A Study on the Diagnostic Abilities of Ultrasound Scans in Assessing the Uterine Fibroids Against Magnetic Resonance Imaging Findings in the Same Subject Population

Background: This study was designed to evaluate the matching percentage among ndings of the ultrasound scans to the magnetic resonance imaging in women with broids in Bahrain. Methods: This descriptive, retrospective study was conducted from January 2016 to December 2018 including all the female patients referred from the Gynaecological Department to the Radiological Department in the Bahrain Defense Force Hospital for magnetic resonance imaging and ultrasound scan for broid evaluation.The data collected included their age, parity and nationality. The StatsDirect software was used to analyze the broids according to the site, size, type and number. Results: The average age of the recruited 205 female patients for the study was 43 years, with 81.5%as Bahraini citizens and a mean parity of 2.3. Ultrasound scan ndings matched the Magnetic Resonance Imaging in the posterior broid sites in 52.5% of cases, followed by anterior (38.1%) and fundal (21.8%). For broid size, Ultrasound scans matched Magnetic Resonance Imaging ndings in 83.3% for broids between >5 to ≤ 10 cm, in 78.3% for broids between >2 to ≤ 5cm, and in 36.8% for broids sized ≤ 2 cm. Fibroids more than 10cm in size were in agreement for 33.33% of broids. Submucosal broids matched in just 29.4% of cases, but for the subserosal broids, it was 44.8%. The matching score for more than 4 broids was 61.8%, followed by single broids (54.8%) and lastly for two broids (34.8%). Conclusions: Fibroid site and size had the highest matching rates amongst all the categories.


Background
Uterine broids (or uterine leiomyomas) are the main gynecological tumors occurring in up to 50% of the women of reproductive age globally [1]. More than 30% of the women over the age of 30 years are affected by the broids, and it is estimated that more than 70% of the women will be affected in their lifetime [2]. Since, most affected women do not encounter any distressing effects or symptoms, uterine broids often go undiagnosed [3]. However, broids do have an approximately 0.1 to 0.8% risk of transforming into malignant sarcomas. 1 Affected women have an increased morbidity and an increased risk of encountering reproductive problems such as infertility and miscarriages [4].
Traditionally, the diagnosis of leiomyomas has been carried out using diagnostic imaging modalities, with ultrasound scans (USS). It has shown relatively high sensitivity and speci city, usually being the rst-line imaging examination [5]. USS is used as the standard con rmatory modality because they are relatively cheap, accessible, harmless and can differentiate uterine broids from a pregnant' uterus or other adnexal tumors [6]. The two main routes used for ultrasonography are the transvaginal ultrasound scans (TVUSS) and transabdominal ultrasound scans (TAUSS) routes, and ideally both should be performed to detect, characterize and map the uterine broids [7]. TVUSS has a higher sensitivity for smaller broids and is capable of visualizing broids as small as 5 mm [7,8]. However, when the uterus is retroverted or bulky, TAUSS is preferred because the uterine fundus could possibly be out of view transvaginally [7]. TAUSS has a wider eld of view, increased transducer movement, more signal penetration ability, and offers the ability to examine other organs [9]. TAUSS itself has a limitation; it is di cult to be used in very obese patients and is less effective when the uterus is greater than 300 ml in total volume [8]. Noticeably, USS has several disadvantages; it has a low e cacy, when there are multiple broids present, since these broids may then produce acoustic shadows through which sound fails to propagate [8]. It may miss smaller broids and be unable to determine the exact number and their location. USS may miss subserosal broids [7,10]. Another disadvantage associated with USS is that it also has considerable inter-observer/intra-observer variability or operator dependence, therefore, in the hands of a less skilled user, broids may be missed [7]. Overall, it could be interpreted that USS results have lesser reproducibility when compared to magnetic resonance imaging (MRI), which is the other diagnostic test for uterine broids [7].
Although MRI is more expensive,it is the most sensitive and accurate modality as it could diagnose, measure, localize and quantify all the lesions including very small broids [11][12][13]. Other than the higher sensitivity in identifying broids than USS, MRI alsodoes not use any ionizing radiation. MRI provides greater spatial resolution, greater contrast resolution, and superior multiplanar capabilities. Combined, this makes MRI a superior evaluator of the uterine zone anatomy (clearly differentiating subserosal, intramural and submucosal broids) and also superior in its localization of small and unusually located uterine broids and cervical broids [14,15]. Additionally, MRI can also diagnose other pelvic abnormalities and pathologies missed on the USS and is useful in assessing broid response to the treatment [7,16].
Since physicians have a high probability of encountering women with uterine broids, and nearly 40% of them may have signi cant morbidity, knowledge of the diagnostic modalities for leiomyomas among them is essential for broid management. The aim of thisstudy is to investigate whether USS was as accurate as MRI in diagnosing broids in the female patients. This is the rst descriptive, retrospective study conducted in the Kingdom of Bahrain evaluating the performances of MRI and pelvic USSs in diagnosing the uterine broids.

Analysis of Data
The software used to analyze the data was StatsDirect statistical package version 3.1.22 (Cambridge, UK, 2018). Using this software, the patientswere analyzed according to their demographics such as age and nationality, parity, entitlement, symptoms, type of imaging conducted, as well as broid characteristics observed. Information on the clinical features of the patients observed was categorized into bleeding, irregular periods, pain, mixed symptoms, and asymptomatic. We also included the percentage of patients who underwent post procedureimaging along with patients who did not and the patients with unknown outcome of the care. The percentage of patients not identi ed with speci c sub-categories of broid site, type, and number were also reported. Continuous variables have been presented as mean ± SD and median with range. Categorical variables have been presented as percentage.

Data extraction and diagnostic equipment
All the o cial radiological reports of MRI and USS were reviewed. The USS was done at both the departments viz. Gynecological Clinics and radiological X-ray departments. However, some USSs in the Gynecological department were retained from the primary hand-written notes. In regard to the diagnostic equipment used, the USSs were performed using two different machines: Hitachi and General Electric (GE). USSs were carried out both trans-abdominally and trans-vaginally. Further, for MRI also, departmenthas two machines with different magnetic powers: GE was 3 teslas and Siemens with 1.5 teslas. Any patient with missing information was excluded from the study.

Patient characteristics
In this study, a total of 205 female patients were diagnosed with broids after undergoing MRI and USS of the pelvis over the period of two years (from January, 2016 to December, 2018). The average age of the patients was 43 years. Among all the recruited patients, nearly 81.5% were Bahraini women and only 18.5% consisted of females from other nationalities. The majority of the women were entitled to free care (73.7%) and the other 26.3% diagnosed received treatment under private care. The predominant clinical features observed were bleeding, irregular periods, and pain, with bleeding (23.9%) being the most common solitary symptom across the patients. Moreover, the highest percentage of the females experienced mixed symptoms (32.7%). 41 patients were found to be asymptomatic and 80% of the patients were with symptoms. Further, 105 patients (51.2%) did not undergo procedure for the removal of broids, 91 patients (44.4%) did, and the procedure history of a few patients (4.4%) was unknown (Table  1).

Fibroid characteristics by MRI
Most of the cases identi ed in the study presented with fundal broid location (38%), followed by anterior (20.5%) and posterior (19.5%) sites. The least being the cervical broid site which was only found in 5.4% of patients.
Fibroid size is an important characteristic observed in the terms of diagnosis. The most common size of broids diagnosed by MRI was > 2 to ≤ 5 cm (44.9% patients). Fibroids > 5 to ≤ 10 cm in size were the second most common subcategory diagnosed (32.2% of patients). Fibroids ≤ 2 cm and > 10 cm in size were seen in the smallest percentage of patients ( Table 2). The location of the broid is also an important diagnostic characteristic.Intramural (38%) was the most common broid type diagnosed by MRI in our patients, followed by subserosal (28.3%). Very few cases were identi ed for the submucosal (8.3%) and mixed (14.6%) broid types. broids (11.2%). The broid numbers in a small percentage of patients were not identi ed in this study.

USS ndings weighed against MRI ndings
In regard tothe broid site, the highest percentage of the matched scan to the MRI were in the posterior broid sites (52.5%), showing that more than half of the scan cases matched the gold standard diagnostic method. Anterior broid sites followed at 38.1% and the fundal at 21.8%. Cervical broid sites had 18.2% of the scans matching to the MRI and the match rates were not good for the lateral broid sites (9.5%) ( Table 3). A trend in the broid size was also noted revolving around the matching of US scans to the MRI ndings. The highest percentages of matching ndings were with broid sizes between > 5 to ≤ 10cm (83.3%), followed by > 2 to ≤ 5 cm and 0 to ≤ 2 cm, respectively. Surprisingly, broids more than 10cm in size were observed to have the lowest match (33.3%). Moreover, the highest matched scans for the broid types were observed in intramural (44.9%) and subserosal (44.8%) compared to the mixed types being only 10% ( Table 3).
The highest agreement found in identifying the broid number was in the presence of multiple broids at 61.8%, followed by single broid at 54.8%. The lowest matching was observed with two broids at 34.8% (Table 3).

Discussion
Previously published studies have compared the merits of USS and MRI ndings with regards to the broid management. It has been cumulatively observed that MRI has more sensitivity and better speci city, indicating the numbers of broids present and their location more accurately [14]. Consequently, MRI has been widely considered to represent the gold standard in such radiological investigations. In one recent study, more than half the patients who underwent USS were found to have additional gynecological problems after undergoing MRI [16]. For these reasons, MRI is seen as the gold standard investigationas it can detect other gynecological abnormalities in addition tothe broids. It has been noted that women with broids can present with non-speci c symptoms and on the contrary, they can also be asymptomatic [1,17]. In this study, 20% of the patients were found to be asymptomatic. Typical symptoms observed in the case of broids were infertility, dysmenorrhea, pelvic pain and menorrhagia, which are broad and non-speci c [3,18].
Inastudy by Dueholm and colleagues (2002), it was reported that the average number of broids located by MRI were 15 ± 2.8 compared to transvaginal USS which located 8 ± 2.6 broids. USS missed more broids than MRI and the discrepancy almost doubled when broids were classi ed according to both the position and uterine wall embedment [19]. Levens' study showed that MRI is twice as sensitive as USS in diagnosing the broids with a rate of 80% compared to 40% [20]. Our data showed that the best correlation between USS and MRI ndings were in the cases with multiple broids over the three categories as shown in Table 3. This nding needs to be considered with caution, as we assessed the matching in the group of patients with over 3 broids without being speci c of their number. It could be explained as when USS detected 4 broids and MRI con rmed the presence of more broids in the same patient, the data entry was still considered as matched. Noticeably, the observation which is accurate and clear is that when there is only one broid, the chance of matching with the scan was over 50%.
Detection of subserous broids using USS showed to have a reduced sensitivity with some broids being undetectable or completely missed in some cases [21]. In our study, USS matched the MRI ndings in 44.8% of the subserosal broids. Our study demonstrated variable matching abilities between USS and MRI in different broid types. The data of the present study further con rmed that submucosal broids are usually di cult to be differentiated on the USS scan. Only 29% of those cases matched the MRI ndings. Other contributing factors would have added to the equation such as the size and the site. Contrary to our nding, a study by Ahmad group found USS to have a higher speci city and positive prediction value (PPV) than MRI in the diagnosis of the submucous broids.
In the study conducted by De La Cruz et al., broids were classi ed according to size, site, and number, but no percentages were given for the rate of occurrence for each individual characteristic of broids. The best matching site was the posteriorly located broid in the present study. This could be due to the angle of the vagina, where the posterior wall would be more accessible and nearer to the TVUSS. Surprisingly, our study showed that the cervical broids had alower matching rate than the MRI. This could simply be due to the mislabeling as anterior or posterior rather than cervical. Determining the accurate size of the broids is crucial forits management plan. In a previous study analyzing broid size, USS underestimated the size of a large 16.6 cm diameter broid to be 12.5 cm [22]. Findings of our study were in line with this previously published data and showed that broids over 10cm correlatedpoorlywith the USS (matching only 33.3% of cases) (Table 3). However, the USS ndings in our study matchedvery well with the MRI ndings in case of the broids between 2 and 10 cm in diameter.
There are limitations to all of these studies, as technical skills play a crucial role in producing accurate USS results. The reality is that different levels of experience and competence in uence the ndings, as do the differences between the models of USS and MRI machines used. In addition, missing data reported as non-identi ed could also skew the results. Increasing the sample size of the study could help to reduce the error, but it will not entirely eliminate it.

Conclusion
Although MRI has been consideredsuperior in broids diagnosis based on its various properties, it does not negate the merits of USS as a preliminary diagnostic tool, especially where MRI is not readily accessible. In the present study, USS and MRI ndings correlated reasonably well in terms of broid site and size.