Radiology turnaround time

Ask a centre owner what their turnaround time is and you will usually get one number: "CT reports in two hours." Ask the referring doctor and you will get a different number. Ask the patient, who has been sitting in the waiting area since 9 am, and you will get a third. All three are measuring something real. They are just starting the clock in different places, and that is the first thing to fix if you want TAT to come down.

Our co-founder Dr. Vishwanath R S puts the stakes plainly in The real total cost of radiology software:

So, let's break turnaround into its stages, look at where the hours actually go in an Indian centre, decide what to measure, and then go through the quick wins.

Turnaround is a chain, not a number

The AHRA's benchmarking task force made this point well in a 2025 piece on defining TAT. It lists several common definitions, from order to final report, exam completion to final report, exam completion to preliminary report, and the often-ignored gap between the technologist ending the exam and the images being available to read. Each one tells you something different.

For an Indian imaging centre, we find it most useful to split the journey into seven stages.

Turnaround is seven stages; most centres measure only the middle three.

Most centres only measure stages 4 to 6, if they measure at all. The hours often hide in the others.

Where the hours actually go

What to measure

You cannot fix what you do not timestamp. Most of the timestamps already exist in a RIS-PACS; someone just has to pull them. A useful TAT measurement is characterised as follows.

1. It measures every stage, not only the total. A single "TAT" hides which stage is broken.
2. It uses the median, and the share within target. The AHRA piece recommends tracking median TAT along with the percentage of cases meeting the goal. The median is not distorted by one case that waited overnight; the percentage tells you how many patients you failed.
3. It splits by modality and priority. Emergency CT, inpatient MRI and outpatient X-ray should each have their own target. An average across all of them means nothing.
4. It splits by time of day. Many centres look fine at 11 am and fall apart after 8 pm.
5. It splits by centre. In a chain, one branch can drag the whole number.
6. It is paired with quality. The same AHRA article warns that TAT incentives should be balanced with measures such as callback rates, peer review and reporting error rates. Fast and wrong is not an improvement.

We deliberately are not giving benchmark numbers here. Targets should come from your own baseline and your referrers' needs, not from a figure in a blog post.

Quick wins

These are the changes that usually move the number in weeks, not months.

1. Auto-send from every modality. Configure each scanner to send images to PACS on completion, not in batches. Check it on every machine, including the old X-ray CR reader in the corner. Stage 3 · Image transfer
2. Alert on missing images. If a study is marked complete in the RIS but no images arrive within a set time, someone should get an alert. Stage 3 · Image transfer
3. One worklist across centres. If radiologists cover more than one site, put every study in one queue, sorted by priority and waiting time. See running a diagnostic chain on one system. Stage 4 · Queue wait
4. Get the history at registration. A two-line clinical history captured at the front desk saves a phone call later. Make it a required field for CT and MRI. Stage 1 · Registration
5. Prefetch priors. Trigger prior retrieval when the study is registered, so it is ready when the radiologist opens the case. More on this in Priors on screen. Stage 5 · Read
6. Cut the typing loop. Direct dictation with speech recognition, or AI-assisted drafting, removes the wait between radiologist and typist. (We describe one way of doing this in Trinity Reporting Engine.) Stage 6 · Finalisation
7. Send reports automatically. When a report is signed, send it, or a link to it, to the referring doctor straight away. Stage 7 · Dispatch
8. Put TAT on a screen. A live view of what is waiting, for how long, and which cases are close to breaching, changes behaviour faster than a monthly report. Measurement

How we approach it

SOMATIQ runs the whole cycle on one platform, from registration, scheduling and billing through acquisition and reporting to report delivery. That matters for TAT mainly because the timestamps for every stage live in one system rather than three, so the table above can be filled from real data instead of guesswork. Administrators can track daily scan volumes and radiologist TAT and act on SLA breaches before they escalate.

Several of the quick wins map directly onto the platform. A modality worklist means the technician does not retype patient details at the scanner; a smart scheduler with reminders helps with the 11 am slot that starts at 11:40; built-in speech-to-text and live co-reporting between radiologists and transcriptionists shorten the finalisation loop. Radiologists can talk to the worklist and move through it hands-free, and once a report is signed, a study link can go to the referring doctor on WhatsApp, with reports also delivered to patients by email and WhatsApp through the patient portal.

Prima Diagnostics in Bengaluru runs all eight of its centres on the platform, and reports that its radiologists are reporting nearly 30% faster, as Digital Health News reported. That gain did not come from anyone reading faster. It came from what disappeared around the reading: hunting for the right scan, chasing previous reports, switching between applications, retyping patient details and waiting for images from another branch. In the terms of the table above, most of it sits in stages 4 to 6.

None of this replaces the basics above. A centre with batch-sending modalities and a handwritten registration register will have slow TAT on any software. The practical upshot is to treat turnaround as seven clocks rather than one: timestamp each stage, find the slowest, and fix that first. In most centres it is not the radiologist's reading time but the waits on either side of it.

Related reading

• Running a diagnostic chain on one system — the central worklist that removes most queue wait.
• Priors on screen — prefetching so the read is not held up.
• Sharing scans with referring doctors — closing the dispatch stage securely.

FAQ