Skip to main content
Skip to main content
Back to Grand Rounds
Grand RoundsWeekly Evidence Brief

Radiology

Edition

30-Second Takeaway

  • Radiomics shows promising lesion- and patient-level prognostic value but needs external validation before routine use.
  • EMR-integrated risk assessment can substantially increase identification of patients meeting genetic counseling criteria.

Week ending May 2, 2026

Radiomics, pragmatic implementation, and consent/workflow trials with direct clinical implications for imaging and research operations

Radiomics predicts survival and lesion response after irinotecan-TACE for colorectal liver metastases.

EUROPEAN RADIOLOGYApr 29, 2026

In 76 patients (176 lesions) from the prospective CIREL registry, baseline intensity radiomic features predicted overall survival (external AUC 0.79). Delta radiomics between baseline and first follow-up best identified lesion-level response (AUC 0.83). Integrating imaging with laboratory values improved lesion-response assessment (AUC 0.86) but did not improve overall survival prediction. These models stratified patients into distinct OS risk groups (median survival low-risk 696 days vs high-risk 453 days).

Radiomics for predicting early response to ablative radiotherapy in oligometastatic disease: heterogeneous evidence.

EUROPEAN RADIOLOGYApr 30, 2026

This scoping review included 29 studies (3946 patients) and reported model AUCs ranging 0.69–0.95 for ART response prediction. Deep learning approaches achieved higher accuracies (reported AUCs up to 1.00) but most studies focused on brain metastases. Methodological quality was moderate with limited external validation and heterogeneous protocols. The evidence is promising but not ready for routine clinical implementation without prospective multicenter validation.

Audiovisual augmentation of electronic consent did not increase consent or comprehension.

JAMA NETWORK OPENApr 30, 2026

In a randomized trial of 1535 participants across four consent modalities, overall consent rate was 57.9% with no modality differences. Consent rates were similar: text-only 58.2% versus physician video 62.4%, patient video 53.0%, animated video 57.8%. Among those who completed the comprehension survey, 86.4% demonstrated understanding, with no modality advantage. Audiovisual enhancements did not improve enrollment or comprehension in this virtual, single-center setting.

References

Numbered in order of appearance. Click any reference to view details.

Additional Reads

Optional additional studies from this edition.

Edition context

Clinical signal

  • For radiomics: consider use in research or multidisciplinary selection discussions, not yet for sole clinical decisions.
  • When adopting EMR risk tools, plan workflows to convert identified patients into counseling visits.
  • In rural trial implementation, pre-specify local champions and simplify technology requirements.