Endometriosis MR mimickers: T2-hypointense lesions.

Endometriosis is a common crippling disease in women of reproductive age. Magnetic resonance imaging (MRI) is considered the cornerstone radiological technique for both the diagnosis and management of endometriosis. While its sensitivity, especially in deep infiltrating endometriosis, is superior to that of ultrasonography, many sources of false-positive results exist, leading to a lack of specificity. Hypointense lesions or pseudo-lesions on T2-weighted images include anatomical variants, fibrous connective tissues, benign and malignant tumors, feces, surgical materials, and post treatment scars which may mimic deep pelvic infiltrating endometriosis. False positives can have a major impact on patient management, from diagnosis to medical or surgical treatment. This educational review aims to help the radiologist acknowledge MRI criteria, pitfalls, and the differential diagnosis of deep pelvic infiltrating endometriosis to reduce false-positive results. Critical relevance statement MRI in deep infiltrating endometriosis has a 23% false-positive rate, leading to misdiagnosis. T2-hypointense lesions primarily result from anatomical variations, fibrous connective tissue, benign and malignant tumors, feces, surgical material, and post-treatment scars. Key points • MRI in DIE has a 23% false-positive rate, leading to potential misdiagnosis.• Anatomical variations, fibrous connective tissues, neoplasms, and surgical alterations are the main sources of T2-hypointense mimickers.• Multisequence interpretation, morphologic assessment, and precise anatomic localization are crucial to prevent overdiagnosis.• Gadolinium injection is beneficial for assessing endometriosis differential diagnosis only in specific conditions.

Endometriosis MR mimickers: T1-hyperintense lesions.

Endometriosis is a chronic and disabling gynecological disease that affects women of reproductive age. Magnetic resonance imaging (MRI) is considered the cornerstone radiological technique for both the diagnosis and management of endometriosis. While MRI offers higher sensitivity compared to ultrasonography, it is prone to false-positive results, leading to decreased specificity. False-positive findings can arise from various T1-hyperintense conditions on fat-suppressed T1-weighted images, resembling endometriotic cystic lesions in different anatomical compartments. These conditions include hemorrhage, hyperproteic content, MRI artifacts, feces, or melanin. Such false positives can have significant implications for patient care, ranging from incorrect diagnoses to unnecessary medical or surgical interventions and subsequent follow-up. To address these challenges, this educational review aims to provide radiologists with comprehensive knowledge about MRI criteria, potential pitfalls, and differential diagnoses, ultimately reducing false-positive results related to T1-hyperintense abnormalities.Critical relevance statementMRI has a 10% false-positive rate, leading to misdiagnosis. T1-hyperintense lesions, observed in the three phenotypes of pelvic endometriosis, can also be seen in various other causes, mainly caused by hemorrhages, high protein concentrations, and artifacts.Key points• MRI in endometriosis has a 10% false-positive rate, leading to potential misdiagnosis.• Pelvic endometriosis lesions can exhibit T1-hyperintensity across their three phenotypes.• A definitive diagnosis of a T1-hyperintense endometriotic lesion is crucial for patient management.• Hemorrhages, high protein concentrations, lipids, and artifacts are the main sources of T1-hyperintense mimickers.

Breast nodule classification with two-dimensional ultrasound using Mask-RCNN ensemble aggregation.

PURPOSE: The purpose of this study was to create a deep learning algorithm to infer the benign or malignant nature of breast nodules using two-dimensional B-mode ultrasound data initially marked as BI-RADS 3 and 4. MATERIALS AND METHODS: An ensemble of mask region-based convolutional neural networks (Mask-RCNN) combining nodule segmentation and classification were trained to explicitly localize the nodule and generate a probability of the nodule to be malignant on two-dimensional B-mode ultrasound. These probabilities were aggregated at test time to produce final results. Resulting inferences were assessed using area under the curve (AUC). RESULTS: A total of 460 ultrasound images of breast nodules classified as BI-RADS 3 or 4 were included. There were 295 benign and 165 malignant breast nodules used for training and validation, and another 137 breast nodules images used for testing. As a part of the challenge, the distribution of benign and malignant breast nodules in the test database remained unknown. The obtained AUC was 0.69 (95% CI: 0.57-0.82) on the training set and 0.67 on the test set. CONCLUSION: The proposed deep learning solution helps classify benign and malignant breast nodules based solely on two-dimensional ultrasound images initially marked as BIRADS 3 and 4.

Three artificial intelligence data challenges based on CT and ultrasound.

PURPOSE: The 2020 edition of these Data Challenges was organized by the French Society of Radiology (SFR), from September 28 to September 30, 2020. The goals were to propose innovative artificial intelligence solutions for the current relevant problems in radiology and to build a large database of multimodal medical images of ultrasound and computed tomography (CT) on these subjects from several French radiology centers. MATERIALS AND METHODS: This year the attempt was to create data challenge objectives in line with the clinical routine of radiologists, with less preprocessing of data and annotation, leaving a large part of the preprocessing task to the participating teams. The objectives were proposed by the different organizations depending on their core areas of expertise. A dedicated platform was used to upload the medical image data, to automatically anonymize the uploaded data. RESULTS: Three challenges were proposed including classification of benign or malignant breast nodules on ultrasound examinations, detection and contouring of pathological neck lymph nodes from cervical CT examinations and classification of calcium score on coronary calcifications from thoracic CT examinations. A total of 2076 medical examinations were included in the database for the three challenges, in three months, by 18 different centers, of which 12% were excluded. The 39 participants were divided into six multidisciplinary teams among which the coronary calcification score challenge was solved with a concordance index > 95%, and the other two with scores of 67% (breast nodule classification) and 63% (neck lymph node calcifications).

Rectal endometriosis: predictive MRI signs for segmental bowel resection.

OBJECTIVE: To retrospectively determine the accuracy of MRI rectal and pararectal signs in predicting the necessity for segmental resection in the case of lesions located in the rectum. METHODS: MR images of consecutive patients treated for rectal endometriosis over a 5-year period were reviewed in consensus by two blinded readers. A systematic analysis of 7 rectal (lesion length, transverse axis, thickness and circumference, and presence of a convex base, submucosal oedema and hyperintense cystic areas) and 4 pararectal (posterior vaginal fornix, parametrial, ureteral and sacro-recto-genital septum involvements) signs was performed for each lesion. MRI results were compared to the surgical procedure performed (shaving versus segmental resection). RESULTS: Among 61 patients studied, 32 received a segmental resection and 29, a shaving. Receiver operating characteristic curve analysis allowed determining cut-off values for length (≥ 32 mm), transverse axis (≥ 22 mm), thickness (≥ 14 mm) and circumference (≥ 3/8 radii). The 7 rectal signs, and only the sacro-recto-genital septum pararectal sign, were significantly associated with segmental resection in univariate analysis, nodular thickness ≥ 14 mm and circumference ≥ 3/8 radii being the most predictive signs (odds ratio 94.5 and 60.4, respectively). These 2 signs remained positively associated with segmental resection in multivariate analysis and, when combined, were predictive of segmental resection with an accuracy of 90.2%. CONCLUSION: Assessing MRI rectal and pararectal signs may accurately predict the need for segmental resection versus a more conservative approach such as shaving for rectal lesion management. KEY POINTS: • MRI analysis of rectal endometriosis, taking into account rectal and pararectal signs, may assist surgeons in the decision-making process, in counselling patients regarding the surgical procedure and in adequately allocating resources. • Among rectal signs, nodular thickness ≥ 14 mm and a circumference ≥ 38% were the most predictive signs of segmental resection. • Among pararectal signs, only the sacro-recto-genital septum involvement was significantly associated with segmental resection.

Bladder endometriosis: Preoperative MRI analysis with assessment of extension to ureteral orifices.

PURPOSE: The purpose of this study was to retrospectively evaluate the performance of magnetic resonance imaging (MRI) in locating endometriosis implants within the bladder wall with assessment of ureteral orifice extension using surgical findings as standard of reference. MATERIALS AND METHODS: MRI examinations of 39 consecutive women (mean age: 31.2±5.5 [SD] years; age range: 22-42years) operated in 3 university hospitals for bladder endometriosis over a 6-year period were reviewed by 2 independent readers. Interobserver agreement was assessed using Kappa tests. Results of consensus reading were used to calculate sensitivity, specificity and accuracy of MRI for the diagnosis, location and extent of endometriosis implants using surgical findings as the standard of reference. RESULTS: Mean bladder repletion volume was 134±110 [SD] mL (range: 21-479mL). The mean largest endometriosis implant diameter was 30±7 [SD] mm (range: 19-41mm). On MR images, 34/39 (87%) endometriosis implants were present in the two anterior thirds of the dome (k=0.45), 31/39 (79%) extended or were present in the posterior third pouch (k=0.92) and 25/39 (64%) extended into the bladder base (k=0.84) with sensitivities of 100% (31/31; 95% confidence interval [CI]: 89-100%), 100% (30/30; 95% CI: 88-100%) and 90% (19/21; 95% CI: 69-98%), respectively, specificities of 83% (5/6, 95% CI: 36-100), 88% (7/8, 95% CI: 47-100%), 87% (13/15; 95% CI: 52-96), respectively and accuracies of 97% (36/37, 95% CI: 86-100%), 97% (37/38; 95% CI: 86-100%), and 89% (32/36; 95% CI: 74-97%), respectively. In 9 (9/25; 36%) patients with bladder base involvement, a zero distance was reported between endometriosis implants and ureteral orifices, all but one presenting with low-to-moderate bladder volumes. In the two patients who needed ureteral resection-reimplantation, ureteral dilation was associated with a zero distance. External adenomyosis was reported in 26/39 (66%) patients (k=0.94). CONCLUSION: A dedicated preoperative MRI work-up for bladder endometriosis helps accurately depict and locate endometriosis implants. Adequate bladder filling is needed to improve appropriate estimate of the distance between endometriosis implants and ureteral orifices to better predict requirement of ureteral resection-reimplantation.

Breast abscesses in lactating women: evidences for ultrasound-guided percutaneous drainage to avoid surgery.

BACKGROUND: Surgical incision with drainage remains the first-line therapy recommendation for breast abscesses greater than 5 cm. PURPOSE: To determine recovery with ultrasound guided (US-guided) procedures for treatment of lactational breast abscesses without surgical incision for drainage. MATERIAL AND METHODS: Institutional review board approval and written informed patient consent were obtained for this retrospective study. From May 1, 2009, to June 1, 2018, 92 consecutive women (mean age 30 years, range 18-45) with 105 abscesses were treated with oral antibiotics and US-guided percutaneous drainage under local anesthesia. A total number of 202 US-guided procedures were performed. Three techniques were used: needle aspiration (diameter 18 to 14G), pigtail catheter insertion (diameter 5 to 7F), and/or vacuum-assisted biopsy/aspiration (diameter 10G). When using needle aspiration or pigtail catheter, a saline irrigation of the cavity was performed according to pus viscosity. RESULTS: The median diameter of abscesses was 4.5 cm (range 1-15), 82/105 (78%) were larger than 3 cm and 40/105 (38%) larger than 5 cm. US-guided management was successful for 101/105 (96%; 95% CI, (91-99%)) abscesses regardless the size. After the first round of procedures, 49/105 (47%) abscesses were recovered, 56/105 (53%) needed more than one drainage with a median number drainages of 2.6 (2-6). In 4/105 cases (4%), women underwent additional surgery under general anesthesia. By excluding abscesses which occurred in the weaning phase (n = 17), breastfeeding carried on for 68/75 (91%) women. CONCLUSION: Unlike previous studies, US percutaneous guided management of lactational abscesses is effective even for abscesses greater than 5 cm and allows continued breastfeeding.

Bowel endometriosis: preoperative diagnostic accuracy of 3.0-T MR enterography--initial results.

PURPOSE: To prospectively determine the accuracy of 3.0-T magnetic resonance (MR) enterography in the preoperative assessment of deep infiltrating endometriosis (DIE) lesions located in the bowel above the rectosigmoid junction. MATERIALS AND METHODS: Institutional review board approval for this study was obtained, and each patient gave written informed consent. Over 18 months, patients with known pelvic DIE who were scheduled for surgery were recruited. Consecutive patients suspected of having bowel endometriosis above the rectosigmoid junction underwent 3.0-T MR enterography. Two blinded readers independently performed a systematic analysis of nine bowel segments above the rectosigmoid junction. MR images were compared with surgical and pathologic findings. Efficacy parameters were calculated with 95% confidence intervals (CIs). Interobserver agreement was assessed with κ statistics. RESULTS: Among the 43 patients enrolled in this study, 33 underwent surgery and were included in the final analyses. Sixteen (48%) patients had bowel DIE lesions located above the rectosigmoid junction at surgery and histopathologic examination. Twenty-six lesions were analyzed, including four ileal, two ileocecal, three cecal, three appendicular, and 14 sigmoid colon lesions. For the diagnosis of these lesions, MR imaging showed sensitivities of 92% (95% confidence interval [CI]: 81.7, 100) for reader 1 and 96% (95% CI: 87.1, 100) for reader 2 and specificities of 100% (95% CI: 98.8, 100) for both readers. The κ value was 0.97. CONCLUSION: These results show 3.0-T MR enterography is accurate in the preoperative diagnosis and mapping of bowel DIE lesions located above the rectosigmoid junction. Online supplemental material is available for this article.