Refining MRI protocols for endometriosis: a comparative study of abbreviated and full MRI sequences

Endometriosis is characterized by the presence of endometrial tissue, including glands and stroma, outside the uterine cavity. 1 It is estimated that 10% of women of childbearing age are affected by endometriosis, with 30%-50% of these cases being symptomatic. 2 The underlying aetiology of endometriosis remains elusive, and its pathogenesis is complex and multifactorial. 2 Endometriosis is primarily diagnosed through the histological examination of biopsied tissues, showcasing endometrial glands and stroma. While the gold standard for diagnosing endometriosis is histological examination of biopsied tissues via laparoscopy showing endometrial glands and stroma, imaging exams can offer valuable insights. 3 Transvaginal ultrasonography (TVS) and MRI are the most commonly used modalities in endometriosis diagnosis. 4 TVS is an accessible and cost-effective modality for diagnosing pelvic endometriosis but is limited by its field of view and operator dependency. 5 In contrast, MRI offers multiplanar reconstruction and superior soft tissue resolution, although it presents challenges such as higher costs, time consuming, limitation of contrast administration, and the need for specialized radiological expertise. 6 , 7 The introduction of abbreviated MRI (aMRI) protocols has emerged as a promising solution to overcome the limitations of traditional imaging methods, particularly in the last decade. 8 These protocols, with reduced sequence acquisitions, provide benefits such as shorter time of image acquisition and interpretation, cost-effectiveness, and diagnostic accuracy comparable to that of full MRI protocols, without the potential risks of contrast-related complications. 8 , 9 Despite these advantages, the use of aMRI in diagnosing endometriosis has been minimally explored in the existing literature. Studies in breast, prostate, and endometrial cancers have demonstrated that the diagnostic accuracy of aMRI is robust and comparable to that of full MRI (fMRI) protocols. 10–12 Given the consistent anatomical involvement pattern typically seen in endometriosis, which can be effectively detected using T2-weighted (T2W) sequences without fat suppression and T1-weighted (T1W) sequences with and without fat suppression, aMRI appears to be a rational choice for diagnosis. 13 In this context, our study aims to evaluate the intra-reader agreement between aMRI and fMRI protocols in diagnosing pelvic endometriosis, aiming to provide valuable insights to the field.

This retrospective cross-sectional study included 446 patients diagnosed with deep infiltrating endometriosis using MRI at Advanced Diagnostic and Interventional Radiology Research Center, Imam Khomeini Hospital Complex between January 2018 and January 2020. Patients were referred for MRI to confirm the diagnosis of endometriosis and assess its extent before undergoing laparoscopic procedures. Inclusion criteria were based on clinical examinations, primary complaints, and/or ultrasound findings suggestive of endometriosis. Patients with inadequate imaging quality, MRI protocols incompatible with the study design, images ruling out endometriosis based on gynaecological assessment, or a history of surgical procedures for endometriosis were excluded. Patients were selected through convenience sampling from those undergoing MRI for suspected endometriosis, adhering to the study’s protocols. The study protocol received approval from the Tehran University of Medical Sciences ethics committee (Approval Code: IR.TUMS.IKHC.REC.1400.097), adhering to the Helsinki Declaration. Patients were fully informed about the study, assured of the confidentiality of their personal information, and provided written consent. Baseline characteristics, including age and chief complaint, were recorded in a study checklist. MRI images, accessible via the hospital’s Picture Archiving and Communication System (PACS), were evaluated by a radiologist with over ten years of experience in gynaecology and obstetrics imaging. Initially, the aMRI protocol, comprising Fast Spin Echo (FSE) T2W non-fat-saturated, diffusion-weighted, and both non-fat-saturated and fat-saturated T1-weighted sequences, was assessed. Subsequently, standard fMRI protocol was interpreted. Reader finalized their decision before proceeding to the next protocol, preventing reversals of decisions. To minimize recall bias, a minimum two-week interval was mandated between evaluations of the two protocols for each patient. Agreement rates between the full and abbreviated protocols were evaluated. All images were acquired using predetermined protocols on Siemens Magnetom Trio 3 Tesla (Germany) or GE Discovery MR750w 3 Tesla (the United States) MRI systems. Pelvic MRIs were performed after a 4-6 h fasting period, 20 mm IM Hyoscine butyl bromide immediately before the scan and 5-10 cc intravaginal gel. The imaging sequences included non-fat-saturated and fat-saturated T1W in the axial plane, non-fat-saturated T2W in axial, coronal, and sagittal planes, as well as diffusion-weighted imaging (DWI) with b -value of 50, 400, 1000 s/mm 2 and apparent diffusion coefficient (ADC) map. Post-contrast gadolinium-enhanced T1W images were obtained in all three planes ( Table 1 ). An intravaginal gel (5-10 cc) was applied. Details of recruited MRI protocols. Abbreviations: DWI = diffusion-weighted imaging; FoV = field of view; GD = gadolinium-enhanced; T1W = T1-weighted; T2W = T2-weighted; TE = echo time; TR = repetition time. Endometriosis presents in various anatomical forms, including superficial endometriosis, deep infiltrative endometriosis (DIE), and endometriomas. Superficial endometriosis consists of small, often microscopic lesions located on or just beneath the peritoneal surface. MRI techniques have limited visibility of these lesions, which are primarily confirmed through laparoscopic procedures. In contrast, DIE involves lesions extending more than 5 mm beneath the peritoneal lining, potentially infiltrating the retroperitoneal space or adjacent pelvic organ walls. Endometriomas are cystic lesions typically found in the adnexal regions and contain senescent haemorrhagic material. 14 MRI detects a spectrum of abnormal signal changes indicative of endometriosis, directly correlating with its distinct pathological features. Enlarged endometrial glands manifest as hyperintense cystic areas in T2W images. Additionally, hyperintense signals on T1W sequences, predominantly appearing hypointense on T2W sequences, suggest haemorrhagic content within endometriotic foci. Irregularly shaped hypointense signals on both T1W and T2W sequences indicate fibrotic transformations. 15–17 Furthermore, MRI can reveal indirect signs suggesting endometriosis-associated adhesions. These signs include intestinal loop angulation, posterior vaginal fornix elevation, posterior stretching of the uterus and ovaries, hydrosalpinx, loculated fluid, and the disappearance of fat lines between pelvic structures. 18 , 19 To minimize interpretation biases, a protocol was established to identify endometriotic indices in each anatomical region as follows: Uterus serosa: Foci with hyperintensity in T1 or hypointensity in T2 weighted sequences at the uterus serosal level (sometimes extending to the serosal layer of fornix) indicated uterus serosal involvement. 18 Ovaries: Unilocular or multiloculated cysts with T1 hyperintensity, not suppressible on fat-saturated images, and T2 hypointensity (T2 shading) were considered main features of ovarian endometriosis (endometrioma). Additional manifestations included a blackish rim around the cyst, blackish spots in the cyst or ovarian parenchyma representing hemosiderin sediment in T2 sequences, and ovarian adhesions (kissing ovaries) ( Figure 1 ). 18 Uterine ligaments: Increased thickness and nodularity in the ligaments, including broad and round ligaments, indicated uterine ligament involvement. 17 Fallopian tube: Tubes filled with hyperintense fluid in T1 sequences, indicative of haematosalpinx, suggested fallopian tube endometriosis. 18 Vagina and rectovaginal septum: Thickening and nodularity of the vagina/rectovaginal septum, predominantly hypointense in T1 and T2 sequences, and obliteration of the rectovaginal septum were characteristic of endometriosis involvement. 18 Anterior cul-de-sac and vesicouterine pouch: Nodules with hyperintensity in T1 or hypointensity in T2 weighted in the anterior base of the uterus, making an obtuse angle with the bladder, along with uterine anteflexion and obliteration of the anterior cul-de-sac due to adhesions or hyperintense T2 foci indicative of dilated endometrial glands, were compatible with endometriosis diagnosis. 17 , 20 Bladder: Hypointense focal thickening of the bladder wall in T2 sequences leading to the disappearance of detrusor muscle signaling. 17 Ureter: Irregular hypointense nodules causing fading of fat wraps between nodules and the ureter in T2 sequences were hallmarks of ureteral endometriosis. 20 Recto-uterine pouch: Plaque-like hypointense thickening of the soft tissue without defined territories in T2 sequences, hyperintense foci in T1 sequences (compatible with haemorrhage), Douglas pouch obliteration, and uterine retroflexion were determinants of recto-uterine pouch endometriosis ( Figure 1 ). 17 , 21 Uterosacaral and torus uterinus: Bilateral or unmatched increased thickening and nodularity at the origin or throughout these ligaments, along with their abnormal appearance (arciform abnormality and tethered appearance), were characterized as endometriosis. 15 , 18 Rectosigmoid: Serosal or muscular layer involvement represented by increased thickness or wall fibrosis, stenosis, adhesion band, rectal angular anterior stretching, and involvement of the rectosigmoid anterior wall with a mushroom cup view (infiltrated fibrotic plaque) in T2 sequences served as diagnostic criteria for rectosigmoid endometriosis ( Figure 1 ). 17 , 22 Other including scar endometriosis or pelvic floor endometriosis: Thickening, nodularity, or direct infiltration of the pelvic floor muscles are considered indicators of endometriosis. These changes can be observed on T2-weighted MRI as hypointense fibrous tissue and on T1-weighted MRI as areas of high signal due to haemorrhagic components ( Figure 2 ). The visibility of fistulous tracts, marked inflammation, or fibrotic changes in the perianal space are considered evidence of endometriosis in this anatomical region ( Figure 3 ). 23 MRI findings of a 35-year-old female with pelvic pain, showing bilateral endometrioma, rectal deep infiltrating endometriosis (DIE), and myometrial DIE. (A-C) Axial T2W (A), T1W fat-saturated without contrast (B), and post-contrast subtracted (C) images demonstrate two cystic lesions with high signal intensity on both T1 and T2 sequences. These lesions exhibit thin wall enhancement post-contrast, consistent with bilateral endometriomas, and illustrate the “kissing ovaries” sign (red arrows). (D-F) Coronal T2W (D), sagittal T2W (E), and sagittal T1W post-contrast (F) images show a low-signal intensity lesion in the recto-uterine pouch, deeply infiltrating the anterior rectal wall with focal mural thickening, indicative of rectal DIE (blue arrow). Additionally, there is simultaneous deep infiltration into the posterior myometrium (green arrow), suggesting myometrial involvement by DIE. MRI findings of a 32-year-old female with dyspareunia, demonstrating deep infiltrative endometriosis (DIE) involving the levator ani, recto-uterine pouch, and anterior rectal wall. (A, B) Axial T1W fat-saturated (A) and T2W (B) images reveal a multiloculated cystic lesion with high T1 signal intensity and T2 shading, located in the retro-rectal space and embedded within the muscle fibres of the levator ani, indicative of DIE (red arrow). (C, D) Sagittal T2W (C) and T1W fat-saturated post-contrast (D) images demonstrate low T2 signal intensity areas in the lesion due to T2 shading, with enhancement seen in the wall and internal septations, without any evidence of a solid mural nodule (red arrows). Additionally, an enhancing low T2 signal intensity lesion in the recto-uterine pouch is noted, infiltrating the anterior rectal wall, consistent with rectal DIE (blue arrow). (E) Coronal T2W image shows the cyst interspersed with the levator ani muscle, causing a pressure effect on the anorectal junction (red arrow). MRI findings of a 38-year-old female with a history of caesarean section and cyclic lower abdominal wall pain, demonstrating scar endometriosis. (A) Axial T2W image shows an irregularly marginated mass at the site of the caesarean scar (red arrow), exhibiting low-signal intensity on T2-weighted imaging. (B) Axial T1W fat-saturated pre-contrast image reveals small foci of high T1 signal intensity within the mass, indicative of blood content (red arrow). (C, D) Axial T1W fat-saturated post-contrast (C) and subtracted (D) images demonstrate heterogeneous enhancement of the mass (red arrows), consistent with scar endometriosis. Data were entered into SPSS Statistics for Windows, Version 23.0 (Armonk, NY: IBM Corp). Descriptive statistics were reported as mean and standard deviation or frequency and percentages. Cohen’s kappa coefficient assessed intra-reader variability in diagnosing endometriosis lesions. A P -value less than .05 was considered statistically significant.

In this study, we evaluated MRI findings from 446 patients diagnosed with endometriosis. The mean age of the study population was 34.13 ± 7.37 years, ranging from 18 to 68 years, with the majority of patients in their fourth decade of life (50.7%) ( Table 2 ). Age distribution of the study population. We observed a varied distribution of endometriosis across different anatomical regions using a full protocol pelvic MRI. The ovary was the most frequently involved site, identified in 88.8% of cases, followed by the recto-uterine pouch with a 65% involvement rate. The uterus and fallopian tubes were also frequently affected, with rates of 60.1% and 41.7%, respectively. Conversely, the least affected sites were the caesarean section scar and bladder, each with a 1.3% involvement rate, as well as the ureter and vesicovaginal septum, which demonstrated minimal frequencies of 2.2% and 2.7%, respectively. Table 3 presents detailed insights into the agreement analysis between aMRI and fMRI interpretations, with perfect concordance (kappa coefficient = 1) observed in assessments of the ovary, bladder, uterus, and caesarean section scar. This suggests identical interpretative outcomes between aMRI and fMRI protocols for these regions. Additionally, the rectum and uterine ligaments demonstrated exceptionally high kappa coefficients of 0.98 and 0.97, respectively, indicating near-perfect agreement, while detection of malignant transformation in the existing ovarian endometriomas, still showed substantial concordance with kappa coefficients of 0.66. Agreement rates between MRI modalities in pelvic endometriosis diagnosis. In our study, solid-enhancing nodules were identified in four patients (0.9%), with two nodules located in the ovary and two in the fallopian tube. These lesions appeared suspicious on T2W imaging; however, DWI assessment showed no significant restriction in two of the nodules. Pathological analysis of these biopsied nodules did not reveal any atypical or malignant transformations. In contrast, the remaining two nodules showed DWI restriction, with one identified as clear cell carcinoma and the other as a borderline seromucinous tumour ( Figures 4 and 5 ). Our study suggests the DWI sequence using three b -values would be a promising diagnostic tool as a reliable alternative to contrast injection in detection of malignant transformation of existing ovarian endometriomas. Transitioning from the full MRI protocol to an abbreviated MRI protocol in our institution has led to a reduction in imaging acquisition time by approximately 30% and a cost savings of about 45%. MRI findings of a 37-year-old female with known endometriosis and a sudden rise in tumour markers, showing a seromucinous tumour superimposed on existing endometrioma. (A, B) Sagittal T2W (A) and axial T2W (B) images show a cystic mass with high T2 signal intensity (red arrow) containing several solid mural nodules (blue arrow). (C, D) Axial fat-suppressed T1W pre-contrast (C) and post-contrast subtracted (D) images reveal high signal intensity on T1-weighted imaging due to internal blood/mucin content (red arrow), with enhancement of the solid nodules (blue arrow). (E, F) Axial diffusion-weighted image ( b -value = 1000 s/mm 2 ) (E) and ADC map (F) demonstrate diffusion restriction in the mural nodule (blue arrow), raising concern for a superimposed malignancy. MRI findings of a 48-year-old female with rising tumour markers, demonstrating suspicious features of malignant transformation of an endometrioma, with pathology confirming borderline serous cystadenocarcinoma. (A-C) Axial T1W fat-saturated (A), axial T2W (B), and axial T1W post-contrast subtracted (C) images show a cystic mass in the right ovary with high signal intensity on T1W and low-signal intensity on T2W due to T2 shading (red arrow). A nodular component in the medial wall of the cyst exhibits enhancement (blue arrow), raising suspicion. (D, E) Axial diffusion-weighted images ( b -value = 1000 s/mm 2 ) (D) and ADC map (E) reveal diffusion restriction in the solid component (blue arrow), a concerning feature for malignant transformation.

In this study, we aimed to assess the diagnostic capabilities of MRI in identifying the extent of endometriosis and to compare the efficacy of aMRI versus fMRI. Our analysis revealed a high level of diagnostic concordance between the two protocols across various pelvic anatomical sites, underscoring the potential of aMRI as a reliable alternative for endometriosis diagnosis. These findings suggest that the use of contrast-enhanced imaging may not be as crucial for accurately detecting endometriosis as previously believed, offering new insights for optimizing MRI protocols in endometriosis evaluation. At our institution, transitioning from the full MRI protocol to an abbreviated MRI protocol has led to a remarkable reduction in imaging acquisition time by approximately one-third and a cost savings of nearly half of the full imaging protocol. Furthermore, by omitting the use of contrast material, we eliminate the risks associated with contrast agents, such as allergic reactions and nephrogenic systemic fibrosis, enhancing patient safety. Importantly, these time and cost efficiencies do not come at the expense of diagnostic accuracy, as demonstrated by our comparable diagnostic outcomes. Diagnosing endometriosis, particularly when it presents with common non-specific symptoms such as abdominal pain, poses significant challenges in clinical practice. 22 Various diagnostic modalities, including ultrasonography, spiral CT scan, and MRI, have been employed. 24 , 25 While ultrasonography is often recommended as the first-line imaging modality for endometriosis diagnosis, it has limitations in examining deep pelvic areas. 22 , 26 CT scan, although diagnostically valuable, has raised concerns due to its high radiation emission, shifting attention towards MRI. 25 Our findings highlight MRI’s capability, especially aMRI protocols, in overcoming these limitations without relying on contrast enhancement. This is supported by literature indicating MRI’s superior diagnostic accuracy for deep pelvic endometriosis, even in the absence of contrast agents. 27–29 Bazot et al 30 reported similar outcomes regarding the limited impact of contrast use on distinguishing endometriosis in various pelvic regions. They observed inter-observer agreements ranging from 0.62 to 0.73 in the rectosigmoid colon, vaginal, and bladder lesions when comparing contrasted versus non-contrast pelvic MRI. Another study by da Silva et al 9 comparing abbreviated pelvic MRI with standard protocol MRI reported diagnostic values of 0.83-0.86 versus 0.83-0.87, respectively. They concluded that abbreviated MRI is comparable to contrast-enhanced MRI. In contrast, Scardapane et al 31 assessed the diagnostic power of contrast-enhanced MRI versus MRI-colonography and reported over 90% sensitivity and specificity for MRI in all pelvic regions, suggesting that MRI-colonography may enhance MRI’s diagnostic capability. Although it is assumed that the accentuation of imaging in abbreviated pelvic MRI might have negative impact on the resolution and quality of the images, we found insignificant difference which has been confirmed in other studies assessing aMRI, 9 or another protocol using low-resolution sequences (single-shot T2 and fast breath-hold T1). 32 Intestinal endometriosis represents the most severe form, with symptoms resembling various pathological conditions, making diagnosis challenging. 33 MRI has shown high sensitivity and specificity exceeding 80% for diagnosing this type of endometriosis. 34 , 35 These findings align with our study, showing a high agreement rate between the two applied modalities, favouring abbreviated pelvic MRI over enhanced standard protocols. The genitourinary area, including the bladder, vagina, uterus, ovaries, and surrounding ligaments, is likely the least complex pelvic region to diagnose endometriosis. Although MRI, with or without contrast, has been favoured for detecting endometriosis in these regions, more accessible and cost-effective modalities like ultrasonography may be more appropriate due to the anterior location of the affected areas. 36 A critical concern is whether the absence of a post-contrast sequence in abbreviated MRI could impair the diagnosis of incidental findings such as malignant adnexal lesions. 37 However, the high-resolution anatomical images provided by T2W volumetric sequences in this imaging protocol make this scenario unlikely. 9 Our study identified four solid-enhancing ovarian and fallopian tube nodules (0.9%) on T1 with gadolinium sequences in contrast-enhanced MRIs. Two of these lesions, which did not show restriction on DWI, were benign upon pathological examination, while the other two, which showed DWI restriction, were malignant clear cell carcinoma and borderline seromucinous tumour. Harth et al 38 found no significant difference in patient outcomes when using contrast for indeterminate adnexal lesions. Hricak et al 39 reported that non-contrast-enhanced images detected 93% of adnexal lesions, compared to 95% with contrast-enhanced images, which was not statistically significant. Further investigations have shown that despite the additional morphological information provided by contrast-enhanced imaging, no significant differences exist between non-contrast and contrast-enhanced MRIs. 40 Although several studies have shown that adding contrast-enhanced images to MRI protocols does not improve diagnostic performance for endometriosis, our own study confirms this finding. Nevertheless, the Society of Abdominal Radiology’s Endometriosis Disease-Focus Panel highly recommends utilizing post-contrast images when interpreting patient scans suspected of having endometriosis, specifically in differentiating endometriosis from neoplasm. 41 We suggest aMRI in all cases, so that contrast-enhanced study should be reserved only in cases of indeterminate adnexal lesions with uncertain probability of being benign or malignant. In this regard, contrast-enhanced images may still have value in identifying suspicious malignancy in adnexal lesion, however, examining only T2W and DWI sequences can also aid in clinical decision-making. The clinical implications of the findings from the current study span several domains. The observed high concordance between aMRI and fMRI for diagnosing endometriosis could significantly influence clinical decision-making. This concordance may reduce the reliance on contrast agents, thereby mitigating associated risks or contraindications. This aspect holds particular importance for patients with renal impairment or other contraindications to contrast administration, highlighting the enhanced safety profile of aMRI as a distinct advantage. Additionally, the cost-effectiveness of aMRI compared to fMRI presents another clinical benefit. The reduction in imaging time and avoidance of contrast agents can lead to cost savings, potentially making MRI a more accessible diagnostic tool. Taken together, these findings suggest that aMRI could be a promising approach for endometriosis diagnosis and management, potentially leading to improved patient outcomes and quality of life. The current study has several limitations. Foremost among these is its retrospective design, which may introduce biases in patient selection and constrain the scope of analysis based on the collected data. Additionally, the study predominantly focused on the diagnostic capabilities of aMRI and fMRI without assessing long-term clinical outcomes or treatment responses associated with these imaging modalities. In terms of statistical analysis, while kappa coefficients were employed to assess intra-reader agreement between aMRI and fMRI interpretations, the study did not address potential inter-observer variability among radiologists. Furthermore, the study did not differentiate between different stages of endometriosis, which could potentially impact the diagnostic accuracy of the imaging protocols. Conversely, the study also exhibited several strengths. These include a comparative analysis between aMRI and fMRI for endometriosis diagnosis, a relatively large sample size, a comprehensive evaluation of endometriosis distribution across various pelvic anatomical sites, and findings with direct clinical implications that could potentially influence diagnostic protocols and decision-making processes.

Our study’s results indicate an excellent degree of concordance between non-contrast and contrast-enhanced pelvic MRI. Additionally, abbreviated non-contrast pelvic MRI provides satisfactory findings for detecting endometriosis across all pelvic areas, and utilizing T1, T2, and DWI sequences can provide reliable and accurate information regarding the detection and characterization of suspicious adnexal lesions in patients with endometriosis. These findings suggest that abbreviated non-contrast MRI could be a viable alternative to contrast-enhanced MRI for endometriosis diagnosis, in that the contrast-enhanced study would be reserved in only a minority of patients with equivocal results especially in the detection of malignant mural nodules in the endometriomas.