few moles, rather than large areas like your entire chest or back. 

Find it: iOS | Android

SkinVision claims to aid early detection of melanoma. The app uses deep learning to analyze photos of your skin and aid in the early detection of skin cancer. The photos are processed through a machine-learning algorithm that filters image layers based on simple, complex, and more abstract functions and patterns through a technology called convolutional neural network (CNN). SkinVision uses it to check small areas of your skin and come back with a high- or low-risk assessment of that area in less than a minute. 

SkinVision is backed by a scientific board of dermatologists, but Dr. Daniel Friedmann, a dermatologist at Westlake Dermatology in Austin, Texas, told CNET that even an app with prominent support of scientists has limitations. 

"I would not recommend that patients avoid these apps, but I would approach their results with cautious skepticism," Dr. Friedmann said, "and counsel patients that suspicious lesions are best evaluated in-office." 

Find it: iOS | Android

SkinVision uses a machine-learning algorithm to analyze spots on the skin.

SkinVision Read more: The easiest way to protect your skin from the sun is already on your phone

The accuracy isn't there yet
Of all the apps discussed here, SkinVision seems to have the most research behind it. Researchers tested the accuracy of the app by analyzing nearly 200 photos of skin lesions that the app had previously analyzed. The researchers found that SkinVision was 81% accurate in detecting melanoma, with 73% sensitivity and 83% specificity, compared to the 88% sensitivity and 97% specificity seen in clinical exams. 

In the name of scientific integrity, it's important to note that some of the study authors had affiliations with SkinVision, and the study was funded in part by SkinVision.

Regardless of affiliations, the 81% accuracy still isn't enough to replace a professional exam: The researchers note that "With an accuracy of 81% in detecting melanoma, the algorithm might have some potential in the future …  but is to date insufficient to detect melanoma accurately."

That consensus holds strong in other research. In another study, researchers from the University of Pittsburgh, analyzed four smartphone apps that claim to detect skin cancer. We don't know the exact apps, as they're named only as Application 1, 2, 3 and 4. Three of the apps used algorithms to send immediate feedback about the person's risk of skin cancer, and the fourth app sent the photos to a dermatologist.

Unsurprisingly, the researchers found the fourth app be the most accurate. The other three apps were found to incorrectly categorize a large number of skin lesions, with one missing nearly 30% of melanomas, classifying them as low-risk lesions.

A 2018 Cochrane review of prior research found that AI-based skin cancer detection has "not yet demonstrated sufficient promise in terms of accuracy, and they are associated with a high likelihood of missing melanomas."

To be fair, much of this research took place a few years ago, and the manufacturers may very well have improved their technology since then. More recently, in 2017, a team of researchers at Stanford University announced that their AI does just as well as an in-person dermatologist in detecting skin cancer -- showing that these apps and algorithms do hold promise.