# Preoperative assessment of tumor size by MRI and ultrasound in cervical cancer: a large-scale retrospective comparative study

## TL;DR

This study compares MRI and ultrasound for measuring cervical cancer tumor sizes before surgery, finding both methods overestimate and have similar accuracy for staging.

## Contribution

A large-scale retrospective analysis comparing MRI and ultrasound accuracy for preoperative cervical cancer tumor size assessment.

## Key findings

- MRI had a smaller mean measurement bias (0.61 mm) compared to ultrasound (1.50 mm).
- Both MRI and ultrasound showed comparable agreement rates (67.6% and 65.8%, respectively) with postoperative pathology for tumor size categorization.
- Tumors larger than 40 mm measured by either modality were significantly associated with large measurement discrepancies (>10 mm).

## Abstract

Accurate preoperative evaluation of tumor sizes is essential for guiding optimal treatment planning in cervical cancer. This study aimed to compare the accuracy of preoperative tumor size measurement between magnetic resonance imaging (MRI) and ultrasound.

A retrospective study was performed involving 925 patients diagnosed with cervical cancer who underwent primary surgical treatment between January 2020 and June 2025. The accuracy of these two imaging modalities was assessed by comparing their measurements to the maximum tumor diameter determined through postoperative pathological analysis.

The Bland–Altman analysis showed that both ultrasound (mean difference: 1.50 mm) and MRI (mean difference: 0.61 mm) overestimated tumor size. In the paired subgroup of 757 patients who underwent both imaging modalities, the agreement rates between imaging and pathology for categorizing tumors into size groups were 65.8% for ultrasound and 67.6% for MRI (p = 0.360). Although MRI showed a significantly smaller mean measurement bias than ultrasound (0.73 mm vs. 1.37 mm; p = 0.012), the proportion of large errors (> 10 mm) was not significantly different. Multivariate analysis indicated that tumors > 40 mm assessed by ultrasound (OR = 2.85) or MRI (OR = 2.72) were significantly associated with increased likelihood of measurement discrepancies > 10 mm.

While MRI exhibited a lower measurement error compared to ultrasound, both modalities showed comparable performance in tumor size staging. Furthermore, for tumors exceeding 40 mm in diameter as determined by preoperative imaging, clinicians are advised to integrate clinical examination to enhance the accuracy of staging.

The online version contains supplementary material available at 10.1007/s00404-026-08304-7.