RadNet Inc. lost an estimated $25 million on its artificial intelligence segment last year, but sees profitability ahead for the burgeoning business line, leaders said Tuesday.

The Los Angeles-based imaging center operator has made a big play in AI recently, including buying two such firms last year for $100 million. Despite early losses, RadNet is seeing “significant strides” in commercializing its AI products, CEO Howard Berger, MD, told investors during the company’s fourth quarter earnings call.

In November, the publicly traded company started implementing its new Enhanced Breast Cancer Diagnostic, or EBCD, offering AI-augmented breast cancer screening to patients for an additional fee. Berger expects all mammography centers to offer the AI add-on by summer 2023, but it will take a little bit longer for the program to operate in the black.  

“Based upon early adoption data from several of our East Coast markets, we are anticipating our losses from AI in 2023 to significantly narrow as a result of EBCD revenue, and we project our AI segment to be profitable in 2024,” Berger said in a statement.

In the meantime, RadNet said it continues to see promising results across its imaging center segment, which gathered more than $1.4 billion in revenue in 2022. When excluding any losses from AI, the company recorded roughly $209 million in 2022 earnings (before interest, taxes, depreciation and amortization). Year-over-year revenues were up roughly $112 million (8.5%), while adjusted earnings fell by $2.9 million (1.4%). Net income totaled $10.7 million, down $14.1 million year-over-year, with the $24.9 million pre-tax loss on AI (with revenue of $4.4 million) a key factor.

“Moving into 2023, the demand for diagnostic imaging remains robust and is growing,” Berger said in the statement, estimating that RadNet has a dozen-plus new centers currently in development.

Until the AI segment turns a profit, RadNet said it will report financial results separately from its imaging centers. Berger said he expects this to continue through 2023, “providing transparency for our stakeholders to track our progress.” In its 2023 guidance, the company said it expects to earn between $16 million and $18 million in revenue from AI, with an anticipated loss of $9 million to $11 million (after adjusting for taxes and other factors).

RadNet bills itself as the “leading” operator of freestanding, fixed-site diagnostic imaging services. All told, the company has a network of 357 centers in markets including California, Maryland, Delaware, New Jersey, New York, Florida and Arizona. Its AI acquisitions, meanwhile, have included Aidence (focused on solutions for pulmonary nodule management and lung cancer screening), Quantib (specializing in prostate cancer and neurodegeneration), and DeepHealth (breast cancer detection).