Abstract

In suspected Cauda Equina Syndrome (CES), MRI is the reference standard for diagnosis and surgical decision-making, but access may be delayed or contraindicated in emergency care. CT is more readily available, motivating interest in CT-to-MRI translation as a potential clinical support tool. However, in high-risk settings, synthetic images must be evaluated beyond visual similarity alone. They must preserve clinically relevant anatomy and pathology, and remain consistent given identical generation conditions. In this pilot study, we evaluate five representative CT-to-MRI translation paradigms for emergency spine imaging through three axes: plausibility, clinical trust, and determinism. Plausibility is assessed using conventional quantitative metrics and semantic representation-level analysis. Clinical trust is assessed through blinded radiologist evaluation of image quality, anatomy, pathology, and confidence. Determinism is assessed using repeated-generation coefficient-of-variation analysis. Our findings show that favourable quantitative performance does not consistently translate into stronger clinician-derived assessment, and that repeat-generation stability provides a separate reliability indication. These results support multi-dimensional, clinically informed evaluation of generative medical imaging systems before further clinical translation.

Links to Paper and Supplementary Materials

Main Paper (Open Access Version): https://papers.miccai.org/miccai-2026-sat/paper/AMAI_027.pdf

SharedIt Link: Not yet available

SpringerLink (DOI): Not yet available

Supplementary Material: Not Submitted

Link to Open Review

Open Review Page: https://openreview.net/profile?id=%7EPaul_Banahan1

BibTex

@InProceedings{BanPau_Determinism_MICCAISAT2026,
        author = { Banahan, Paul AND Mathur, Prateek AND Treacy, Molly Godson AND McDermott, Edward AND Kyle, Eimear AND Caldwell, David AND Lawlor, Aonghus AND MacMahon, Peter},
        title = { { Determinism, Plausibility, and Clinical Trust in CT-to-MRI Translation for Emergency Spine Imaging } },
        booktitle = {Medical Image Computing and Computer Assisted Intervention -- MICCAI 2026 Workshops and Challenges},
        year = {2026},
        publisher = {Springer Nature Switzerland},
        volume = {LNCS 17273},
        month = {pending},
        page = {pending}
}


back to top