MRI in the Diagnosis of Endometriosis and Related Diseases

Histologically, DIE has been defined as endometriosis infiltrating the peritoneum to a depth > 5 mm [ 14 ]. As described above, most endometriotic lesions visible on MRI are DIE, because superficial peritoneal implants are usually not visible [ 15 ]. DIE is frequently found in organs and structures surrounding an ovarian endometriotic cyst, where it affects their function and causes chronic symptoms and infertility. Several types of pain are related to DIE, including severe dysmenorrhea, intense dyspareunia, noncyclic chronic pelvic pain, painful defecation during menstruation, urinary tract symptoms, and gastrointestinal symptoms [ 19 ]. On microscopic examination, a characteristic finding of DIE is fibromuscular hyperplasia around the ectopic endometrial glands. The endometrial glands and stroma infiltrate the adjacent tissue and induce smooth muscle proliferation as well as a fibrous reaction, resulting in solid nodule formation, fibromuscular tissue thickening, and anatomical structure distortion [ 60 ]. Reflecting these microscopic features of fibrotic tissue, MR images show hypointense nodular lesions or soft tissue thickening with irregular, indistinct, or stellate margins on both T1WI and T2WI [ 5 60 61 62 ]. Sometimes, bright lesions on T1WI can be found, representing hemorrhagic foci. Diagnostic values of MRI for DIE are quite high: sensitivity, 90%; specificity, 91%; positive predictive value (PPV), 92%; negative predictive value (NPV), 89%; and accuracy, 90% [ 63 ]. Accurate diagnosis of DIE and evaluation of its spread and severity support appropriate treatment selection for a diversity of patients. In fact, MRI findings might vary from institution to institution. Thus, it is important to obtain the characteristic findings of DIE from several institutions [ 5 15 42 64 ]. This section presents a review of MRI findings of DIE in representative pelvic organs and structures, along with the diagnostic ability of each imaging method. The posterior cul-de-sac is a frequently involved site of endometriosis. Complete posterior cul-de-sac obliteration was assigned the highest score in the rASRM classification [ 65 ]. In cases of posterior cul-de-sac obliteration during laparotomy, advanced laparoscopic skills are required for the separation of adhesions and for increased risk of bowel endometriosis, wherein segmental resection and anastomosis might be necessary [ 66 67 ]. Therefore, preoperative diagnosis may help gynecologists in surgical planning. On transvaginal US, the ‘sliding sign’ is known to be useful for diagnosing posterior cul-de-sac obliteration, providing a sensitivity of 92.9%–100%, specificity of 90.9%–100%, and accuracy of 93.1%–100% [ 68 ]. The sliding sign uses gentle pressure by the transvaginal US probe to assess whether the anterior rectum glides freely across the posterior aspect of the cervix and posterior vaginal wall, and whether the rectosigmoid glides freely over the posterior aspect of the posterior upper uterus [ 67 ]. On MRI, five related findings were obtained: retroflexed uterus, elevated posterior vaginal fornix, intestinal tethering or a tethered appearance of the rectum in the direction of the uterus, faint strands between the uterus and intestine, and fibrotic plaques or nodules covering the serosal surface of the uterus ( Figs. 3 , 8 ) [ 69 ]. Among these findings, the retroflexed uterus had the highest sensitivity (mean sensitivity, 60.8%). Fibrotic plaques on the uterine serosal surface showed the highest specificity, PPV, and accuracy (84.3%, 76.8%, and 64.5%, respectively) [ 69 ]. The most accurate combination was intestinal tethering in the direction of the uterus and fibrotic plaque on the uterine serosal surface [ 69 ]. Macario et al. [ 70 ] reported that a retrouterine fibrous mass, intraperitoneal fluid displacement, and adherence of bowel loops showed the best performance (92.8%, 93.1%, and 86.1% accuracy, respectively). The cul-de-sac can be difficult to evaluate using laparoscopy in cases of strong adhesions. These MRI findings are useful for identifying DIE in the obliterated posterior cul-de-sac and to avoid underestimation. Torus uterinus is a small transverse thickening where the USL is attached. Originating from the torus uterinus, the USL courses dorsocranially toward the sacrum. Torus uterinus and USL (86%) are the most frequent sites of DIE and the second most frequent site of endometriosis (46%) after ovarian endometriosis (67%) [ 63 71 ]. The involvement of the USL is asymmetrical. For example, 40% of patients have unilateral involvement, while 60% have bilateral involvement [ 63 ]. Endometriosis of the USL can be more often observed close to the torus uterinus [ 42 ]. On US, endometriosis of the USL appears hypoechoic or with a mixed echotexture [ 24 ]. On MRI, the involvement of the USL is found to have low SI on T2WI, representing fibrous tissues ( Fig. 9 ) [ 63 ]. T2-low intensity thickening or a mass with regular or irregular margins is observed at the torus uterinus ( Figs. 8 , 9 , 10 ) [ 63 ]. Observation of the USL might be difficult because of the retroflexed uterus or endometriotic cyst at the posterior uterus by laparoscopy [ 63 ]. The diagnostic capability of MRI for detecting endometriosis in USL is superior to that of transvaginal US. A meta-analysis showed that the pooled sensitivity of MRI was 0.70, specificity was 0.93, US sensitivity was 0.67, and specificity was 0.86 [ 72 ]. A recent study also found significant superiority of MRI for the detection of DIE of USL, with an accuracy of 71% in US and 84% [ 73 ]. For better visualization of the total USL, Bazot et al. [ 28 ] reported the utility of taking thin-section oblique axial and perpendicular views to the cervix. Reportedly, the addition of oblique axial T2WI improves the accuracy of USL endometriosis on MRI, especially for left USL endometriosis, with good agreement between readers (Kappa = 0.51) [ 28 ]. When bilateral ovaries are located as mutually adjacent in the posterior cul-de-sac or at the cornua of the uterus because of adhesion by endometriosis, the condition is termed ‘kissing ovaries’ ( Figs. 8 , 11 ) [ 15 ]. US examination has revealed that this finding is strongly related to disease severity, frequency of infertility, and operation time [ 74 ]. Kissing ovaries and retropositioned ovaries (ovaries located behind the uterus) on MRI show a significant correlation with DIE (sensitivity, 67%; specificity, 68%; PPV, 55% in the detection of DIE) [ 75 ]. As Williams et al. [ 75 ] described, ovaries are anchored in position by three structures: the mesovarium, utero-ovarian ligaments, and suspensory ligaments. Kissing ovaries result from the distortion of these anchoring structures following intense inflammation and fibrotic changes stimulated by endometriosis [ 75 ]. The prevalence of round ligament of the uterus (RLU) involvement is reported to be 3%–5%, with left-side predominance [ 76 ]. Anatomically, the round ligament originates anterolateral of the uterine horn, coursing under the broad ligament toward the inguinal canal. Frequently involved sites have also been reported. As reported by Gui et al. [ 77 ], RLU is divisible into intra-pelvic and extra-pelvic parts. Most cases of endometriosis involving the RLU are located extrapelvically, at the distal part of the ligament near the inguinal region. Therefore, the clinical symptom is the presence of a painful, palpable inguinal mass, with or without menstrual variation in the size or severity of symptoms [ 77 ]. Another report described the most frequent site as the proximal third of the ligament (approximately 14%) [ 15 ]. Predominance of right side involvement is also known to occur, probably because of clockwise intra-peritoneal fluid circulation from the left paracolic gutter to the right side through the pelvic floor [ 77 ]. However, questions regarding side predominance (right or left) persist [ 77 ]. Laparoscopically, endometriosis is diagnosed by the presence of shortening, thickening, and deviation of the RLU [ 78 ]. Similarly, a thickened and shortened RLU was observed on MRI. It can be irregular with a nodular appearance ( Fig. 9 ) [ 77 ]. The SI depends on the degree of hemorrhagic foci and fibrous tissue, but most cases tend to show hypointense lesions on both T1WI and T2WI. The rectosigmoid colon, which is the most frequent site of gastrointestinal endometriosis, accounts for up to 65.7% of disease incidence, followed by the ileocecal junction (20%) and rectum (15%) [ 42 79 ]. Depending on the disease severity, such as the percentage of circumference involved, surgical procedures including segmental resection, discoid resection, and rectal shaving can be considered [ 80 ]. The characteristic MRI findings of rectal endometriosis are the ‘fan-shaped’ and ‘mushroom cap’ appearances ( Fig. 6 ) [ 64 81 ]. The ‘fan-shaped’ appearance is defined as fan-shaped bowel wall thickening with hypointense T2WI (similar signal to muscle) and slightly high SI at the mucosal side [ 64 ]. Occasionally, intense signal spots are visible on T1WI. The ‘mushroom cap’ appearance refers to focal thickening of the rectal wall to resemble an umbrella-like head of a mushroom [ 81 ]. Both signs correspond to hypertrophy of the muscularis propria and smooth muscle hyperplasia, which might be induced by stimulation of invading endometrial tissue located within the muscle layer or submucosal layer of the bowel [ 64 81 ]. Consequently, when these are observed on MR images, muscular invasion of endometrial tissue is expected [ 64 ]. Reportedly, the sensitivity, specificity, PPV, and NPV, as well as accuracy of fan-shaped appearance for infiltration of the muscular layer of the bowel wall are 100%, 75%, 96%, 100%, and 96% [ 64 ]. Distention of the rectal lumen by US gel was proposed for better visualization of the disease, because of better contrast between the rectal lumen and the wall [ 82 ]. However, the diagnostic capability of colon wall infiltration was not significantly different before and after US gel application [ 82 ]. Motion artifacts due to the distended rectum and lack of US gel at the upper part of the rectum are regarded as causes of ineffectiveness [ 82 ]. Recently, Rousset et al. [ 83 ] demonstrated that greater than 135° circumference involvement and greater than 14 mm lesion thickness are specific MRI findings predicting the need for segmental resection for rectal endometriosis. Thus, MRI can provide important information that supports decisions made in formulating a treatment strategy. Urinary tract involvement occurs in 1.0%–6.4% of endometriosis patients, depending on the series [ 84 ]. The most frequently involved area is the posterior bladder wall [ 42 ]. Symptoms may be very subtle or be as concerning, such as dysuria or hematuria [ 85 ]. Endometrial cells that seed on the anterior uterine wall or bladder wall often during earlier surgery, such as caesarean section, are thought to infiltrate into the muscular layer and manifest as mural masses [ 85 ]. The lesion may be clinically mistaken for cancer. On MRI, bladder endometriosis is depicted as hypointense nodular formation on both T1 and T2WI, sometimes accompanying high signal foci reflecting ectopic endometrial glands and hemorrhage ( Fig. 10 ) [ 60 85 ]. In correlation with surgical and pathologic findings, MRI showed a sensitivity of 88%, specificity of 98%, PPV of 88%, NPV of 98%, and an accuracy of 98% [ 63 ]. In 90% of cases with bladder endometriosis, several other sites with endometriosis have been reported, including those in the ovaries and retrocervical areas, as well as adenomyosis [ 85 ]. Bladder endometriosis is sometimes found as a continuous lesion with uterine adenomyosis occurring via the vesicouterine space [ 85 ]. Although not usual, the sciatic nerve can also be involved in endometriosis in patients with DIE. Peripheral nerve involvement or compression by endometriosis is regarded as a cause of chronic pelvic pain symptoms, such as dysmenorrhea, non-cyclical pain, and deep dyspareunia [ 86 ]. Although some aspects of the pathogenesis of pain in endometriosis remain unclear, pathological intraneural and perineural invasion are observed more often in endometriosis patients with severe pelvic pain [ 86 ]. Additionally, presacral neurectomy is known to relieve these symptoms, although it presents with some risk of adverse effects [ 87 ]. Recognizing nerve involvement by endometriosis using conventional MR sequences might be difficult, but diffusion tensor images (DTIs) and magnetic resonance neurography might facilitate its visualization [ 88 89 90 ]. Manganaro et al. [ 89 ] reported an irregular and disorganized appearance of the sacral nerve root (S1, S2, and S3) on DTI in most patients with endometriosis. Furthermore, fractional anisotropy values calculated from DTI in the S1, S2, and S3 roots were significantly lower in patients than in control subjects [ 89 ]. Zhang et al. [ 90 ] used contrast-enhanced 3D short time inversion recovery sampling perfection with application-optimized contrasts using different flip angle evolution sequences to visualize compression and adhesion of the sacral plexus nerve as a result of ectopic endometriotic lesions. These new techniques may provide additional information for future clinical applications.

Laparoscopy with histological confirmation of ectopic endometrial tissue is the gold standard for the diagnosis of endometriosis [ 2 ]. Ultrasonography (US) is the first-line imaging modality for the assessment of pelvic endometriosis because of its easy accessibility, low degree of invasiveness, and cost-effectiveness [ 23 ]. Nevertheless, it is constrained by limitations such as limited field of view and operator dependence. Today, MRI is recognized as a valuable tool for diagnosis and presurgical planning. Although it is more expensive and time consuming, MRI has advantages that US does not: it is more objective, and the images can cover a large field-of-view with multiple directions [ 24 ]. Due to its excellent contrast resolution, the combination of several MR sequences can provide detailed information about the locations and histological characteristics of endometriosis. The European Society of Urogenital Radiology provides recommendations on the optimal MRI protocol and guidelines for the diagnosis of pelvic endometriosis based on evidence from the literature and the consensus of expert opinion ( Table 2 ) [ 9 ]. For patient preparation, 3–6 hours of fasting and bladder emptying one-hour before the examination are recommended [ 9 ]. A moderately distended bladder is preferable because an overly distended bladder might lead to oversight in detecting small endometriosis of the bladder [ 9 ]. Regarding the timing of the menstrual cycle, Botterill et al. [ 25 ] found no significant difference between menstruating and non-menstruating MRI scans for the evaluation of the extent or severity of endometriosis. An anti-peristaltic agent is recommended to prevent bowel motion artifacts, unless contraindicated [ 9 ]. Vaginal or rectal opacification by a gel is regarded as ‘optional’ [ 9 ]. Distention of the vagina or rectum by a gel might make pelvic endometriosis easier to find, especially for Douglas pouch or rectosigmoid colon endometriosis. Fiaschetti et al. [ 26 ] reported improved sensitivity for the detection of DIE with vaginal opacification, although Bazot et al. [ 27 ] found no significant difference. The invasiveness of the procedure and additional time necessary for patient preparation are also hurdles to be overcome. As a standard MR protocol, T2-weighted images (WIs) (axial, sagittal, and oblique axial) and T1WI (with and without fat suppression), as well as single-shot-fast-spin-echo images were considered. Oblique axial T2WI is useful for assessing USL endometriosis [ 28 ]. Contrast-enhanced T1WI, diffusion-weighted imaging (DWI), and susceptibility-weighted imaging are not recommended for the evaluation of ovarian endometriotic cysts and DIE [ 9 ]. Apparent diffusion coefficient (ADC) values were also not regarded as useful because no significant difference was found among various pelvic lesions, either with or without endometriosis [ 29 ]. However, contrast enhancement and DWI are recommended for patients with suspected ovarian malignancy [ 29 ]. Relevant details are presented below.

Endometriosis, a chronic inflammatory disease involving the growth of endometrium-like epithelium and stromal cells outside the uterus [ 1 ], is a common gynecological disease found in 2%–8% of the general population [ 2 ]. More than 80% of affected patients are of reproductive age [ 3 ]. Female with this disease usually present with pelvic pain, dysmenorrhea, and dyspareunia, all of which can strongly affect quality of life [ 3 4 ]. It is also closely related to infertility [ 3 ]. Early diagnosis and precise evaluation of the spread of lesions and severity of adhesions before and after treatment are important for formulating a treatment strategy. Three main entities of pelvic endometriosis have been identified: ovarian endometriotic cysts, superficial peritoneal endometriosis, and deep infiltrating endometriosis (DIE) [ 5 ]. Among them, magnetic resonance imaging (MRI), which plays an important role in the diagnosis of ovarian endometriotic cysts and DIE, is also useful for detecting characteristic appearances of various conditions or diseases associated with ovarian endometriotic cysts: decidual change induced by pregnancy, changes after hormonal treatment, and endometriosis-related tumors [ 5 6 7 8 9 ]. Regarding DIE, the limited diagnostic ability of physical examinations, such as digital vaginal examination by gynecologists, has been reported (36% accuracy, 46%–54% specificity) [ 5 ]. Moreover, MRI can assist gynecologists in evaluating the precise location of the endometriosis and severity of adhesions, which are strongly related to clinical symptoms such as dysmenorrhea, dyspareunia, pelvic pain, and infertility [ 5 ]. This review article presents a summary of the MRI findings of ovarian endometriotic cysts and their physiologically and pathologically related conditions. We also reviewed the key points of MRI in the diagnosis of DIE at different anatomical sites ( Table 1 ).

Endometriosis, a common chronic inflammatory disease affecting female of reproductive age, can severely affect patients’ quality of life as well as their fertility. Because of its high contrast and objectivity, MRI can contribute to the early and accurate diagnosis of ovarian endometriotic cysts and DIE while obviating the need for invasive procedures and radiation exposure. Furthermore, MRI plays a role in evaluating severity, leading to optimal treatment selection and preoperative planning. It is noteworthy that ovarian endometriotic cysts sometimes present with an atypical appearance, such as decidual change or polypoid endometriosis. Complications such as rupture, abscess, and endometriosis-related tumors should also be noted. For the diagnosis of DIE and secondary adhesion among pelvic organs, we look for ‘fibrosis’ with T2 hypointense nodules or thickened structures. Hemorrhagic foci on T1WI can be helpful, but they are infrequent. Indirect imaging findings of adhesion, such as kissing ovaries or intestinal tethering, are also important. Radiologists should be familiar with both common and uncommon locations of endometriosis, their characteristic imaging findings, and their relationship to disease severity and treatment selection.

Typical symptoms attributed to endometriosis are dysmenorrhea, dyspareunia (which can be related to the menstrual cycle), and infertility [ 3 4 ]. Although DIE is known to be closely related to these symptoms, ovarian and superficial peritoneal endometriosis are often asymptomatic [ 19 ]. Deeper endometriosis (greater than 10 mm) is associated with pain [ 3 19 20 ]. Dysmenorrhea and dyspareunia are the two most frequent symptoms of pain caused by endometriosis [ 10 ]. Dysmenorrhea is related to excessive intraperitoneal prostaglandin production by the ectopic endometrium, causing myometrial hypertonus and secondary ischemia [ 21 ]. Treatments were considered for patients with the symptoms presented above. Surgery is unnecessary if clear therapeutic benefits of the intervention cannot be found [ 10 ]. Overall, medical therapy with oral contraceptives (OCs) and progestins enables satisfactory long-term pain control in approximately two-thirds of symptomatic female [ 22 ]. In addition, OCs dramatically reduce the rate of recurrence of endometriotic cysts without damaging future fertility [ 22 ]. Regarding its relationship with infertility treatment, the benefit of surgery is limited, even in cases of advanced endometriosis [ 22 ].

The exact etiology of endometriosis remains unclear. The most prevalent theory is the retrograde menstruation phenomenon [ 3 10 ]. Sampson [ 11 ] proposed that the reflux of endometrial tissue through the fallopian tube by retrograde menstruation results in implantation and growth on the peritoneal surface. Many supporting observations are presented below. Endometriotic lesions are most commonly found in the area of the tubal ostia and the deepest portion of the Douglas pouch. Endometriosis is more common in female with early menarche and heavy or long menstrual flow [ 3 ]. Other theories include endometrial stem cell implantation, Mullerian remnant abnormalities that explain cul-de-sac endometriosis, and coelomic metaplasia. Mullerian remnant abnormalities of aberrant migration of the Mullerian duct into the posterior pelvic floor and coelomic metaplasia explain ovarian endometriosis by metaplastic changes of the coelomic epithelium into the endometrium [ 10 ]. Recent reports have implicated the ampullary portion of the fallopian tube as the origin of endometriosis, which is similar to that of high-grade serous carcinoma [ 12 ]. Other factors, such as female hormones and inflammation, are also considered to be deeply associated with the pathogenesis of endometriosis [ 10 ]. Estradiol promotes the growth of ectopic endometrial tissues. The presence of ectopic tissue causes inflammation with overproduction of prostaglandins, cytokines, and chemokines, consequently causing pain, fibrosis, and adhesion formation [ 10 ]. Research elucidating the condition’s putative association with genetic factors has also progressed, with the WNT4, CDKN2B-AS1, and GREB1 genes being strong targets for additional studies investigating endometriosis [ 10 ]. These genetic factors contribute to approximately half of all conditions carrying the risk of endometriosis [ 13 ].

Extra-pelvic endometriosis is a rare condition of endometriosis found at sites distant from the pelvic organs [ 91 ]. Although the ‘extra-pelvic’ area is large and the exact frequency of this disease remains unclear, abdominal wall endometriosis and thoracic endometriosis are frequent [ 91 92 ]. As for abdominal wall endometriosis, the frequent sites are the inguinal and umbilical regions [ 91 92 ]. A study of 105 patients with ‘abdominal/pelvic wall mass’ by Yarmish et al. [ 93 ] revealed that 24.8% (26/105) of them had histologically proven abdominal wall endometriosis. Such patients are often admitted to the hospital with cyclic abdominal pain, and are therefore referred to dermatology and not to gynecology [ 29 93 ]. In addition, the influence of factors such as prior surgery, including cesarean section and hysterectomy, are well known. A systematic review article noted that an earlier surgical history was recognized in approximately 80% of patients, but not in 20% of patients [ 94 ]. Spontaneous inguinal endometriosis is, in most cases, an extension of pelvic endometriosis along the round ligament [ 95 ]. Regarding imaging diagnosis, US is the first modality to be used because it is a superficial lesion. On US, the lesion shows hypoechoic round or oval nodules with ill-defined borders and a hyperechoic rim [ 96 ]. In fact, CT showed a sensitivity of 69% and specificity of 97% by a combination of the following three imaging features: location below the umbilicus, homogeneous density, and the presence of linear infiltration irradiating peripherally from a central soft tissue nodule, designated as the ‘gorgon’ sign [ 93 ]. The ‘gorgon’ sign reflects histological features of histiocytic infiltration and fibrosis because of chronic hemorrhage, as well as a few glands found in pelvic endometriosis [ 93 ]. Based on these histological findings, MRI showed a hypointense solid mass on T2WI with hyperintense hemorrhagic cysts or hyperintense hemorrhagic cysts on T1WI ( Fig. 12 ) [ 97 ]. Fat-saturated T1WI is recommended for the diagnosis of abdominal wall endometriosis [ 97 ]. Thoracic endometriosis is characterized by the presence of endometriotic lesions in the thoracic cavity [ 98 ]. The most common clinical presentation of thoracic endometriosis is catamenial pneumothorax (70%), followed by catamenial hemothorax and lung nodules [ 99 ]. Because CT is unable to detect lesions easily, especially in cases of small lesions, the gold standard for the diagnosis of thoracic endometriosis is video-assisted thoracoscopic surgery ( Fig. 13 ) [ 92 ]. For the evaluation of diaphragmatic endometriosis using MRI, the lesions were detected as hyperintense nodules on T1WI, which were mostly located at the right side and posterior with respect to the vena cava [ 100 ]. The MRI sensitivity was 83%, and the kappa value of the two readers was 0.86 [ 100 ].

Endometriosis is divided into three types: ovarian endometriotic cysts, superficial peritoneal endometriosis, and DIE. Among these, superficial peritoneal endometriosis is the least severe. In this type, endometrium-like tissue attaches to the peritoneum. Most cases are occult on imaging. The ovaries are the most frequent site of ectopic endometrial tissue, and ovarian endometriotic cysts occur. DIE is defined histologically as endometriosis infiltrating the peritoneum to a depth > 5 mm. The consensus of the Society of Abdominal Radiology Disease Focused Panel on endometriosis is that most endometriotic lesions visible on MRI are DIE [ 14 15 ]. Several classification and staging systems are used clinically, depending on the purpose, such as the revised American Society for Reproductive Medicine (rASRM) classification, ENZIAN classification, endometriosis fertility index, and American Association of Gynecological Laparoscopists classification [ 16 ]. Among them, rASRM classification is the most widely used. It is useful for physicians to report the severity of endometriosis (stages I–IV) in patients in simple terms [ 16 17 ]. This scoring system is based on the summation of values assigned to the following items: the size of the endometriosis lesions in the peritoneum and ovaries, adhesion on the ovaries and fallopian tubes, and partial or complete posterior cul-de-sac obliteration. The rASRM classification does not incorporate DIE infiltration into the retroperitoneal structures. As a supplement to the rASRM, the ENZIAN classification was developed to classify DIE [ 17 ]. Di Paola et al. [ 18 ] reported an excellent correlation between MRI-based ENZIAN scores and histopathological ENZIAN scores, particularly for rectovaginal septum, uterosacral ligament (USL), and rectosigmoid locations.