RPS 1407b - Multimodality approach in imaging of the uterus and endometriosis

Purpose:

To assess the non-inferiority of a DIXON sequence (DiS) in relation to VT1 in endometrioma evaluation, between readers, using subjective and quantitative analysis.

Methods and materials:

100 consecutive MRIs performed between April 2016-April 2018 were evaluated, half with the word endometriomas in the report and half with normal exams.

The exams were randomly distributed to two readers, a fellow with one year experience and a junior radiologist with 3 years of experience in pelvic imaging. The readers did two sessions: first, they were instructed to look only at DiS, and second, after a month interval to reduce memory bias, only at the VT1 sequence. The following criteria were evaluated: the presence of an endometrioma, size, signal ratio (SR), and laterality.

The SR was assessed with a region of interest (ROI) placed in the lesion and in the ipsilateral vertebral musculature.

Results:

Of the 50 endometriomas cases, 3 were excluded due to incomplete data.

The agreement was perfect between readers in the 50 cases without endometriomas.

The intraobserver agreement was 96.9% for the first reader and 92.8% for the second, with kappa (k) values of 0.94 and 0.85, respectively. Interobserver agreement was 94.9% using VT1 and 92.8% for the DiS, with k values of 0.89 and 0.85, respectively.

The signal ratio was 1.78 using DiS and 2.05 for VT1, with p value=0.005.

Conclusion:

The DIXON sequence is effective as the conventional T1 sequence for the evaluation of endometriomas, showing good agreement between observers, even without a higher signal ratio, as we postulated previously.

Limitations:

A small sample and single institution.

Ethics committee approval

n/a

Funding:

No funding was received for this work.

Purpose:

To retrospectively test the performance and reproducibility of MR Enzian classification and to develop a new MR classification, including lateral endometriosis, and to evaluate their ability to predict postoperative complications.

Methods and materials:

Between 2017/01/01 and 2018/12/31, 150 women (mean age 34,5 years, 20-52 years) with deep endometriosis (DE) at MR imaging with a subsequent surgery were recruited. Two radiologists (junior and senior) independently reviewed MR images and rated DE according to MR Enzian classification and a new radiological classification named deep pelvic endometriosis index (dPEI), grading the severity of DE in low (≤2 compartments), moderate (3 or 4 compartments), and high extensive (≥5 compartments).

Results:

MR and surgical Enzian classification were concordant for A lesions in 78.7% (118/150), for B lesions in 34.7% (52/150), and for C lesions in 82.7% (124/150). Operative time and hospital stay were longer in A2 compared to A0, B2 to B0, C3 to C2, and C2 to C0 (p<0.001) in high extensive compared to moderate extensive DE, and in moderate to low extensive DE (p<0.01). Patients with a vagina or rectosigmoid involvement had, respectively, 6 and 3 times more frequent complications according to Clavien Dindo (p<0.001). Postoperative dysuria was correlated to the presence of A lesions (OR=6.82, p=0.001), parametrial lesions (OR=6.6, p=0.0002), a moderate or high extensive DE according to dPEI (OR=4.15, p=0.001), unilateral lateral pelvic involvement (OR=3.6, p=0.03), and C lesions (OR= 2.6, p=0.04).

Conclusion:

MR imaging is accurate to predict postoperative complications, helping clinicians to preoperatively inform patients of the specific risk of surgery for DE.

Limitations:

A highly experienced centre. No long term follow up.

Ethics committee approval

Our institutional ethics committees approved the study and granted a waiver of informed consent.

Funding:

No funding was received for this work.

Purpose:

With increasing MRI usage for female pelvic pain evaluation, we are seeing an increase of minimal deep endometriosis diagnosis showing very subtle/subjective findings.

The study's purpose was to evaluate subtle/minimal endometriosis findings on MRI, assessing inter and intraobserver concordance between radiologists with different expertise levels.

Methods and materials:

We conducted a retrospective study selecting 100 female pelvic MRI exams from 2015-2018. We searched for 50 normal MRI reports and 50 other examinations were chosen using keywords describing incipient/subtle endometriosis in the radiology reports.

The MRIs were independently analysed by three radiologists with different experience levels (two assistants with 15 and 3 years of expertise and one fellow with 1 year of expertise). They were blinded to clinical data and original reports. The readers filled a questionnaire determining the presence (or absence) and location of pelvic endometriosis. After a 2 months interval (memory bias), all the readers re-evaluate the exams. The imaging criteria was revised with all the readers with practical cases before the study started.

Results:

The intraobserver analysis was poor or weak in most compartments, with kappa (k) values varying between 0.24-0.48 for torus uterinus (TU-mean: 0.37) and 0.21-0.41 for utero-sacral ligaments (USL-mean: 0.36). The other areas reached lower k values, with the exception of ovaries (mean: 0.62).

The interobserver analysis was also low, with intraclass correlation values of 0.039 for TU and 0.032 for USL. Even if the studies were analysed by the presence of at least one area positive, the ICC was also poor, reaching 0.018.

Conclusion:

Inter and intraobserver agreement were poor for assessing subtle/minimal endometriosis with pelvic MRI. These findings were independent of the reader's experience level.

Limitations:

A single-centre study with a small number of patients and no pathological confirmation.

Ethics committee approval

Ethics approval, informed consent waived.

Funding:

No funding was received for this work.

Purpose:

To assess the accuracy of a new transvaginal ultrasound (TVS) mapping system in defining the size and location of deep infiltrating endometriosis (DIE) with laparoscopic and histologic confirmation.

Methods and materials:

Our study included 51 patients showing features of DIE by TVS with subsequent laparoscopic and histologic evaluation. Data was collected from October 2017 to September 2019. All patients were examined by the same radiologist with GE E8 (GE Healthcare) ultrasound machine using a transvaginal 5-9 MHz transducer. Conventional 2D and 3D volume acquisition using volume contrast imaging (VCI) and tomographic ultrasound imaging (TUI) was used. Accurate mapping of the extent of DIE was recorded using a dedicated mapping sheet during TVS by the radiologist and at laparoscopy by the same surgeon who was blinded to TVS mapping. Results were correlated with histopathology findings.

Results:

Laparoscopy detected DIE in all 51 patients. Depending on the locations of DIE, the accuracy of TVS ranged from 73%-98%. The lowest sensitivity (47%) and accuracy (73%) was seen in a TVS diagnosis of vaginal DIE with a positive likelihood ratio (LR+) of 2.8, while the highest accuracy (98%) was seen in detecting bladder lesions with LR+ of 47. 5 cases showed additional scar endometriosis with 100% accuracy.

Conclusion:

We believe that the new TVS mapping is accurate in detecting the extent of DIE, which may help in preoperative evaluation and intraoperative management to tailor the surgical approach, thus improving patient outcomes.

Limitations:

The high incidence of DIE in the sample due to patient selection and setting of the study as a referral centre for endometriosis.

Ethics committee approval

Informed consent was obtained from all patients.

Funding:

No funding was received for this work.

Purpose:

To compare findings of conventional MRI, DWI, and DCE-MRI in benign and malignant endometrial lesions.

Methods and materials:

44 adult female patients with sonographically suspected endometrial lesions underwent MR imaging of the abdomen and pelvis (T1, T2WI, DWI, and DCE MRI). Conventional MR images were analysed for morphological features. The presence or absence of restricted diffusion was assessed on DWI and ADC values were calculated. Using DCE images, enhancement characteristics were studied and time-signal intensity (TIC) curves were generated. Appropriate tests for statistical analysis were applied. The ROC curve was applied to determine the ADC threshold value to differentiate between the malignant and benign lesions. The imaging diagnosis was finally compared with histopathological examination or follow-up imaging.

Results:

Of 44 patients, 16 had benign lesions and 28 harboured malignant lesions. Conventional MR findings found to have a significantly higher frequency in benign lesions were a brightly hyperintense signal on T2WI, smooth tumour/endo-myometrial interface, and the presence of well-defined cystic areas within the mass (p<0.01). Diffusion restriction was seen only in 2/16 benign cases, however, was demonstrated in all the malignant cases (p<0.01). Using a ROC curve, cut off ADC value of <0.859 x 10-3 mm2/sec was seen to have a sensitivity, specificity, PPV, and NPV of 100%, 89.3%, 100%, and 93.2%, respectively, for predicting malignant endometrial lesions. On dynamic imaging and TIC analysis, the increasing trend of enhancement was found to have a sensitivity of 86.2% and specificity of 81.2% for benign lesions.

Conclusion:

Newer functional MR techniques of DWI and DCE-MR can significantly add to conventional MR imaging in differentiating benign from malignant endometrial lesions.

Limitations:

A small sample size.

Ethics committee approval

Approval from an institutional ethical committee obtained.

Funding:

No funding was received for this work.

Purpose:

To assess the usefulness of diffusion kurtosis imaging (DKI) as a noninvasive method for detecting and evaluating endometrial cancer (EMC).

Methods and materials:

18 consecutive patients with a histological diagnosis of EMC by biopsy and 19 healthy volunteers were enrolled. They underwent an MR exam with diffusion-weighted imaging (DWI), acquired using multiple b values (from b 0 to b 2500 s/mm2) for each subject.

DKI parameters were calculated for each voxel, to obtain K (kurtosis) and D (diffusivity). We also evaluated conventional ADC.

All DWI images were processed using an in-house program developed with MatLab software (R2016a; MathWorks, Natick, MA).

For K and D maps, regions of interest (ROIs) were manually drawn approximately equivalent in size to tumoural tissue and myometrium for EMC and to endometrium and myometrium for healthy patients, using DWI and T2W as references.

Mean ± SD of K, D, and ADC values for uterine wall layers and EMCs were calculated and differences analysed using an Anova test.

Results:

ADC (*10-3 mm2/s): EMC 0.87±0.22; healthy 0.96±0.23; p -

D (*10-3 mm2/s): EMC 1.05±0.30; healthy 1.14±0.25; p -

K (u.a.): EMC 1.12±0.30; healthy 0.63±0.18; p 0.015.

K significantly discriminated between a tumour and healthy tissue (p=0.015), whereas D and ADC didn’t discriminate.

Conclusion:

Kurtosis imaging seems to be more sensitive compared to conventional diffusion imaging in detecting EMC.

These results are promising and the difference between EMC and healthy layers may be used to precisely predict myometrial infiltration during MR examination.

Limitations:

A limited number of patients.

Ethics committee approval

The study was approved by an ethics committee. Every patient gave written informed consent.

Funding:

No funding was received for this work.

"Purpose:

The revised classification of endometriosis by the American Society for Reproductive Medicine was widely used in evaluating pelvic endometriosis. However, it was not well correlated to the severity of symptoms and obstetrical outcomes as it heavily depended on the size of the ovarian endometriotic cyst. Some trials had been made to make classifications reflecting clinical importance. However, there is no widely accepted system to date. The purpose of this study is to establish a new evaluation scale for the severity of pelvic endometriosis, including adhesion and deeply infiltrating endometriosis (DIE), which are more clinically meaningful.

Methods and materials:

60 cases with pelvic endometriosis were retrospectively evaluated using the same scoring system with MR and laparoscopy, independently. Three radiologists evaluated MR images including sagittal and axial T1-, T2-, and fat-saturated T1-weighted images obtained with 1.5T superconducting MR systems. Two gynaecologists also evaluated recorded video during laparoscopic surgery. Both decisions were made by consensus reading. The score included adhesions between uterus and bladder, uterus and ovaries, uterus and rectum, ovaries and rectum, and bilateral ovaries and fibrotic plaque on the posterior uterine aspect, respectively. The Pearson’s coefficient of correlation of the total score between MR and laparoscopy was calculated. We also studied the concordance rate of each score.

Results:

The Spearman’s rank correlation coefficient between the MR and laparoscopy was 0.81. The concordance rate of each score was 0.52-0.97.

Conclusion:

Our new MR scoring system emphasised adhesion and DIE was feasible to evaluate the severity of endometriosis.

Limitations:

A limited number of cases.

Ethics committee approval

Approved by the ethics committees of Kurashiki Medical Center, Medical Topia Soka Hospital and Chugai Pharmaceutical Co., LTD.

Funding:

No funding was received for this work.

"