Spectral CT separated accessory spleen from pancreatic tumour on one phase

Design

Retrospective single-centre diagnostic performance study

Population

85 patients with hypervascular pancreatic lesions on spectral CT

Primary outcome

Discrimination of accessory spleen from neoplasm by spectral metrics and visual reading

Effect

Effective-atomic-number area under the curve 0.97 to 0.99; reader accuracy 88% vs 61%

An intrapancreatic accessory spleen in the pancreatic tail can mimic a hypervascular neoplasm, and resolving it often needs extra imaging such as scintigraphy. This retrospective study tested whether spectral CT on a single portal-venous phase could make the call in 85 patients (34 accessory spleens, 51 neoplasms).

The accessory-spleen lesions tracked the spleen's spectral signature: relative lesion-to-spleen ratios were lower than in neoplasms across every image series. Effective-atomic-number and iodine maps discriminated best, with areas under the curve around 0.97 overall and up to 0.99 for pancreatic-tail lesions. Adding spectral imaging to conventional review lifted radiologists' accuracy from 61% to 88% and markedly improved agreement between readers.

For a department with spectral CT, this offers a way to settle an indeterminate hypervascular tail lesion on the scan already acquired, avoiding further tests. It is retrospective and single-centre, so it points to a useful application rather than a validated protocol.

• Retrospective study of 85 patients: 34 intrapancreatic accessory spleens, 51 hypervascular neoplasms.
• Iodine and effective-atomic-number maps discriminated best, with areas under the curve around 0.97 to 0.99.
• Spectral imaging raised reader accuracy from 61% to 88% over conventional CT alone.
• Inter-observer agreement improved markedly (ICC 0.93 vs 0.65).
• It uses the single portal-venous phase already acquired, potentially avoiding further workup.

Why it matters

It can resolve a common incidental mimic on the scan already in hand, sparing a patient scintigraphy or biopsy.

Don't overread it

This was retrospective and single-centre; it shows promise on existing scans rather than a validated diagnostic threshold.

The statistics, in plain English

An area under the curve near 0.97 indicates strong discrimination; the jump in reader accuracy from 61% to 88% is large, but as a retrospective single-centre study the numbers may be optimistic compared with routine practice.

Read the rest in the app

You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.