Endometriosis diagnosis: a retrospective study on the use of transvaginal ultrasound, sonovaginography and magnetic resonance imaging

Objective: To provide estimates of the diagnostic accuracy of transvaginal ultrasound, sonovaginography and magnetic re - sonance imaging for the diagnosis of ovarian endometriosis and deep endometriosis.

This was an observational, retrospective study carried out in a Portuguese tertiary center. Consecutive patients who underwent surgical procedures between 2010 and 2023 for diagnosis and treatment of endometriosis were included. The in - dex tests were transvaginal ultrasound, sonovaginography and magnetic resonance imaging. Accuracy of these tests to iden - tify lesions of endometriosis was assessed for different sites, relative to surgical findings. Data was obtained from imaging and surgical reports. Sensitivity , specificity , positive and negative predictive values were calculated for each diagnostic method.

Out of 301 patients who underwent surgery , 195 (64.8%) were evaluated with transvaginal ultrasound, 99 (32.9%) had a sonovaginography performed and 167 (55.5%) were evaluated with magnetic resonance imaging. All but 3 (1.0%) patients had endometriosis confirmed through surgical visualization. The sensitivity and specificity for ovarian endome - triosis were 95.2% and 83.1% for transvaginal ultrasound and 91.0% and 80.5% for magnetic resonance imaging, respec - tively . Regarding deep endometriosis, the sensitivity and specificity were 84.1% and 54.5% for sonovaginography and 90.3% and 60.0% for magnetic resonance imaging, respectively .

Transvaginal ultrasound was highly accurate for the diagnosis of endometriomas and, when used alongside so - novaginography it had important value in the diagnosis of deep endometriosis as well. Both sonovaginography and magne - tic resonance imaging showed similar performance in diagnosing deep endometriosis and stand as reliable diagnostic tools.

Endometriosis; Diagnosis; Diagnostic imaging; Ultrasonography; Magnetic Resonance Imaging. Endometriosis diagnosis: a retrospective study on the use of transvaginal ultrasound, sonovaginography and magnetic resonance imaging Diagnóstico de Endometriose: uma análise retrospetiva do uso de ecografia transvaginal, sonovaginografia e ressonância magnética pélvica Eva Maria de Jesus Paiva 1, Ana Mafalda da Costa Castro Neves 2, Pedro Viana Pinto 3, Ana Margarida Póvoa 4, Andreia T eixeira 5 Faculdade de Medicina da Universidade do Porto Centro Hospitalar Universitário de São João Resumo Objetivo: Obter estimativas da capacidade diagnóstica da ecografia transvaginal, sonovaginografia e ressonância magné - tica para o diagnóstico de endometriose ovárica e profunda. Métodos: Este foi um estudo observacional retrospetivo que decorreu num centro terciário português. Foram incluídos pacientes sujeitos a procedimentos cirúrgicos para o diagnóstico e tratamento de endometriose entre 2010 e 2023. A ca - pacidade da ecografia transvaginal, sonovaginografia e ressonância magnética para a identificação de lesões de endome - triose foi avaliada para diferentes localizações, tendo por referência os achados cirúrgicos. Os dados foram obtidos a partir dos relatos imagiológicos e cirúrgicos. Para cada método diagnóstico foram calculadas as seguintes medidas: sensibilida - de, especificidade, valores preditivos positivos e negativos. DOI: 10.69729/aogp.v19i4a05 Eva Maria de Jesus Paiva et al. Acta Obstet Ginecol Port 2025;19(4):268-279 269

E ndometriosis is a chronic disease characterized by the presence of endometrium-like tissue outside the endometrium, frequently accompanied by an asso- ciated inflammatory reaction 1. The exact prevalence of this disease is unknown, but it is thought to affect around 10% of women in their reproductive years 2. Endometriosis manifests mainly as three different subtypes: superficial implants in the peritoneum, ova - rian endometriosis (endometriomas) and deep-seated infiltration into nearby tissues and pelvic organs (deep infiltrating endometriosis, DE). According to an inter - national working group, DE is defined as “endome - trium-like tissue lesions in the abdomen, extending on or under the peritoneal surface; they are usually nodu - lar, able to invade adjacent structures, and associated with fibrosis and disruption of normal anatomy” 1. The - se subtypes may differ in symptom presentation and diagnostic approach 1,3 . The clinical manifestations ran - ge from dysmenorrhea and deep dyspareunia to dysu - ria, dyschezia and infertility 4. No symptom is specific or pathognomonic of endometriosis and its severity does not correlate to the surgical stage of the disease, which can remain undiagnosed for 8 to 12 years 5. This delay leads to long- lasting pain, reduced quality of life, psychological stress, and compromised fertility 6. Diagnosing endometriosis remains a challenge for clinicians 7. Laparoscopic observation and biopsy are considered the gold standard for the diagnosis although identifiable lesions may not always be histologically confirmed 4,5 . Physical examination alone has low diag - nostic accuracy , so imaging is widely used to establish a diagnosis and evaluate the extent of the disease as surgery is invasive and costly 3,4 . Surgical intervention is usually reserved for women with disease non-res - ponsive to medical treatment, infertility and severe di - sease (interfering with organ function) so noninvasive imaging exams assume a preponderant role 6. Transvaginal ultrasonography (TVUS) allows for real- -time assessment of the uterus, adnexa and surrounding structures while granting dynamic evaluation of organ mo - bility hence it is the first- line imaging modality to use when suspecting of endometriosis 8-10 . TVUS detection is greatly dependent on the experience of the operator and the lo - cation of the endometriotic lesions, being especially use - ful in patients with endometriomas or DE 9,11,12 . Sonovagi - nography (SVG), introduced by Dessole et al in 2003, is a Resultados: De 301 doentes submetidos a cirurgia, 195 (64,8%) realizaram ecografia transvaginal, 99 (32,9%) efetuaram sonovaginografia e 167 (55,5%) foram avaliados com ressonância magnética. O diagnóstico de endometriose foi confir - mado, através de visualização direta, em todos os pacientes exceto 3 (1,0%). A sensibilidade e especificidade da ecografia transvaginal para o diagnóstico de endometriose ovárica foram 95,2% e 83,1% e da ressonância magnética foram 91,0% e 80,5%, respetivamente. Relativamente ao diagnóstico de endometriose profunda, obtivemos sensibilidade e especificida - de de 84,1% e 54,5% para a sonovaginografia e de 90,3% e 60,0% para a ressonância magnética, respetivamente. Conclusão: A ecografia transvaginal apresentou um desempenho excelente no diagnóstico de endometriomas e, quando usada em conjunto com a sonovaginografia, revelou grande valor para o diagnóstico de endometriose profunda. Tanto a sonovaginografia como a ressonância magnética obtiveram resultados semelhantes para o diagnóstico de endometriose profunda e afirmam-se como métodos de imagem confiáveis. Palavras-chave: Endometriose; Diagnóstico; Imagiologia diagnóstica; Ultrassonografia; Imagem de Ressonância Magnética. 1. Faculdade de Medicina, Universidade do Porto, Porto, Portugal. 2. Department of Obstetrics and Gynecology , Centro Hospitalar Universitário de São João, Porto, Portugal. 3. Department of Obstetrics and Gynecology , Centro Hospitalar Universitário de São João, Porto, Portugal; Department of Anatomy , Faculdade de Medi - cina, Universidade do Porto, Porto, Portugal; Department of Obstetrics and Gynecology , Faculdade de Medicina, Universidade do Porto, Porto, Portugal. 4. Department of Obstetrics and Gynecology , Centro Hospitalar Universitário de São João, Porto, Portugal; Department of Obstetrics and Gynecology , Fa - culdade de Medicina, Universidade do Porto, Porto, Portugal. 5. MEDCIDS – Department of Community Medicine, Information and Decision in Health; Faculdade de Medicina, Universidade do Porto, Porto, Portu - gal; CINTESIS@RISE – Center for Health Technology and Services Research; Faculdade de Medicina, Universidade do Porto, Porto, Portugal; ADiT-LAB, Instituto Politécnico de Viana do Castelo. Endometriosis diagnosis: a retrospective study on the use of transvaginal ultrasound, sonovaginography and magnetic resonance imaging 270 Acta Obstet Ginecol Port 2025;19(4):268-279 contrast-enhanced transvaginal ultrasound using saline solution or gel to create an acoustic window between the probe and surrounding structures. It is used as an addi - tional tool to identify and characterize DE 9,13-15 . Magnetic resonance imaging (MRI) is recommended as a further step in the diagnostic workup of endometriosis following an equivocal ultrasonography evaluation and in case of ex - tensive disease 16 . Infiltrating lesions are more precisely mapped as it has a larger field of view and high tolerabili - ty therefore significantly helping preoperative planning 9. The high accuracy of such non-invasive modalities guarantees their use as standard, enabling early diag - nosis which in turn facilitates medical treatment and ensures optimal planning for adequate surgical mana - gement, anticipating its difficulty and helping deter - mine when a multidisciplinary team is needed 7,17 . In this study , we aim to provide estimates of the diag - nostic accuracy of these non-invasive imaging methods for the diagnosis of endometriomas and deep endo - metriosis in patients who had surgical procedures in a tertiary referral center and to compare the performan - ce of these modalities in accurately mapping lesions and assessing their extent.

This was an observational, retrospective cohort study carried out in a tertiary center in Portugal and com - pliant with the STARD (Standards for Reporting of Diagnostic Accuracy) guidelines. The study was approved by the local ethics committee. All patients who underwent surgical procedures (laparoscopy or laparotomy) for diagnosis and treatment of endome - triosis between 2010 and 2023 were eligible for inclu - sion and we reviewed the unit’s surgical database to se - lect them. These patients were followed in this hospi - tal and had clinical or imaging suspicion of ovarian or deep endometriosis. We excluded those who had ex - trapelvic endometriosis or adenomyosis alone. The archived data about clinical history , transvagi - nal ultrasound, sonovaginography , magnetic resonan - ce imaging and surgery results were collected from elec - tronic medical records (SClinico and BHealth). The topography of the endometriotic lesions was ob - tained from imaging and surgical reports regarding an - terior compartment involvement (distal ureters and bladder), posterior compartment involvement (utero - sacral ligaments; uterine torus, rectovaginal septum and posterior vaginal wall), intestinal involvement (rec - tum, sigmoid and rectosigmoid transition), and pre - sence of ovarian endometriomas. For each location, data were classified as positive or negative for the de - tection of endometriosis in every test. For every patient, the following anamnestic data were collected: age, height and weight, smoking status, age at menarche, parity and gravidity , history of previous surgeries and medical conditions. Clinical symptoms of endometriosis such as dysmenorrhea, dyspareunia or infertility were also documented as well as if the patient was referred due to an incidental imaging finding. As per this study’s retrospective methodology , all those performing imaging studies or surgery had access to clinical information regarding the patient at the time of execution of the respective procedures. Pa - tients underwent each imaging exam as their assistant gynecologist saw appropriate, considering each case. All reports were analyzed, and the relevant findings were adequately noted. Some reports did not provide information on endometriosis findings regarding eve - ry topographic location considered in this study . In tho - se cases, we retrieved the available information and missing results for each location were excluded from the analysis. When the report expressed uncertainty regarding the presence of endometriosis in a given lo - cation, we considered it a positive finding. Transvaginal ultrasound and sonovaginography Transvaginal ultrasound and sonovaginography were mostly performed in the gynecology department by the same two gynecologists, with extensive expertise in US diagnosis of endometriosis; only a minority of patients had an US executed by their doctor within the clinical appointment. During the first years of the study, a Voluson™ 730 (General Electric, USA) transvaginal 5-9 MHz probe was used; from 2021 a Voluson™ S8 (General Electric, USA) was used. No bowel preparation was used before sonography or sonovaginography . Sonographic assessment was routinely done accor - ding to IDEA group consensus since its publication 11 . Anatomic sites in the anterior and posterior com- partments were evaluated for possible endometriotic Eva Maria de Jesus Paiva et al. Acta Obstet Ginecol Port 2025;19(4):268-279 271 findings like a hypoechoic thickening, hypoechoic irregular nodules, pouch of Douglas obliteration and pelvic adhesions. Obliteration of the vesicouterine or rectovaginal pouches was assessed by gently pressing the uterus with the probe (anterior and posterior sli - ding signs). Endometriomas were described according to the International Ovarian T umor Analysis (IOTA) terminology and typically appear on ultrasound as uni - locular, round and with regular margins, fluid-filled cysts with ground glass homogeneous echogenicity (Fi - gure 1A) but can also have atypical features like multi - ple locules and papillations 18 . DE lesions in the anterior compartment were diagnosed when finding hypoe - choic lesions extending farther than the bladder sero - sa or ureter strictures. In the posterior compartment, DE was identified when observing hypoechoic solid nodules with heterogeneous size and margins or hy - poechoic thickening of the vagina or intestinal wall (Fi - gure 1B). Sonovaginography was performed when clinicians deemed it appropriate to better characterize DE lesions. Acoustic window was obtained by placing 50 ml of ul - trasound gel in the vagina to distend its walls and enhance visualization of the vaginal walls and fornix, uterosacral ligaments, pouch of Douglas and rectova - ginal septum. Each examination also included a tran - sabdominal ultrasound with the bladder filled in order to better evaluate the presence of any relevant lesion. Magnetic resonance imaging Pelvic MRI was performed as per hospital protocol, using hyoscine butylbromide to reduce intestinal pe - ristalsis and contrast if suspecting of an atypical endo - metrioma in order to exclude potential malignancy . Pa - tients did not undergo any bowel preparation. The pro - tocol also included administration of sterile vaginal ul - trasound gel. The majority of the exams were performed using a Siemens MAGNETOM Vida™ de - vice. A minority of the patients had MRI reports from external facilities, but these cases were discussed in group meetings before surgery . From 2020 all cases were reviewed by a multidisciplinary specialized group, including radiologists specialized in pelvic imaging and gynecologists with both ultrasound and surgical expe - rience in dealing with endometriosis. DE can be detected on MRI by an array of findings ran - ging from ectopic endometrial nodules to thickened struc - tures, fibrosis or changes to the usual pelvic anatomy; FIGURE 1. Imaging findings of endometriosis. Endometriosis diagnosis: a retrospective study on the use of transvaginal ultrasound, sonovaginography and magnetic resonance imaging 272 Acta Obstet Ginecol Port 2025;19(4):268-279 the implants typically appear as hyperintense areas foci on T1 with fat suppression and hypointense in T2-weighted MR images on the respective locations (Figure 1C). Fur- thermore, obliteration of pelvic pouches, disappearance of fat tissue planes and ureteral dilation are also signs of the disease. Endometriomas were diagnosed as hyperin - tense ovarian cysts on T1-weighted images and hypoin - tense on T2W (Figure 1D). Surgery All surgical interventions were performed in this hospital by a skilled set of gynecologists, with experience in ad - vanced laparoscopic surgery . This was considered the re - ference standard for the analysis as it is still considered the gold standard for the diagnosis of endometriosis 5. Diag - nosis of DE was made when at least one of the above-men - tioned structures was involved; the location of all suspi - cious lesions was recorded and removed for histological confirmation. T ypical lesions like dark “powder-burn” no - dules, implants or cysts with thick content were visually identified on the pelvic structures and considered as DE when extending > 5 mm beneath the peritoneum. Spicu - lated nodules and thickening of pelvic structures were also evaluated for the presence of the disease. Ovarian endo - metriomas were identified as blood-filed cysts often ad - herent to the uterus or surrounding ligaments. Statistical analysis Statistical analysis was performed using SPSS version 29. For each diagnostic tool, the sensitivity , specificity , posi - tive predictive value and negative predictive value with their respective 95% confidence interval were calculated for individual and grouped locations. We performed a se - condary analysis of diagnostic accuracy , calculating sen - sitivity and specificity in the group of patients who un - derwent TVUS complemented with SVG; TVUS versus MRI for endometrioma diagnosis and SVG versus MRI for DE detection. To compare the diagnostic perfor - mance of these methods, the McNemar’s test was used and a p value <0.05 was considered significant.

Table I presents the demographic data of our population as well as their clinical symptoms and surgical findings. In total, 301 patients were submitted to surgery due to en - dometriosis and were included in the analysis. The mean age of our patients was 36.6 years (range 18-61) with a median BMI of 23.9 Kg/m 2. We found dysmenorrhea to be the most common symptom of our patients (69.0%), followed by deep dyspareunia (44.9%). At least 64 pa - tients (22.9%) had infertility and 40 (13.3%) had un - dergone previous endometriosis surgery . Regarding surgeries performed, 285 (94.7%) patients had laparoscopic surgery while the remaining 16 (5.3%) had a laparotomy performed. All but 3 (1.0%) patients had endometriosis confirmed through surgical visuali - zation. Based on surgical records, we found that 180 (59.8%) patients had endometriomas and 207 (68.8%) DE. Both subtypes were present in 90 (29.9%) patients. Regarding temporal trends, 25.2% of surgeries were performed between 2010 and 2014, 35.2% between 2015 and 2019, and 39.5% between 2020 and 2023. Transvaginal ultrasound Out of the 301 patients included in this study , 195 (64.8%) were evaluated with TVUS. The diagnostic per - formance of TVUS in predicting endometriosis in dis - tinct locations is presented in Table II. Regarding ovarian endometriosis, the sensitivity of TVUS for endometrio - mas was 95.2% (120/126) (95% CI 91.5-99.0) with spe - cificity of 83.1% (49/59) (95% CI 73.5-92.6); however, regarding DE (both anterior and posterior compartment lesions), only 24 out of 124 patients with surgical fin - dings of DE were correctly diagnosed. Sonovaginography A total of 99 (32.9%) patients had an SVG performed, and its diagnostic performance is reported in Table II. Regarding overall DE, SVG correctly identified 74 out of 88 patients who had DE diagnosed during surgery . In the anterior compartment, SVG missed DE in 18 patients which resulted in a sensitivity of 14.3% (3/21) (95% CI 0.0-29.3); however, for the posterior compartment, the overall sensitivity was 83.9% (73/87) (95% CI 76.2-91.6). In this compartment, diagnostic sensitivity of endometriosis was highest for uterine torus, rectova - ginal septum and posterior vaginal wall at 73.6% (53/72) (95% CI 63.4-83.8) compared to uterosacral ligaments (37.1% (26/70); 95% CI 25.8-48.5) and intestinal endometriosis (36.4% (16/44); 95% CI 22.2-50.6). Eva Maria de Jesus Paiva et al. Acta Obstet Ginecol Port 2025;19(4):268-279 273 Comparison of TVUS and SVG – SVG as a complement of TVUS SVG was used as a complement of TVUS when sus - pecting DE, as was considered pertinent. For that rea - son, out of 99 patients who performed gel SVG, 94 also had available records of TVUS. In this smaller group, we compared the diagnostic performance of both ul - trasound techniques separately and considering their findings together, as that is more resembling of what happens in clinical practice. Our results can be found in Table III. TVUS alone was more sensitive for endo - metrioma detection reaching a sensitivity of 95.7% (44/46) (95% CI 89.8-100) and specificity of 91.3% (42/46) (95% CI 83.2-99.5) while SVG alone was more sensitive for the diagnosis of DE achieving a sensitivi - ty of 83.1% (69/83) (95% CI 75.1-91.2) and specifici - ty of 54.5% (6/11) (95% CI 25.1-84.0). Magnetic resonance imaging Amongst our population, 167 (55.5%) patients were evaluated with MRI. The diagnostic performance of this imaging study in diagnosing endometriosis in specific pelvic locations is reported in Table IV . For endome - trioma detection, MRI showed a sensitivity of 91.0% (61/67) (95% CI 84.2-97.9) and specificity of 80.5% (62/77) (95% CI 71.7-89.4); for DE, MRI correctly TABLE I. DEMOGRAPHIC CHARACTERISTICS, CLINICAL SYMPTOMS AND SURGICAL FINDINGS. Patient characteristics N=301 Age, years, M (SD) 36.6 (6.5) Weight (N=236), Kg, Med [Q1;Q3] 63.0 [56.0;71.0] Height (N=188), m, Med [Q1;Q3] 1.63 [1.60;1.67] BMI (N=178), Kg/m 2, Med [Q1;Q3] 23.9 [21.2;26.6] Age at menarche (N=195), years, M (SD) 12.2 (1.7) Gravidity (N=279), Med [Q1;Q3] 1 [0;2] Parity (N=279), Med [Q1;Q3] 0 [0;1] Smokers (N=146), n (%) 28 (19.2) Symptoms Dysmenorrhea (N=274), n (%) 189 (69.0) Deep dyspareunia (N=274), n (%) 123 (44.9) Infertility (N=279), n (%) 64 (22.9) Previous endometriosis surgery, n (%) 40 (13.3) Previous abdominal/pelvic surgery (n=281), n (%) 127 (45.2) Surgical findings Ovarian endometriosis, n (%) 180 (59.8) Right, n (%) 56 (18.6) Left, n (%) 75 (24.9) Bilateral, n (%) 49 (16.3) Deep endometriosis, n (%) 207 (68.8) Anterior compartment, n (%) 52 (17.3) Bladder, n (%) 35 (11.6) Ureters, n (%) 19 (6.3) Posterior compartment, n (%) 198 (65.8) Uterosacral ligaments, n (%) 150 (49.8) Uterine torus, rectovaginal septum and posterior vaginal wall, n (%) 158 (52.5) Bowel, n (%) 99 (32.9) Rectum, n (%) 84 (27.9) Rectosigmoid transition, n (%) 7 (2.3) Sigmoid colon, n (%) 29 (9.6) BMI: body mass index Endometriosis diagnosis: a retrospective study on the use of transvaginal ultrasound, sonovaginography and magnetic resonance imaging 274 Acta Obstet Ginecol Port 2025;19(4):268-279 TABLE II. DIAGNOSTIC PERFORMANCE OF TRANSVAGINAL ULTRASOUND AND SONOVAGINOGRAPHY FOR PREOPERATIVE DIAGNOSIS OF ENDOMETRIOSIS. Sensitivity Specificity PPV NPV Accuracy Site TP (n) FP (n) TN (n) FN (n) N (95% CI) (95% CI) (95% CI) (95% CI) (95% CI) (%) (%) (%) (%) (%) TVUS Ovarian endometriosis 120 10 49 6 185 95.2 83.1 92.3 89.1 91.4 (91.5-99.0) (73.5-92.6) (87.7-96.9) (80.9-97.3) (87.3-95.4) Deep endometriosis 24 1 38 100 163 19.4 97.4 96.0 27.5 38.0 (12.4-26.3) (92.5-100) (88.3-100) (20.1-35.0) (30.6-45.5) Anterior compartment 4 1 122 23 150 14.8 99.2 80.0 84.1 84.0 (1.4-28.2) (97.6-100) (44.9-100) (78.2-90.1) (78.1-89.9) Bladder 3 1 128 17 149 15.0 99.2 75.0 88.3 87.9 (0.0-30.7) (97.7-100) (32.6-100) (83.0-93.5) (82.7-93.2) Ureters 1 1 137 7 146 12.5 99.3 50.0 95.1 94.5 (0.0-35.4) (97.9-100) (0.0-100) (91.6-98.7) (90.8-98.2) Posterior compartment 23 1 44 95 163 19.5 97.8 95.8 31.7 41.1 (12.3-26.6) (93.5-100) (87.8-100) (23.9-39.4) (33.6-48.7) Uterosacral ligaments 11 0 62 82 155 11.8 100 100 43.1 47.1 (5.3-18.4) (100-100) (100-100) (35.0-51.1) (39.2-55.0) Uterine torus, rectovaginal septum 17 3 62 75 157 18.5 95.4 85.0 45.3 50.3 and posterior vaginal wall (10.6-26.4) (90.3-100) (69.4-100) (36.9-53.6) (42.5-58.1) Bowel 6 1 99 48 154 11.1 99.0 85.7 67.3 68.2 (2.7-19.5) (97.1-100) (59.8-100) (59.8-74.9) (60.8-75.5) SVG Deep endometriosis 74 5 6 14 99 84.1 54.5 93.7 30.0 80.8 (76.5-91.7) (25.1-84.0) (88.3-99.0) (9.9-50.1) (73.1-88.6) Anterior compartment 3 3 73 18 97 14.3 96.1 50.0 80.2 78.4 (0.0-29.3) (91.7-100) (10.0-90.0) (72.0-88.4) (70.2-86.6) Bladder 2 3 78 13 96 13.3 96.3 40.0 85.7 83.3 (0.0-30.5) (92.2-100) (0.0-82.9) (78.5-92.9) (75.9-90.8) Ureters 0 2 85 7 94 0.0 97.7 0.0 92.4 90.4 (0.0-0.0) (94.6-100) (0.0-0.0) (87.0-97.8) (84.5-96.4) Posterior compartment 73 6 6 14 99 83.9 50.0 92.4 30.0 79.8 (76.2-91.6) (21.7-78.3) (86.6-98.3) (9.9-50.1) (71.9-87.7) Uterosacral ligaments 26 6 19 44 95 37.1 76.0 81.3 30.2 47.4 (25.8-48.5) (59.3-92.7) (67.7-94.8) (18.8-41.5) (37.3-57.4) Uterine torus, rectovaginal septum 53 15 10 19 97 73.6 40.0 77.9 34.5 64.9 and posterior vaginal wall (63.4-83.8) (20.8-59.2) (68.1-87.8) (17.2-51.8) (55.5-74.4) Bowel 16 5 49 28 98 36.4 90.7 76.2 63.6 66.3 (22.2-50.6) (83.0-98.5) (58.0-94.4) (52.9-74.4) (57.0-75.7) TP: true positive; FP: false positive; TN: true negative; FN: false negative; PPV: positive predictive value; NPV: negative predictive value Eva Maria de Jesus Paiva et al. Acta Obstet Ginecol Port 2025;19(4):268-279 275 diagnosed 131 patients with a sensitivity of 90.3% (131/145) (95% CI 85.5-95.2) and specificity of 60.0% (9/15) (95% CI 35.2-84.8). MRI had lower sensitivity and higher specificity in the anterior compartment in contrast to the posterior compartment. Ovarian Endometriosis – Comparison of TVUS and MRI In an effort to compare the performance of TVUS and MRI in diagnosing ovarian endometriosis, we analyzed the 91 patients who had both exams performed pre- operatively . In this group, 51 (56.0%) patients had ova - rian endometriomas according to surgery . We found that they performed similarly as sensitivity and speci - ficity were 90.2% (46/51) (95% CI 82.0-98.4) and 87.5 % (35/40) (95% CI 77.3-97.8) for TVUS and 94.1% (48/51) (95% CI 87.7-100) and 77.5% (31/40) for MRI (95% CI 64.6-90.4), respectively (Table V). Deep Endometriosis – Comparison of SVG and MRI In the same way , for comparison of the diagnostic ac - curacy of SVG and MRI in identifying DE lesions, we analyzed the patients who undertook both these tests prior to surgery . Out of 49 patients, all except for 2 (4.1%) had DE upon surgical evaluation. Lesions were observed in the posterior compartment of all the pa - tients and 12 (24.5%) of them had simultaneous fin - dings of DE in the anterior compartment. In Table V we summarize the diagnostic performance of SVG and MRI in this group of patients. In both exams, each patient without DE was classi - fied as having it on the posterior compartment and for this reason, specificity for this location was 0.0%. Con - sidering overall DE, MRI and SVG performed similar - ly . Only for the detection of lesions on the uterosacral ligaments, MRI appeared to be slightly more sensitive with 63.4% (26/41) (95% CI 48.7-78.2) compared to 31.7% (13/41) (95% CI 17.5-46.0) when using SVG.

In this study , we analyzed the diagnostic accuracy of TABLE III. DIAGNOSTIC PERFORMANCE OF TVUS, SVG AND TVUS+SVG FOR PREOPERATIVE DIAGNOSIS OF ENDOMETRIOSIS. Site TVUS SVG p TVUS+SVG Ovarian endometriosis Sens 95.7 (89.8-100) 78.3 (66.3-90.2) 0.002 95.7 (89.8-100) Spec 91.3 (83.2-99.5) 95.7 (89.8-100) 91.3 (83.2-99.5) Deep endometriosis Sens 3.6 (0.0.7.6) 83.1 (75.1-91.2) <0.001 83.1 (75.1-91.2) Spec 90.9 (73.9-100) 54.5 (25.1-84.0) 54.5 (25.1-84.0) Anterior compartment Sens 0.0 (0.0-0.0) 10.5 (0.0-24.3) 0.125 10.5 (0.0-24.3) Spec 98.6 (96.0-100) 95.9 (91.5-100) 95.9 (91.5-100) Bladder Sens 0.0 (0.0-0.0) 7.1 (0.0-20.6) 0.250 7.1 (0.0-20.6) Spec 98.7 (96.3-100) 96.2 (92.0-100) 96.2 (92.0-100) Ureters Sens 0.0 (0.0-0.0) 0.0 (0.0-0.0) 1.000 0.0 (0.0-0.0) Spec 98.8 (96.5-100) 97.6 (94.4-100) 97.6 (94.4-100) Posterior compartment Sens 3.7 (0.0-7.7) 82.9 (74.8-91.1) <0.001 82.9 (74.8-91.1) Spec 91.7 (76.0-100) 50.0 (21.7-78.3) 50.0 (21.7-78.3) Uterosacral ligaments Sens 1.5 (0.0-4.4) 37.3 (25.7-48.9) <0.001 37.3 (25.7-48.9) Spec 100 (100-100) 76.0 (59.3-92.7) 76.0 (59.3-92.7) Uterine torus, rectovaginal septum Sens 4.4 (0.0-9.3) 72.1 (61.4-82.7) <0.001 72.1 (61.4-82.7) and posterior vaginal wall Spec 96.0 (88.3-100) 40.0 (20.8-59.2) 40.0 (20.8-59.2) Bowel Sens 2.4 (0.0-7.2) 34.1 (19.6-48.7) <0.001 34.1 (19.6-48.7) Spec 100 (100-100) 90.4 (82.4-98.4) 90.4 (82.4-98.4) Data reported as % (95% CI). Sens: Sensitivity; Spec: Specificity Endometriosis diagnosis: a retrospective study on the use of transvaginal ultrasound, sonovaginography and magnetic resonance imaging 276 Acta Obstet Ginecol Port 2025;19(4):268-279 TABLE IV. DIAGNOSTIC PERFORMANCE OF MAGNETIC RESONANCE IMAGING FOR PREOPERATIVE DIAGNOSIS OF ENDOMETRIOSIS. Sensitivity Specificity PPV NPV Accuracy Site TP (n) FP (n) TN (n) FN (n) N (95% CI) (95% CI) (95% CI) (95% CI) (95% CI) (%) (%) (%) (%) (%) Ovarian endometriosis 61 15 62 6 144 91.0 80.5 80.3 91.2 85.4 (84.2-97.9) (71.7-89.4) (71.3-89.2) (84.4-97.9) (79.7-91.2) Deep endometriosis 131 6 9 14 160 90.3 60.0 95.6 39.1 87.5 (85.5-95.2) (35.2-84.8) (92.2-99.1) (19.2-59.1) (82.4-92.6) Anterior compartment 20 5 101 12 138 62.5 95.3 80.0 89.4 87.7 (45.7-79.3) (91.3-99.3) (64.3-95.7) (83.7-95.1) (82.2-93.2) Bladder 15 5 108 8 136 65.2 95.6 75.0 93.1 90.4 (45.8-84.7) (91.8-99.4) (56.0-94.0) (88.5-97.7) (85.5-95.4) Ureters 3 3 118 8 132 27.3 97.5 50.0 93.7 91.7 (1.0-53.6) (94.8-100) (10.0-90.0) (89.4-97.9) (87.0-96.4) Posterior compartment 127 9 9 15 160 89.4 50.0 93.4 37.5 85.0 (84.4-94.5) (26.9-73.1) (89.2-97.6) (18.1-56.9) (79.5-90.5) Uterosacral ligaments 68 11 31 28 138 70.8 73.8 86.1 52.5 71.7 (61.7-79.9) (60.5-87.1) (78.4-93.7) (39.8-65.3) (64.2-79.3) Uterine torus, rectovaginal septum 101 21 19 15 156 87.1 47.5 82.8 55.9 76.9 and posterior vaginal wall (81.0-93.2) (32.0-63.0) (76.1-89.5) (39.2-72.6) (70.3-83.5) Bowel 50 3 65 23 143 66.7 95.6 94.3 72.2 80.4 (56.0-77.3) (90.7-100) (88.1-100) (63.0- 81.5) (73.9-86.9) TP: true positive; FP: false positive; TN: true negative; FN: false negative; PPV: positive predictive value; NPV: negative predictive value TVUS, SVG and MRI, three wide - ly used imaging exams for the diagnosis of endometriosis. Their diagnostic performance varies across published literature as the - se are all dependent on operator or interpreter experience. TVUS and pelvic MRI were highly accurate for the diagnosis of ovarian endometriosis. Our results align with those from a recent sys - tematic review which reported a sensitivity of 70.86-96% and spe - cificity of 71-96% for TVUS as well as a sensitivity of 76.9-94% and specificity of 71-93.9% for MRI 19 . Even though our study was retros - pective, our ultrasound results for endometrioma detection also match those from a recent multi - center prospective study , streng- thening our findings 20 . Regarding DE detection using TVUS, our re - sults for its diagnostic sensitivity are lower than reported in the mentioned studies which we attri - bute to their more robust prospec - tive methodology . Furthermore, as this center regularly uses SVG for DE assessment, TVUS reports may be less descriptive for these kinds of lesions. We attribute this lower- than- expected result to these fac - tors, as well as to others such as the lengthy data collection period. Over the past decade, substantial research and innovation have dri - ven major advances in imaging technology and more standardised reporting practices. In the future, it would be interesting to investiga - te the extent to which these deve - lopments, such as IDEA group consensus adoption, since its pu - blication may have impacted diag - nostic acuity of TVUS over time in Eva Maria de Jesus Paiva et al. Acta Obstet Ginecol Port 2025;19(4):268-279 277 our centre. As with any operator-dependent imaging exam we are also always conditioned by possible tech - nical errors, which can introduce variability . Nonethe - less, the specificity of TVUS for DE detection was high, suggesting that a positive result is highly indicative of DE but a negative finding does not exclude it, in line with the most recent ESHRE guidelines for diagnostic imaging 4. SVG showed good diagnostic accuracy for overall DE although, when considering specific sites, it was lo - wer than what is described in the literature 15 . Barra et al obtained a sensitivity of 89.4% and specificity of 79.4% for posterior DE using gel SVG while our sensi - tivity was 83.9% and specificity of 50.0% for the same location 21 . Our lack of specificity seems to be mainly driven by higher false positive rates for endometriosis in the uterine torus, rectovaginal septum and posterior vaginal wall. Even though we have a good detection rate for overall lesions, this lack of specificity means that there is still an opportunity to improve with this diagnostic tool. Based on our comparison of the diag - nostic accuracy of TVUS and SVG, we do believe we can conclude that, by performing these exams simul - taneously as a routine, our rates of detection of ovarian endometriosis (sensitivity 95.7% (95%CI 89.8-100) and specificity 91.3% (95%CI 83.2-99.5)) and DE (sensitivity 83.1% (95%CI 75.1-91.2) and specificity 54.5% (95%CI 25.1-84.0)) are reassuring for our pa - tients. For overall DE diagnosis by MRI, we obtained a sen - sitivity of 90.3% (95%CI 85.5-95.2) and specificity of 60.0% (95%CI 35.2-84.8) which is quite comparable to what Nisenblat et al reported of overall sensitivity of 94% (95%CI 90-97) and specificity of 77% (95%CI 44-100%) 17 . Similarly to SVG, our specificity was wor - se particularly due to higher false positive rates for the uterine torus, rectovaginal septum and posterior vagi - nal wall. When focusing on specific locations, we ob - tained slightly worse results than those described in the literature 8,17,22 . This suggests that while MRI shows TABLE V. DIAGNOSTIC PERFORMANCE OF TVUS AND MRI FOR PREOPERATIVE DIAGNOSIS OF OVARIAN ENDOMETRIOSIS AND DIAGNOSTIC PERFORMANCE OF SVG AND MRI FOR PREOPERATIVE DIAGNOSIS OF DEEP ENDOMETRIOSIS. Site TVUS MRI p Ovarian endometriosis (N=91) Sens 90.2 (82.0-98.4) 94.1 (87.7-100) 0.146 Spec 87.5 (77.3-97.8) 77.5 (64.6-90.4) SVG RM Deep endometriosis (N=49) Sens 89.4 (80.6-98.2) 85.1 (74.9-95.3) 0.774 Spec 0 (0.0-0.0) 0.0 (0.0-0.0) Anterior compartment Sens 16.7 (0.0-37.8) 50.0 (21.7-78.3) 0.453 Spec 94.6 (87.3-100) 97.3 (92.1-100) Bladder Sens 10.0 (0.0-28.6) 50.0 (19.0-81.0) 0.219 Spec 94.9 (88.0-100) 94.9 (88.0-100) Ureters Sens 0 (0.0-0.0) 0.0 (0.0-0.0) 0.500 Spec 95.5 (89.3-100) 100 (100-100) Posterior compartment Sens 89.4 (80.6-98.2) 85.1 (74.9-95.3) 0.774 Spec 0 (0.0-0.0) 0.0 (0.0-0.0) Uterosacral ligaments Sens 31.7 (17.5-46.0) 63.4 (48.7-78.2) 0.043 Spec 57.1 (20.5-93.8) 85.7 (59.8-100) Uterine torus, rectovaginal septum Sens 73.7 (59.7-87.7) 78.9 (66.0-91.9) 1.000 and posterior vaginal wall Spec 10.0 (0.0-28.6) 20.0 (0.0-44.8) Bowel Sens 50.0 (30.8-69.2) 69.2 (51.5-87.0) 1.000 Spec 82.6 (67.1-98.1) 100 (100-100) Data reported as % (95% CI). Sens: Sensitivity; Spec: Specificity Endometriosis diagnosis: a retrospective study on the use of transvaginal ultrasound, sonovaginography and magnetic resonance imaging 278 Acta Obstet Ginecol Port 2025;19(4):268-279 an adequate overall accuracy for diagnosing DE, there is room for optimizing its performance at specific site detection of lesions to help achieve a more precise mapping of the disease. Regarding the anterior compartment, the lower pre - valence of disease limits the interpretation of our re - sults. Reassuringly , the high specificity found in both SVG and MRI, allows us to be confident when disease is diagnosed in these exams. In the direct comparison between SVG and MRI, re - sults were quite similar. However, for disease in the ute - rosacral ligaments, MRI seems to have a higher sensi - bility with similar specificity to SVG. This means that, in cases with high clinical suspicion of disease in this location, MRI seems to have an important role. Ultrasound has several advantages to MRI, such as being a dynamic, innocuous and cost-effective exam. As such, its role as a first-line modality for screening and diagnosis of endometriosis is well established 10 . TVUS serves a purpose beyond being a diagnostic test, as it can help to understand the extent and site of lesions by taking into consideration real-time tenderness and mobility of the structures, giving an idea about the exis - tence of adhesions and obliteration of the pelvic pou - ches 7. However, it relies extensively on the operator’s experience and its images cannot be adequately revie - wed after the examination by other clinicians. Further - more, TVUS may be refused or impossible to perform in some patients. On the other hand, MRI provides ac - curate identification of endometriotic lesions, particu - larly in the posterior compartment and some major be - nefits of its use are the simultaneous multiple plane as - sessment and its capability to allow the retrospective re - view of images, which is most useful for preoperative workup 9. Moreover, it allows the examination of organs and regions not accessible to the ultrasound probe. We identify several limitations in this study . Firstly , we have limitations inherent to the retrospective de - sign of the study and the use of medical records, thus our results rely on the quality and quantity of infor - mation recorded. Additionally , the time interval bet - ween performing the different tests was not considered. This could result in the disease progressing while pa - tients were undergoing various imaging exams or awai - ting surgery , leading to an underestimation of the re - sults. Despite the fact that visual inspection of lesions during surgery is not fully specific, we also did not con - sider histopathology as an additional reference stan - dard, as our reports did not have sufficient quality , even though the specimens were adequately marked during surgery . The use of the same operators for both TVUS and SVG reduces inter-operator variability but may in - troduce incorporation bias when both exams are per - formed by the same clinician. By contrast, the limited number of scans performed by other clinicians could also contribute to discrepancies in findings. Besides, even though the majority of the TVUS and SVG were performed by the same 2 ultrasonographers, not all MRIs were performed by the same radiologists; howe - ver, from 2020, even the exams performed outside of the hospital could be reviewed by a multidisciplinary team, minimizing this limitation. Furthermore, clini - cians were not blinded to the patient’s clinical history , which can introduce heterogeneity in our data although it better reflects day-to-day practice. Lastly , this study may be subject to patient selection bias as it only in - cludes women who underwent surgery for endome - triosis diagnosis and treatment. Finally , some key strengths of our study are its large sample of patients from 14 years of clinical care in a center with an experienced team in dealing with diag - nosing and treating this disease. The imaging exams performed in this hospital are standardly described to ensure a complete and comprehensive evaluation and surgeries were performed mainly by the same team. In conclusion, our results suggest that TVUS is highly accurate for the diagnosis of ovarian endome - triosis and remains the cornerstone imaging modality for this condition. Both SVG and MRI showed similar performance in diagnosing deep endometriosis and stand as reliable diagnostic tools. Lesions in the utero - sacral ligaments were best identified by MRI, making it an important auxiliary exam. Rather than being equi - valent modalities, SVG and MRI should be viewed as complementary techniques that enhance our unders - tanding of deep lesions and help us optimize surgical planning.

1. Tomassetti C, Johnson NP , Petrozza J, Abrao MS, Einarsson JI, Horne AW , et al. An international terminology for endometriosis, 2021(). Hum Reprod Open. 2021;2021(4):hoab029. Eva Maria de Jesus Paiva et al. Acta Obstet Ginecol Port 2025;19(4):268-279 279 2. Shafrir AL, Farland LV , Shah DK, Harris HR, Kvaskoff M, Zon - dervan K, et al. Risk for and consequences of endometriosis: A cri - tical epidemiologic review. Best Pract Res Clin Obstet Gynaecol. 2018;51:1-15. 3. Allaire C, Bedaiwy MA, Yong PJ. Diagnosis and management of endometriosis. Cmaj. 2023;195(10):E363-e71. 4. Becker CM, Bokor A, Heikinheimo O, Horne A, Jansen F , Kie - sel L, et al. ESHRE guideline: endometriosis. Hum Reprod Open. 2022;2022(2):hoac009. 5. Kiesel L, Sourouni M. Diagnosis of endometriosis in the 21st century . Climacteric. 2019;22(3):296-302. 6. Zondervan KT, Becker CM, Missmer SA. Endometriosis. N Engl J Med. 2020;382(13):1244-56. 7. Pascoal E, Wessels JM, Aas-Eng MK, Abrao MS, Condous G, Jurkovic D, et al. Strengths and limitations of diagnostic tools for en - dometriosis and relevance in diagnostic test accuracy research. Ul - trasound Obstet Gynecol. 2022;60(3):309-27. 8. Guerriero S, Saba L, Pascual MA, Ajossa S, Rodriguez I, Mais V , et al. Transvaginal ultrasound vs magnetic resonance imaging for diagnosing deep infiltrating endometriosis: systematic review and meta-analysis. Ultrasound Obstet Gynecol. 2018;51(5):586-95. 9. Quesada J, Härmä K, Reid S, Rao T, Lo G, Yang N, et al. En - dometriosis: A multimodal imaging review. Eur J Radiol. 2023;158: 110610. 10. Piketty M, Chopin N, Dousset B, Millischer-Bellaische AE, Roseau G, Leconte M, et al. Preoperative work-up for patients with deeply infiltrating endometriosis: transvaginal ultrasonography must definitely be the first-line imaging examination. Hum Reprod. 2009;24(3):602-7. 11. Guerriero S, Condous G, van den Bosch T, Valentin L, Leo - ne FP , Van Schoubroeck D, et al. Systematic approach to sonogra- phic evaluation of the pelvis in women with suspected endome - triosis, including terms, definitions and measurements: a consensus opinion from the International Deep Endometriosis Analysis (IDEA) group. Ultrasound Obstet Gynecol. 2016;48(3):318-32. 12. Daniilidis A, Grigoriadis G, Dalakoura D, D’Alterio MN, An - gioni S, Roman H. Transvaginal Ultrasound in the Diagnosis and Assessment of Endometriosis-An Overview: How, Why , and When. Diagnostics (Basel). 2022;12(12). 13. Dessole S, Farina M, Rubattu G, Cosmi E, Ambrosini G, Nar - delli GB. Sonovaginography is a new technique for assessing recto - vaginal endometriosis. Fertil Steril. 2003;79(4):1023-7. 14. Reid S, Lu C, Hardy N, Casikar I, Reid G, Cario G, et al. Office gel sonovaginography for the prediction of posterior deep in - filtrating endometriosis: a multicenter prospective observational stu - dy . Ultrasound Obstet Gynecol. 2014;44(6):710-8. 15. Arezzo F , Cormio G, La Forgia D, Kawosha AA, Mongelli M, Putino C, et al. The Application of Sonovaginography for Imple - menting Ultrasound Assessment of Endometriosis and Other Gy - naecological Diseases. Diagnostics (Basel). 2022;12(4). 16. Bazot M, Bharwani N, Huchon C, Kinkel K, Cunha TM, Guerra A, et al. European society of urogenital radiology (ESUR) guidelines: MR imaging of pelvic endometriosis. Eur Radiol. 2017;27(7):2765-75. 17. Nisenblat V , Bossuyt PM, Farquhar C, Johnson N, Hull ML. Imaging modalities for the non-invasive diagnosis of endometriosis. Cochrane Database Syst Rev. 2016;2(2):Cd009591. 18. Timmerman D, Valentin L, Bourne TH, Collins WP , Verrelst H, Vergote I. Terms, definitions and measurements to describe the sonographic features of adnexal tumors: a consensus opinion from the International Ovarian Tumor Analysis (IOTA) Group. Ultra - sound Obstet Gynecol. 2000;16(5):500-5. 19. Bau□ic A, Coroleuc□ C, Coroleuc□ C, Comanda□u D, Matasa - riu R, Manu A, et al. Transvaginal Ultrasound vs. Magnetic Reso - nance Imaging (MRI) Value in Endometriosis Diagnosis. Diagnos - tics (Basel). 2022;12(7). 20. Leonardi M, Uzuner C, Mestdagh W , Lu C, Guerriero S, Za - jicek M, et al. Diagnostic accuracy of transvaginal ultrasound for de - tection of endometriosis using International Deep Endometriosis Analysis (IDEA) approach: prospective international pilot study . Ul - trasound Obstet Gynecol. 2022;60(3):404-13. 21. Barra F , Leone Roberti Maggiore U, Evangelisti G, Scala C, Alessandri F , Vellone VG, et al. A prospective study comparing rec - tal water contrast-transvaginal ultrasonography with sonovagino - graphy for the diagnosis of deep posterior endometriosis. Acta Obs - tet Gynecol Scand. 2021;100(9):1700-11. 22. Indrielle-Kelly T, Frühauf F , Fanta M, Burgetova A, Lavu D, Dundr P , et al. Diagnostic Accuracy of Ultrasound and MRI in the Mapping of Deep Pelvic Endometriosis Using the International Deep Endometriosis Analysis (IDEA) Consensus. Biomed Res Int. 2020;2020:3583989. AUTHOR CONTRIBUTIONS Eva Paiva: Conceptualization, Investigation, Methodology , Project ad - ministration, Visualisation, Writing – original draft. Ana Mafalda da Costa Castro Neves: Resources, Validation, Writing – review & edi - ting. Pedro Viana Pinto: Conceptualization, Investigation, Methodo - logy , Project administration, Visualization, Writing – original draft. Ana Margarida Póvoa: Resources, Validation, Writing – review & editing. Andreia Teixeira: Resources, Validation, Writing – review & editing. CONFLICT OF INTEREST The authors declare that they have no conflict of interest. CORRESPONDENCE TO: Eva Maria de Jesus Paiva E-mail: [email protected] https://orcid.org/0009-0000-0960-6218 RECEIVED: 22/01/2025 ACCEPTED: 02/12/2025