Ultrasound and magnetic resonance imaging can be helpful for the detection and characterization of placenta accreta spectrum (PAS) disorders, but both imaging modalities suffer from limitations at least partly due to the subjective, qualitative evaluation of medical images. Radiomics is regarded as a possible solution to increase the value of imaging in this setting, with recent evidence supporting this hypothesis. We aimed to systematically review the literature on radiomics applications for patients with PAS disorders and evaluate the methodological quality of these radiomics studies.

We defined a search string and run it on different databases to identify studies on human subjects published in English. To qualitatively summarize the findings of selected studies, we retrieved details on study design (e.g., prospective or retrospective), aim, patient population (e.g., sample size), imaging modalities and radiomics workflow (e.g., segmentation and feature extraction approaches). We appraised the methodological quality of selected works using the Radiomics Quality Score (RQS).

In total, we identified and analyzed 10 articles. All were retrospectively designed and MRI-based. Most had more than 100 patients (6/10). There were four prognostic (i.e. prediction of bleeding volume or of appropriate management) and six diagnostic (PAS vs not PAS classification) studies. The highest RQS score was of 17/36, while the lowest was of -6/36, with a median of 8. Commonly encountered methodological concerns were the absence of feature robustness testing to repeated segmentation as well as lack of data openness.

The methodological quality of radiomics studies focused on PAS disorders is heterogeneous. Additionally, current literature fails to address relevant research questions. Further, more robust research is needed to establish the possible role of radiomics in the management of patients with PAS disorders.