AI helps uncover woman’s abnormal cells, elevated breast cancer risk

NEWPORT NEWS, Va.— An artificial intelligence program that flags potential abnormalities in mammograms helped identify an area with cells linked to a higher risk for breast cancer in a Virginia Beach woman’s breast.

“I would do anything to prevent myself from ever having cancer,” said Hannah Bocks. “That’s in God’s hands, but doing my part [gives me peace of mind].”

Bocks felt a lump in her breast about a year ago, and it scared her.

“My sister is five years younger than me, and at 31, she was diagnosed with breast cancer—Invasive carcinoma,” Bocks said during a conversation with me about her concerns.

“She’s in remission,” said Bocks with tears of joy in her eyes. “[But] because it happened to her so young, it’s kind of always scared our family.”

Bocks told me she had her first mammogram about a year ago at age 40 to investigate what she described as a large lump in her breast. She said doctors told her she had dense breast tissue, and the lump in her breast was nothing more than a cyst.

However, Bocks was still concerned. She had a right to be, too.

While doing research about mammograms and their ability to detect breast cancer, I learned screening mammograms can miss one in eight breast cancers, according to the American Cancer Society. The research also revealed that mammograms are more likely to miss cancer in people with dense breast tissue, like Bocks.

“That tissue density can affect the ability for the mammogram to see [cancer],” said Dr. Benjamin Pettus, a radiologist within the Riverside Health System.

Bocks, whose role as the Assistant Chief Engineer at WTKR-TV ensures the technology is in place to get our stories on the air, saw my report last year about Riverside Health System’s use of an artificial intelligence program to help radiologists detect signs of cancer the naked eye may miss.

“I saw it on our newscast and instantly was like, I need to do this!” Bocks said. “I need to experience this myself because of my worries from last year, and just put my mind at peace.”

Bocks made an appointment with Dr. Pettus at Riverside Health System in January and allowed me and photojournalist Lydia Johnson to document the process.

After receiving a 3D mammogram— which studies reveal have improved screening outcomes— Dr. Pettus reviewed Bocks’ scans with the help of the Transpara Breast Care artificial intelligence portal. The use of the technology comes at no extra cost to any patient at Riverside, he said.

The artificial intelligence flagged an area of Bocks’ scan as an “Intermediate Risk”.

Dr. Pettus said after reviewing Bocks’ previous mammogram to compare how the area of concern has changed.

“[The AI’s scoring is] helpful information to me because this patient has multiple cysts everywhere, but unlike [her] other cysts that have a really defined edge to me, the one it’s scoring does have a subtlety questionable edge,” the doctor told her.

Based on his own experience reading mammograms, the artificial intelligence’s scoring, and Bocks’ family history, Dr. Pettus ordered an ultrasound of her breast. The area still looked suspicious in that screening, too, so Dr. Pettus ordered a biopsy.

“It’s been decided that I’m going to follow up with a biopsy just to be 100 percent sure of what we are looking at,” Bocks shared with me in a candid video following her appointment.

The biopsy, according to Dr. Pettus, indicated the presence lobular carcinoma in situ, or LCIS. It is not cancer—even though the term “carcinoma” gave Hannah a scare— but it is abnormal cells that indicate an elevated risk for developing breast cancer.

Watch previous coverage: WTKR engineer looks to AI technology to screen lump for breast cancer

WTKR engineer looks to AI technology to screen lump for breast cancer

“About 20 percent of the women that we find [these abnormal cells], somewhere in their future they may have a breast cancer develop somewhere in either breast,” said Dr. Pettus.

I said to Dr. Pettus, “This is the same area that the AI flagged and [indicated] something’s up here. It wasn’t cancer, but there is something [to be concerned about]?”

“There’s something about that area that was different,” Dr. Pettus responded. “So again, it’s good that we found it.”

Dr. Pettus referred Bocks to a surgeon following the biopsy results. Bocks told me she and the surgeon decided the best course of action was to surgically remove the area with the abnormal cells.

The surgery is set for April. Bocks told me she’s grateful about how this could decrease her chances of developing cancer, or at least giving her a head start.

I asked Bocks, “What does your sister think about all this?”

“Oh, she’s like, get the spot out,” she told me. “Let’s be proactive.”

While some people with these abnormal cells decide— like Bocks did—to have the area surgically removed, others may decide to have more frequent screenings or undergo a preventive double mastectomy.

Watch related story: Virginia Beach woman diagnosed with breast cancer while pregnant after years of voicing concerns

Virginia Beach woman diagnosed with breast cancer while pregnant after years of voicing concerns

Dr. Pettus told me artificial intelligence programs can help radiologists detect signs of cancer in mammograms that screening by a radiologist may miss, and studies I reviewed support Dr. Pettus’ findings.

He’s been using the technology to help screen mammograms this year, and last month, Riverside Health System announced that all their patients’ mammograms will be analyzed by the Transpara Breast Care artificial intelligence portal at no additional cost.

However, a study I read by the Radiological Society of North America found that automation bias “impaired the performance of radiologists with varying levels of expertise.”

I asked Dr. Pettus, “[Do] you want to caution other doctors not to just use [artificial intelligence] as a default?”

“I don’t think we should ever do that,” he replied. “I personally think it’s best to have it as a secondary safety measure after we’ve done all of our work.”