Cervical carcinoma (CC) is the fourth most common cancer type and cause of cancer mortality in women, with an estimated 604,000 new cases and 342,000 deaths globally in 2020 alone [1].International Federation of Gynecology and Obstetrics (FIGO) staging represents the most widely used grading system for CCs, and has demonstrated a role in not only treatment guidance, but also prognosis prediction [2,3]. According to current guidelines, surgery is considered the primary treatment of choice for early and localized lesions (FIGO stages IB1, IB2, and IIA1), while concurrent chemoradiotherapy is indicated in the presence of locally advanced or node-positive lesions (FIGO stages IB3 and ≥ IIA2) [4]. Accurate clinical staging is thus vital for the treatment planning of this disease [5].

However, compared with surgical staging, clinical staging remains inferior in terms of accuracy. Error rates as high as 32% and 65% have been reported for stage IB and III diseases [6]. At present, the clinical staging of cervical cancer is mainly based on the gynecologist's pelvic examination before treatment, and also depends on the experience of the gynecologist. It is not accurate if the patient has pelvic inflammatory disease, endometriosis or obesity. At the same time, some limitations of FIGO preoperative staging depend on physical examination and imaging examination, and there are subjective factors, resulting in the classification method is not objective enough [7].

Conventional MRI, with its high resolution for soft tissues, has been reported as an alternative preoperative staging technique; however, its limitations have been reported [2,8]. For example, the accuracy of two-thirds of the vagina and the assessment of parametrial invasion both rely on standardized scans parallel or perpendicular to the long axis of the cervix [9,10].In particular, reactive tissue changes such as peritumoral edema can lead to an increase in T2-weighted signals, resulting in an overestimation of the lesion size and thus the over-staging of the disease [3] .Inadequacies in the sensitivity of this approach for micrometastatic lymph nodes have also been reported [11,12]. Altogether, these highlight the need for a non-invasive, objective, and effective staging method for CC [13].

Radiomics is an emerging field demonstrating the ability in extracting quantitative imaging features beyond the capabilities of even a trained naked eye. Such an advancement has allowed for a shift from qualitative image interpretation to a more objective and quantitative approach [[14], [15], [16]]. T2-weighted imaging is known to provide detailed information on tumor morphology and degree of stromal invasion, while contrast-enhanced T1-weighted imaging is useful for vascular distribution and tumor localization. The superiority of CE-T1W imaging in terms of tumor localization, tumor margin, and contrast-to-noise ratio has been demonstrated for the assessment of CCs [17].

The aim of our study was thus to develop and validate a multiparametric MRI-based radiomics model for the prediction of FIGO stage in CC patients.