Medical photography is failing patients with darker skin – The Verge

Doctors aren’t usually in the business of publishing photography guides. But Jenna Lester, a dermatologist at the University of California San Francisco, was growing frustrated with the poor quality images she’d receive of her dark-skinned patients. It wasn’t just a cosmetic issue — the bad photos meant darker-skinned people weren’t getting the same quality of care.

So in January, Lester co-authored a paper in the British Journal of Dermatology that gives a step-by-step guide to photographing skin of color accurately in clinical settings. Lester, who herself is Black, said, “I feel like these issues and my life is constantly me saying, ‘Hey, what about us?’ ‘What about these patients?’”

Medical photographs are vital to documenting disease in textbooks and journals and training medical students. If they’re not accurate — if, for example, they misidentify important characteristics like texture and color — doctors don’t get as clear a picture of what disease looks like on different people. It then becomes harder for medical professionals to spot potential conditions in their patients. This contributes to disparities in health outcomes for patients of color.

“Across different aspects of medicine, patients with darker skin are getting lower access to screening, delayed diagnoses — and when they are diagnosed, it’s more often at a point where the disease is more severe or higher stage,” says Daniel Cho, a plastic surgery fellow at the Children’s Hospital for Philadelphia. For instance, the skin cancer known as melanoma, although comparatively rarer in Black people, is significantly more likely to kill them. The 5-year survival rates in Black and white individuals are 67 percent and 92 percent, respectively.

Bad photos of dark skin aren’t specific to medicine. Bias has plagued photography since the days of film cameras. Before the advent of commercial digital photography in the late 1990s, lab technicians in camera stores would color-calibrate film by comparing a customer’s photos to a standard called a “Shirley Card.” Named after Shirley Page, the original model who was pictured, these cards were used to ensure the color, hue, and tone of a customer’s image remained consistent and balanced against the so-called default photo. That default was whiteness, as almost all the Shirley Card models were white. This made it harder for photographers to accurately depict details of darker hues. The chemical coatings on photos also weren’t designed to be sensitive to non-Caucasian flesh tones.

The result was that in photos with dark- and light-skinned people together, the darker-skinned person’s features were often erased, lacking the fine detailing afforded to the fair-skinned people. In the mid-1990s, Kodak began introducing multiracial Shirley Cards in an attempt to expand their global reach, which improved the issue.

Since then, digital photography technologies have evolved. For instance, cameras that can balance two different skin tones within the same image now exist and make it easier to represent a greater variety of skin tones at the same time. But the idea of whiteness as the standard in photography has prevailed — creating serious problems, especially in medical photography.

Within highly visual medical fields like dermatology and plastic surgery, physicians rely heavily on photographs of a patient’s skin to aid in diagnoses and treatment plans. But images of conditions on light skin are overrepresented in medical journals and textbooks. That’s problem number one, Lester says. “Problem number two is if you only see one tone of non-white skin. And problem number three is that if that tone of non-white skin looks nothing like it would in real life, how do we make a diagnosis?”

The issue of bad photos of dark-skinned patients is widespread. “I see terrible photos all the time of dark skin,” Lester says. “When I’m flipping through a patient’s chart, and I’m trying to monitor disease over time, the photos that other people have taken have not been great.”

For instance, a bad photo of inflammation on dark skin may make the already subtle condition difficult to see. In a person of color, the use of a flash or bad lighting can drown out the problem, and the inflammation may seem to blend into the skin. On the other hand, a well-captured image of the same taken near a window using natural light could allow a clear distinction between healthy and unhealthy skin.

Lester says the best photos come out of health systems with specialized photography departments, which are rare. Most clinical photos are taken by well-intentioned doctors who haven’t been trained in the nuances of photographing patients of different races. There are fundamental differences in the physics of how light interacts with different skin tones that can make documenting conditions on skin of color more difficult, says Chrystye Sisson, associate professor and chair of the photographic science program at Rochester Institute of Technology, the only such program in the nation.

Interactions between light, objects, and our eyes allow us to perceive color. For instance, a red object absorbs every wavelength of light except red, which it reflects back into our eyes. The more melanin there is in the skin, the more light it absorbs, and the less light it reflects back.

“If you wrote with a red pen on white paper, the white paper reflects all those wavelengths back at you to appear white, and the red pen stands out in comparison,” says Sisson. Darker paper, though, absorbs more light. If someone writes with the same pen on that darker paper, the red reflection doesn’t stand out as much. That’s why some red skin conditions like inflammation or rosacea can appear more subtle on darker skin.

But standard photographic setups don’t account for those differences. For instance, although many medical journals recommend using flash for photographing patients, it can compromise detail when done on dark skin. The contrast of the bright flash of the camera against dark subjects can make dark things appear darker than they are, reducing detail in dark skin. The reflection of the flash back from the skin can also obscure the skin tone underneath and leave a shine, an effect that is more pronounced on darker skin. “Whatever standard setup you come up with needs to take that into account,” Sisson says.

These biases may also be increasingly cemented within new technologies, like artificial intelligence algorithms that are trained to detect conditions using clinical photographs. Google Health announced in May that it is hoping to pilot an AI-powered dermatology assistant tool to assess skin conditions and provide access to authoritative information on them. The company trained a deep learning algorithm on a set of over 16,000 pictures of various skin conditions. But only 3.5% of these cases depicted dark and deep brown skin. Out of these photos of dark skin, it’s impossible to tell how many of them were taken using techniques that make sure they’re truly representative.

Clinical images are also frequently used in teaching materials, meaning the next generation of medical professionals is being trained with a scarcity of representative photos. And as telehealth becomes increasingly important, patients will need to be taught how to take accurate images of their conditions — which becomes hard when even their doctors don’t know how.

The consequences of these compounded issues are what motivated Lester to compile the set of tips for photographing skin of color in a medical context — instructing people to pay close attention to lighting and choosing the color of background drapes carefully, along with other techniques. Lester hopes that her guidelines will prompt others in the field to question what “standard” practice means and who it was built for and serve as a stepping stone in bridging health equity gaps.

“Recognizing that there are nuances and differences is an important first step in making the necessary adjustments to having a photograph that looks representative,” she says.

Source

Leave a Reply

Your email address will not be published. Required fields are marked *