“We Want to Utilize AI in the Workflow to Bridge That Gap,” Says RadNet Eastern Operations CEO

“There’s a pretty severe shortage of technologists and radiologists that we need for every exam, and at the same time there’s an increased demand for imaging services,” said Steve Forthuber, President and CEO of Eastern Operations at RadNet to AIM Media House. According to numbers reported by The 2025 ASRT Radiologic Sciences Staffing and Workplace Survey, vacancy rates near record highs, including 19.4% in computed tomography and 17.4% in magnetic resonance imaging.

AI adoption in radiology has accelerated during the same period. Independent reporting indicates that roughly two-thirds of U.S. radiology departments now use AI tools in some capacity, with adoption rates roughly doubling since 2019. Still, many deployments of AI remain limited to point solutions rather than full workflow integration.

RadNet says it performs more than 10 million outpatient imaging procedures annually across more than 430 centers in the United States. In recent months, the company expanded into Indiana and Southwest Florida through acquisitions of regional imaging practices and centers. AI-powered workflow tools are to be introduced into acquired centers. “We want to utilize AI… in the workflow to try and bridge that gap,” he said.

AI Deployment Across a Constrained Workforce

Technologist shortages have been cited nationally as a barrier to expanding MRI, CT, and ultrasound access. RadNet’s DeepHealth subsidiary received FDA 510(k) clearance for TechLive, a remote scanning solution that enables centralized operation and supervision of MRI, CT, PET/CT, and ultrasound procedures. Company materials state that more than 300 imaging systems are connected to TechLive.

In a pilot across 64 centers in RadNet’s New York market, DeepHealth reported a 42% decrease in MRI room closure hours over one year. The company attributed the change to expanded remote technologist coverage and scheduling optimization. These figures are company-reported metrics.

Forthuber described the integration process following acquisitions. Data migration and IT connectivity come first, followed by pilot deployments. Radiologists and technologists evaluate workflows before broader rollout. He said integration typically occurs within four to six months of closing.

“We don’t come in and force it on them,” he said.

Forthuber said that dataset is used internally to refine and redeploy AI tools. Radiologists retain final responsibility. AI assists with measurements, prioritization, and reporting.

Forthuber also described AI-enabled workflow changes as affecting career pathways inside imaging centers. He cited examples where patient service representatives transition into in-suite assistant roles that support remote technologists, and later pursue certification in advanced modalities.

“That is a lifechanging opportunity for that individual who went from, you know, 20-some dollar an hour job to a 40 or $50 an hour job,” he said. He sees the transition as a shift in role composition rather than role elimination, with routine tasks reduced and clinical responsibilities expanded for staff who pursue additional training.

Earlier Detection and Stage Shift

RadNet has deployed an AI-supported breast screening workflow called Enhanced Breast Cancer Detection (EBCD). The system flags discordance between the interpreting radiologist and the AI model. If disagreement occurs, a second radiologist reviews the case before the initial reader makes the final determination.

According to RadNet and its DeepHealth subsidiary, the workflow increased overall breast cancer detection rates by 21% and by 23% in women with dense breast tissue. Peer-reviewed analyses show that treatment costs increase substantially with advancing cancer stage. Earlier-stage detection is associated with lower treatment expenditures and improved survival outcomes.

In lung cancer screening, the National Health Service in England reported that approximately three-quarters of cancers identified through its Targeted Lung Health Check program were detected at an early stage during pilot phases. DeepHealth lung AI tools are used within portions of that ecosystem.

“Anytime you can deal with disease earlier… you’re going to radically reduce the cost of care and most importantly you’re going to create much better outcomes for the patient,” Forthuber said.

Company leadership has stated that image-based AI tools are being developed for additional screening categories, including prostate and cardiac imaging. Regulatory and reimbursement requirements determine how autonomous systems can be deployed. Current FDA-cleared imaging AI tools function primarily as assistive technologies.

Other imaging providers and health systems are expanding AI deployments, often through vendor partnerships. Hospital networks are integrating third-party AI tools into PACS environments, and healthcare technology companies have announced partnerships to deploy AI-enabled imaging platforms for early disease detection. GE HealthCare has announced plans to acquire imaging software company Intelerad, reflecting investment in cloud-based imaging infrastructure and analytics.

RadNet’s model combines an outpatient imaging network with internally developed AI tools and standardized rollout processes across newly acquired centers.