Do MRI structured reports with FIGO classifications of leiomyomas contain adequate information for clinical decision making?

LT, JA, PC, PM, SD, and FA contributed to the conceptualization; LT, HA, SD, and FA, contributed to the methodology; LT, AA, SS, JA, SD, and FA collected the data; LT, HA, and FA contributed to formal analysis and to writing the orginal draft; LT, HA, AA, SS, JA, PC, PM, SD, and FA performed the reviewing and editing; LT, HA, AA, SS, JA, PC, PM, SD, and FA visualized the study; and it was supervised by LT, PC, PM, SD, and FA.

Out of the total of 1195 pelvic MRI scans identified, a subset of 60 MRI scans belonging to 60 individual patients was randomly selected. Within this group, 20 were randomly selected for inclusion in the NR category. Among the remaining 40 MRI scans, 20 were assessed using the SR template, while the other 20 were evaluated with the SR‐FIGO template. There were notable variations in the number of KF (median 16, interquartile range [IQR] 12–17) across the reports ( P  < 0.001). SR‐FIGO exhibited the highest number of KF (median 17, IQR 16–18), followed by SR (median 16.5, IQR 15.25–17) and NR (median 9, IQR 8–12). On pairwise comparisons, significant differences were observed in the number of KF between NR and SR ( P  = 0.001) and between NR and SR‐FIGO ( P  = 0.001), but no significant difference was found in the number of KF between SR and SR‐FIGO ( P  = 0.063). Several KF, such as the description of the uterine position and the junctional zone, number of leiomyomas, and their enhancement, were notably more prevalent in SR and SR‐FIGO compared with NR. However, in pairwise comparisons, no significant differences were observed between SR and SR‐FIGO ( P values between 0.271 and 0.897) (Table  1 ; Figure  2 ). Axial T1‐weighted postcontrast fat‐saturated image in a 34‐year‐old woman with pelvic pain found to have multiple leiomyomas, including a non‐enhancing subserosal leiomyoma with less than 50% intramural component, classified as a FIGO Stage 6 leiomyoma according to the FIGO (the International Federation of Gynecology & Obstetrics) PALM‐COEIN leiomyoma subclassification system. Non‐enhancing leiomyomas are less likely to respond to uterine artery embolization. As such, it was deemed relevant to include in the magnetic resonance imaging structured reports whether leiomyomas were enhancing. No significant difference in answers to questions 1 to 5 among the three different reports was identified for the gynecologist ( P values between 0.055 and −0.344). However, for the gynecologist‐in‐training, significant disparities in responses to questions 3 and 5 were noted among the three types of reports ( P  = 0.015 and P  = 0.004 respectively), with pairwise comparisons revealing significant differences only between NR and SR‐FIGO for questions 3 ( Does the report give you enough information to decide on the type of surger y?) and 5 ( If reviewed, how much time (in minutes) was required to review the images? ) ( P  = 0.010 and P  = 0.009 respectively), but no significant difference between NR and SR and between SR and SR‐FIGO ( P values between 0.102 and 0.691). When the responses of the gynecologist and gynecologist‐in‐training to each question were compared, there was statistically significant difference for question 1 among the three types of reports ( Does the report provide enough information for clinical decision making of the management of the patient? ) ( P  < 0.001), with significant difference on pairwise comparison for SR ( P  = 0.007) and SR‐FIGO ( P  = 0.024), with the gynecologist deeming SR and SR‐FIGO to provide enough information for clinical decision making significantly more commonly than the gynecologist‐in‐training (Table  2 ). No significant difference was observed between the gynecologist and gynecologist‐in‐training's responses to questions 2 to 5 ( P values between 0.201 and 0.761). Comparing their confidence scores when answering these questions, there was no difference across the three different types of MRI reports ( P values between 0.076 and 1 for the gynecologist and between 0.186 and 0.733 for the gynecologist‐in‐training). Similar results were observed when the median confidence score for each question was compared between gynecologist and gynecologist‐in‐training ( P values between 0.376 and 0.654) except for question 4 ( Is review of images required? ), with gynecologists showing higher median confidence ( P  < 0.001; gynecologist: median 5, IQR 5; gynecologists‐in‐training: median 4, IQR 4–5). Comparison of gynecologist and gynecologist‐in‐training answers for narrative reports, structured reports, and structured reports with PALM‐COEIN FIGO subclassification of leiomyomas. a Abbreviations: NR, narrative reports; SR, structured reports; SR‐FIGO, structured report with PALM‐COEIN FIGO (the International Federation of Gynecology & Obstetrics) subclassification of leiomyomas. Data are presented as n for gynecologists/ n for gynecologists in training. On pairwise comparison of SR, P  = 0.007. On pairwise comparison of SR‐FIGO, P  = 0.024.

Uterine leiomyomas are a common condition affecting women, often requiring medical, surgical, or interventional management because of associated symptoms and complications. Our study aimed to evaluate the effectiveness of structured MRI reporting, specifically incorporating SR‐FIGO, in comparison to NR, to ascertain whether these templates provide sufficient information for clinical decision making. In our study, structured reports (SR or SR‐FIGO) consistently included KF more frequently than NR. This observation aligns with previous studies that have shown structured reporting templates to improve the incorporation of pertinent information in radiology reports. For instance, Franconeri et al. 7 demonstrated that implementing a structured template for MRI reporting of uterine leiomyomas led to more comprehensive reports with a greater inclusion of KF. Our results corroborate these findings within the context of MRI reporting for uterine leiomyomas, suggesting that structured templates contribute to the thoroughness of reports. Notably, the gynecologist‐in‐training required significantly more time to review the MRI images for the NSR and faced difficulty in determining the type of surgery based solely on NSR. This observation resonates with existing literature, which underscores the challenges posed by non‐structured reports, potentially leading to prolonged interpretation times and potentially inadequate information for clinical decision making. Smith et al. 12 observed similar trends in abdominal imaging, associating NSR with extended interpretation times and reduced confidence in decision making. Our study expands upon these findings to the realm of uterine leiomyoma MRI reporting, underscoring the advantages of structured templates in facilitating efficient and informed decision making. The comparison between the answers of the gynecologist and the gynecologist‐in‐training revealed that although the experienced gynecologist felt that they had sufficient information for clinical decision making with a structured report, the gynecologist‐in‐training did not find either SR or SR‐FIGO reports adequately informative. This may be due to the greater ability of the experienced gynecologist to extract relevant information from the MRI reports compared with the gynecologist‐in‐training. This finding, although in a different domain, is somewhat comparable to what was observed in a study by Alessandrino et al. 13 on MRI reports of patients with multiple sclerosis, in which less experienced neurologists were less capable of obtaining valuable information for clinical decision making from non‐structured reports and had to evaluate the images more frequently compared with experienced neurologists. We found a significant difference between the responses to question 1 ( Does the report provide enough information for clinical decision making of the management of the patient? ) between the gynecologist and gynecologist‐in‐training with the gynecologist deeming SR and SR‐FIGO to provide enough information for clinical decision making significantly more commonly than the gynecologist‐in‐training. Although the present study demonstrates that SR of MRI scans in patients with leiomyomas provides more information compared with NR, it remains unclear whether inclusion of the PALM‐COEIN FIGO classification of leiomyomas in the SR influences the gynecologist's clinical decision making ability. Further studies with larger sample size and longitudinal follow ups are warranted to ascertain if the PALM‐COEIN FIGO classification enhances patient management and outcomes. A study by Brown et al. 14 used patient outcomes such as recurrence rates and treatment success to evaluate the effectiveness of structured reporting in prostate imaging. Applying similar methodologies in uterine leiomyoma imaging could provide valuable insights into the clinical impact of structured reporting. The study was limited by its retrospective design, small sample size, and reliance on subjective evaluations of MRI reports. Future research incorporating objective measures and assessing patient outcomes post‐treatment could provide a more comprehensive understanding of the impact of structured MRI reporting on leiomyoma management. Our study highlights the advantages of structured MRI reporting, particularly when incorporating the PALM‐COEIN FIGO classification, in providing a more comprehensive evaluation of uterine leiomyomas. These structured templates offer a standardized approach, aiding clinicians in treatment planning and decision making. Our findings are consistent with existing literature on the benefits of structured reporting in radiology, demonstrating its applicability to uterine leiomyoma imaging. Future studies should explore the impact of structured reporting on patient outcomes and the potential for enhanced inter‐rater agreement with the use of standardized templates. As the PALM‐COEIN FIGO classification gains prominence in leiomyoma management, incorporating it into structured MRI reports could further improve clinical decision making in this realm, especially for obstetricians/gynecologists‐in‐training. In conclusion, the present study demonstrates the potential benefits of incorporating the PALM‐COEIN FIGO classification into structured reporting, but further research is needed to determine its impact on clinical decision making and patient outcomes.

Uterine leiomyomas are the most common tumor of the female reproductive tract with estimates suggesting that up to 70%–80% of women will develop uterine leiomyomas by age 50 years. 1 , 2 These benign growths are associated with a spectrum of symptoms, including abnormal uterine bleeding, dysmenorrhea, bulk symptoms, and even infertility, making them a considerable cause of morbidity among affected individuals. Despite their common occurrence, effective management of fibroids remains a challenge, often necessitating medical, surgical, or interventional treatments. 3 Ultrasound serves as the initial imaging modality to diagnose leiomyomas in symptomatic patients, but magnetic resonance imaging (MRI) is the preferred tool for treatment planning because it allows for better characterization of the number, location, and types of leiomyomas. In particular, MRI's capability to accurately identify submucosal fibroids and discern alternative pelvic pathologies underscores its indispensable role in clinical decision making. 4 The FIGO (the International Federation of Gynecology & Obstetrics) PALM‐COEIN classification system was developed in 2011 and updated in 2018 to standardize description and classification of uterine leiomyomas, which is essential for treatment planning. 5 , 6 Despite the existence of structured reporting templates in various medical institutions, a universally accepted structured MRI template specifically tailored for reporting uterine leiomyomas is currently lacking. 7 , 8 Structured reports are increasingly used in various fields of radiology, potentially increasing standardization and providing relevant information that will aid in clinical decision making. 9 , 10 , 11 Various studies have shown that structured reporting of pelvic imaging in patients with leiomyomas improves clarity and is perceived as more helpful for treatment planning by the referring providers. 7 , 11 Nonetheless, it is unclear if adding the FIGO PALM‐COEIN classification to a structured MRI report will provide relevant information to the referring providers and will ultimately impact patient management. In this study, we aim to evaluate if a structured report incorporating the FIGO PALM‐COEIN classification of leiomyomas contains adequate information for clinical decision making compared with non‐structured reports (NR).

FA is a subinvestigator for Ascelia Pharma AB. All other authors have no conflicts of interests.

A database search of our picturing archiving communication system (PACS) was conducted from January 1, 2009, to January 1, 2022, with the following inclusion parameters: female patients aged 18 years or older, MRI performed with contrast, mention of leiomyoma/fibroid/myoma(s) in the study indication, and outpatient imaging. Patients were excluded if they were pregnant at the time of the study or had undergone previous leiomyoma‐related procedures (i.e. history of uterine artery embolization, myomectomy). This study was determined to be exempt by the Institutional Review Board at University of Miami. Multiplanar, multisequence pelvic MRI scans were acquired on various 1.5‐ and 3.0‐Tesla scanners. Our departmental MRI protocol included turbo spin echo (TSE) T2‐weighted axial, sagittal, coronal sequences (variable TE and TR, slice thickness = 5 mm, field of view (FOV) = 200 mm), TSE T2‐weighted fat‐suppressed axial and sagittal sequences (variable TE and TR, slice thickness = 5 mm, FOV = 300 mm), a two‐dimensional (2D) in‐phase and out‐of‐phase gradient echo (GRE) T1‐weighted axial sequence (TE = 2.38 ms, TR = 180 ms, slice thickness = 5 mm, FOV = 300 mm), a set of diffusion‐weighted sequences (b values 50–500, slice thickness = 5 mm, FOV = 300 mm), and a half Fourier single‐shot turbo spin echo (HASTE) T2‐weighted coronal sequence (variable TE and TR, slice thickness = 5 mm, FOV = 300 mm). A set of GRE dynamic, high‐resolution 3D T1‐weighted fat suppressed axial sequences (variable TE and TR, slice thickness = 3 mm, FOV = 300 mm) were acquired before and 30 s, 1 min, and 5 min after the intravenous administration of MultiHance (gadobenate dimeglumine; Gd‐BOPTA, Bracco SpA, Milano, Italy) contrast medium (0.2 mL/kg body weight, up to 20 mL), with a power injection at a rate of 2 mL/s, followed by a saline flush of 20 mL. Additionally, a GRE dynamic, high‐resolution 3D T1‐weighted fat‐suppressed sagittal sequence (variable TE and TR, slice thickness = 3 mm, FOV = 300 mm), was obtained 5 min after intravenous administration of contrast. A total of 19 imaging key features (KF) were deemed relevant for clinical decision making and surgical planning by a consensus of an abdominal fellowship‐trained diagnostic radiologist (B1B1), an interventional radiologist (C1C1), and a gynecologist specialized in treatment of uterine leiomyomas (A1A1). These are specified in Table  1 and include: position and size of the uterus, description of the junctional zone, number of leiomyomas, their enhancement as well as the size, location, and appearance, and the myometrial thickness overlying submucosal leiomyomas. Comparison of key features for narrative report, structured reports, and structured reports with PALM‐COEIN subclassification of leiomyomas. a Abbreviation: UAE, uterine artery embolization. Data are presented as number (percentage). Two MRI reporting templates for leiomyomas, modeled after those outlined in the literature, 7 , 8 were collaboratively developed by a fellowship‐trained abdominal diagnostic radiologist (B1B1), a gynecologist specialized in treating patients with uterine leiomyomas (A1A1), and an interventional radiologist (C1C1) (Appendix  S1 ): (1) a uterine leiomyoma‐specific structured MRI reporting template (SR) adapted from the template proposed by Franconeri et al. 7 and (2) a uterine leiomyoma‐specific MRI template report, integrating the PALM‐COEIN FIGO classification by Gomez et al. 8 as an additional parameter (SR‐FIGO). Cases were reported using the SR and the SR‐FIGO templates by a fellowship‐trained radiologist (Y1Y1) not involved in the creation of the templates after 1 h of training on the structure of the templates by the radiologist involved in the creation of the template (B1B1), and the findings were reviewed by the same radiologist for accuracy. The study involved the anonymization of the MRI reports for uterine leiomyomas. An objective evaluation was conducted to identify the presence of KF in each report. The three MRI report types—NR, SR, SR‐FIGO—were compared for the presence of 19 KF deemed relevant for management of leiomyomas. Each KF was noted and assigned a value of 1 if mentioned in the report, irrespective of whether it represented a positive or negative finding. Conversely, if a KF was not mentioned, it was recorded as absent and given a value of 0 in the evaluation. Reports were scored by a radiologist‐in‐training (X1X1), not involved in the creation of the SR and SR‐FIGO templates. A gynecologist specialized in the management of leiomyomas (A1A1) and a fifth‐year postgraduate gynecologist‐in‐training (Z1Z1) were given de‐identified MRI reports, accompanied by an evaluation questionnaire for each report (Figure  1 ). Each questionnaire consisted of five questions, each with a five‐point scale to rate their responses, aimed at gauging their confidence. The objective of the questionnaire was to assess whether the reports provided adequate information to determine the optimal management approach for each patient (medical, surgical, or interventional), specifying surgical approach and type of surgery. Participants were also asked to indicate if image review was essential for determining the next steps in patient management and to estimate the time spent on image review. Magnetic resonance imaging reports evaluation form. The three reporting templates (NR, SR, SR‐FIGO) were compared for the presence of 19 KF deemed relevant for management of leiomyomas using Kruskal–Wallis test, after testing for normal distribution. Pairwise comparison between reports was performed with Mann–Whitney U test. χ 2 test (or Fisher exact test when applicable) was used to compare the answers of the gynecologist and gynecologist‐in‐training to the questions for the three different types of reports. Kruskal–Wallis test was used to compare the confidence scores of gynecologist and gynecologist‐in‐training when answering each question about the three different types of MRI reports. A P value less than 0.05 was considered statistically significant. Statistical analysis was performed with SPSS (IBM, Armonk, NY, USA).

Appendix S1.