Imaging Center Call Center: What AI Voice Agents Fix

An imaging center call center is not a scheduling desk with a phone attached. It is four different jobs pointed at the same queue: taking referrals from ordering offices, confirming coverage before a scan, walking patients through prep, and chasing the paperwork that has to be on file before the appointment counts.

Every one of those calls has a deadline attached to a scanner slot. Miss the coverage check and the scan happens without authorization. Miss the prep call and the patient shows up having eaten breakfast. Either way the slot is gone, and a single MRI hour is worth more than the whole staff cost of the calls that protect it.

The four call types, and why they collide

Referral intake comes from other offices, usually by fax or portal, and usually incomplete. Somebody has to call back for the missing indication, the ordering provider’s NPI, or the right body part. That call is administrative but it blocks everything downstream.

Coverage and authorization calls run on payer clocks. Advanced imaging is one of the most heavily managed categories in outpatient care, and CMS has run a program specifically around appropriate use criteria and referral documentation for advanced imaging services (CMS). The center is holding the bag on paperwork the ordering office generates.

Prep calls are simple and unforgiving. NPO windows, contrast instructions, metal screening, arrival times. The content rarely changes. The volume never stops.

Reminder and reschedule calls absorb whatever hours are left. When they do not happen, the no-show rate climbs and the schedule develops holes nobody can backfill on short notice.

The collision happens because these queues share one team. On a busy Monday, prep calls lose to authorization calls, and the center trades a preventable no-show for a preventable denial.

What gets automated and what does not

Pretty Good AI builds voice agents for the administrative layer of that queue, integrated with athenahealth. The line is firm. The agent handles logistics. It does not decide whether a scan is appropriate, it does not interpret a result, and it does not answer questions about what a finding means. Those go to a human, every time.

On inbound calls, the agent identifies whether the caller is a patient, a referring office, or a payer, and routes accordingly. It captures referral details, checks eligibility and benefits, and files what it collects into athenaOne so staff are not rekeying from a notepad.

On outbound calls, it runs prep instructions for scheduled scans, confirms arrival windows, verifies the patient still has the ordered study and the right site, and calls back when a coverage check comes back pended. Anything that turns clinical, meaning a question about symptoms or about what a scan will show, is handed to staff with the context already captured.

That handoff is the part most centers underweight. A patient who asks a clinical question and gets stalled by an automated system will remember that, not the twelve smooth calls before it.

The economics of a protected scanner hour

Imaging is a fixed-cost business with a variable schedule. The scanner, the tech, and the lease are paid whether the slot fills or not, so every avoidable gap is close to pure margin loss.

Take a center running two MRI scanners at 10 slots a day each. A 10 percent no-show and cancellation rate is two slots a day, roughly 500 slots a year. At a conservative average net of $400 per study, that is $200K in unbilled capacity. Cutting that rate by even a third pays for a lot of automation.

Denials add the second layer. Insurers deny a meaningful share of in-network claims and consumers appeal very few of them (KFF). For an imaging center, an authorization gap caught before the scan costs a phone call. Caught after the scan, it costs the study.

Staffing is the third. Front-office and patient access roles are among the hardest to keep filled across health care, and turnover in those seats resets institutional knowledge every few months (MGMA). Automation does not fix hiring, but it makes the desk survivable when a seat is open.

How to phase it in

Start with outbound prep and reminder calls. The script is stable, the risk is low, and the effect on the no-show rate shows up inside a month. That gives you a clean before-and-after number to argue with.

Add eligibility and benefit checks next, because they run on a schedule you already know, and the write-back into athenaOne is what saves the staff time rather than the call itself.

Move inbound routing last. Inbound is where callers are least predictable and where a bad handoff does the most damage to the referring relationship. Get the escalation path right before you point volume at it.

Measure three things at each stage: percentage of scheduled scans with a completed prep contact, percentage with coverage confirmed before arrival, and average hold time during the 8am referral rush. If those three move, the queue is healthier. If they do not, the tool is decorating the problem.

Key Takeaways

• Treat the imaging queue as four distinct workloads. Automating the wrong one first produces motion without margin.
• Prep and reminder calls are the safest starting point and the fastest to show a measurable drop in empty slots.
• Confirm coverage before the scan, not after. A pre-scan phone call is cheap, a post-scan denial is the whole study.
• Require write-back into athenaOne so eligibility and referral details land in the chart instead of a queue somebody has to re-enter.
• Define the escalation path first. Any call that turns clinical goes to a human with the context already captured.

An imaging center call center succeeds on throughput, not on heroics. The centers that run clean are the ones where the boring calls, prep and coverage and callback, happen every single time, so the staff still have hours left for the calls that need a person.

Sources

• CMS, Appropriate Use Criteria Program: https://www.cms.gov/medicare/quality/appropriate-use-criteria-program
• KFF, Claims Denials and Appeals in ACA Marketplace Plans: https://www.kff.org/private-insurance/issue-brief/claims-denials-and-appeals-in-aca-marketplace-plans/
• MGMA Stat, Stabilized but Not Solved: Staff Turnover in Medical Practices in 2026: https://www.mgma.com/mgma-stat/stabilized-but-not-solved-staff-turnover-in-2026