Shirley McCarthy

Diagnosis of uterine anomalies: relative accuracy of MR imaging, endovaginal sonography, and hysterosalpingography.

The purpose of this study was to compare the relative accuracy of magnetic resonance (MR) imaging (n = 26), endovaginal sonography (EVS) (n = 14), and hysterosalpingography (HSG) (n = 20) in the classification of müllerian duct anomalies in 26 patients. There were 24 cases of surgically proved anomaly, and two patients had normal uteri (one with a vaginal septum). MR imaging allowed diagnosis of 24 of 24 cases (accuracy, 100%), and EVS was correct in 11 of 12 cases (accuracy, 92%). HSG was correct in only four cases. In the diagnosis of septate uterus, MR imaging demonstrated a sensitivity and specificity of 100% and EVS demonstrated a sensitivity of 100% and a specificity of 80%. Both MR imaging and EVS demonstrated a sensitivity and specificity of 100% in distinguishing those anomalies that did not require surgery. The high accuracy of MR imaging and EVS permit noninvasive differentiation of uterine anomalies and can spare women diagnostic laparoscopy, promoting cost-effective diagnosis.

Peliosis Hepatis: Spectrum of Imaging Findings

OBJECTIVE It is important to recognize the imaging characteristics of peliosis hepatis because peliotic lesions may mimic several different types of focal hepatic lesions CONCLUSION We illustrate the spectrum of imaging findings of peliosis hepatis, including sonography, CT, MR, and angiography.

MRI vs. ultrasound for suspected appendicitis during pregnancy.

To compare the sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) of ultrasound (US) and MRI in evaluation of pregnant patients with a clinical suspicion of appendicitis.

Does pelvic magnetic resonance imaging differentiate among the histologic subtypes of uterine leiomyomata

OBJECTIVE To evaluate if pelvic magnetic resonance imaging (MRI) is reliable for differentiating leiomyoma subtypes. DESIGN Prospective study. SETTING Academic center. PATIENT(S) Forty-five patients underwent MRI before surgery for leiomyomata. INTERVENTION(S) One radiologist blinded to patient history and histologic diagnosis recorded the MRI characteristics and classification of the largest leiomyoma. MAIN OUTCOME MEASURE(S) Comparison of MRI and histologic diagnoses. RESULT(S) Leiomyoma subtypes were diagnosed accurately by MRI in 69% of cases. Magnetic resonance imaging had a 95% sensitivity and 72% specificity for diagnosing an uncomplicated leiomyoma and a 10% sensitivity and 100% specificity for a cellular leiomyoma. For cystic leiomyomata, the sensitivity was 80% and specificity was 98%, and for hemorrhagic leiomyomata, 100% and 86%, respectively. Magnetic resonance imaging correctly diagnosed all malignant tumors and did not incorrectly diagnose a leiomyoma as a leiomyosarcoma in any case. Ill-defined MRI margins were significantly more likely to be leiomyosarcoma, whereas well-defined margins were characteristic of benign lesions. Hemorrhagic leiomyomata were significantly more likely to be hyperintense on T1-weighted images than other subtypes. CONCLUSION(S) Although MRI is only fairly accurate in differentiating the subtypes of benign uterine smooth muscle tumors, signal intensities and margin characteristics are useful to distinguish accurately benign from malignant tumors.

MR evaluation of clinically suspected adnexal masses

Objective Our goal was to determine the sensitivity, specificity, predictive value, and accuracy of pelvic MRI in the prospective evaluation of women with a clinically suspected pelvic mass. Materials and Methods One-hundred three patients with clinically suspected pelvic masses were prospectively evaluated by pelvic MRI. Masses were analyzed for size, location, morphological characteristics, and signal behavior. Masses were classified as benign, malignant, or indeterminate and in every case an attempt was made to generate a specific diagnosis according to previously reported characteristic MR appearances. Surgical follow-up was obtained within 3 months of MRI examination for all patients, excepting 11 patients with typically appearing leiomyomas. Results One hundred twenty pelvic masses were confirmed. Magnetic resonance was 100% sensitive and 99% specific in prospectively diagnosing dermoids, 96% sensitive and 100% specific in diagnosing subserosal leiomyomas, and 92% sensitive and 91% specific in diagnosing endometriomas. Conclusion When physical examination or ultrasound examination is inconclusive, pelvic MRI can aid in the evaluation of women with a suspected pelvic mass. By identifying these common benign gynecologic conditions, unnecessary surgery can be avoided.

High-field MRI and US evaluation of the pelvis in women with leiomyomas

Magnetic resonance imaging (MRI) and real-time transabdominal ultrasonography (US) were performed on 23 women with uterine leiomyomas. The uterus, ovaries, and cul de sac were evaluated. Accurate determination of uterine volume was possible in all cases by MRI, but was limited on US in uteri larger than 140 cc. Marked enlargement also prevented visualization of contour abnormalities in eight patients on US, but none on MRI. The endometrial stripe and junctional zone could not be adequately visualized in 21/23 US examinations, whereas they were identified in all 23 MRI (8 normal and 15 distorted). Individual leiomyomas were clearly depicted on 4 US and 19 MR scans, the smallest being 1.1 cm and 0.8 cm, respectively. Of the 31 fibroids present on MRI: 13 were intramural, 4 subserosal, and 14 submucosal. MRI successfully identified 44/46 ovaries as compared to 21/46 on US. Cul de sac fluid was noted in seven women by MRI alone. This data suggests that MRI is superior to US in examination of the entire pelvis in women with leiomyomas.

Scaphoid fractures and Kienbock's disease of the lunate: MR imaging with histopathologic correlation

Thirteen patients with scaphoid fractures and four patients with Kienbocks disease of the lunate underwent magnetic resonance imaging (MRI) prior to surgery. A total of 28 specimens had MR-histologic correlation. Biopsy specimens obtained by curettage provided pathologic correlation. MRI proved accurate in prospective evaluation of bone viability for both scaphoid fractures and Kienbocks disease. Normal marrow signal was shown to correlate with the presence of osteoid and osteocytes on light microscopy and a surface layer of fluorescence reflecting tetracycline uptake in viable bone. Decreased marrow signal corresponded to non-viable trabeculae with scant osteoid, without osteocytes and no tetracycline labeling. By virtue of its accurate identification of avascular necrosis, MRI may prove valuable in predicting prognosis for patients with scaphoid fractures and Kienbocks disease of the lunate.

MRI appearance of mesenchymal tumors of the uterus

PURPOSE Uterine leiomyomas are the most common uterine neoplasms. Statistically, a uterine mass with unusual imaging features is more likely to represent a leiomyoma than other uncommon uterine mesenchymal neoplasms such as leiomyosarcoma or endometrial stromal tumors. Several prior studies have attempted to identify objective imaging characteristics that differentiate these entities. The purpose of this study was to test these criteria on our patient population. METHODS AND MATERIALS This retrospective study was approved by the institutional Human Investigations Committee and was performed in compliance with HIPAA regulations. Four patients with uterine leiomyosarcoma, two with stromal tumors of uncertain malignant potential (STUMP), one with endometrial stromal sarcoma, and two with mixed endometrial stromal and smooth muscle tumors were included in the study. Seventeen additional control cases of leiomyomas were selected as controls. Cases were blindly evaluated by two experienced readers. Objective criteria included T1 and T2 signal characteristics, enhancement pattern, the presence of cystic changes, and ill defined margins. Subjective criteria included individual reader gestalt. All cases had pathologic correlation. RESULTS None of the objective criteria were associated with the presence or absence of uterine mesenchymal neoplasm. Ill defined margins came closest to having statistical significance (p=0.06). Reader gestalt was statistically associated with the presence of mesenchymal neoplasm for one of our readers (p=0.02) but not for the other (p=0.07). CONCLUSION We found poor accuracy for objective imaging criteria in distinguishing leiomyomas with atypical imaging features from more clinically significant uterine mesenchymal neoplasms.

Pelvic imaging using a T1W fat-suppressed three-dimensional dual echo Dixon technique at 3T.

To compare two T1‐weighted (T1W) fat‐suppressed sequences for 3D breath‐hold pre‐ and postcontrast fat‐suppressed T1W imaging of the female pelvis at 3T.

Fast spin echo STIR imaging.

Objective Our goal was to evaluate the image quality, contrast characteristics, and possible clinical utility of STIR images obtained using a fast SE (FSE) technique. Materials and Methods The signal and contrast characteristics of FSE STIR images were evaluated using a lipid/water phantom and normal volunteers. Based upon these results, optimal FSE STIR imaging parameters were chosen. Conventional STIR and FSE STIR images were then obtained (while maintaining an equal number of section locations between the two sequences) in a series of 14 patients with known musculoskeletal abnormalities. These images were compared side by side by two experienced MR radiologists for image quality and lesion detection. Results There were no statistically significant differences between the FSE STIR images and conventional STIR images in lesion detection, image quality, motion artifact, or final diagnosis. Conclusion STIR imaging provides optimal contrast for detection of many pathologic abnormalities. This is especially true for musculoskeletal tumors and infection. The long imaging time and reduced number of sections obtainable with conventional SE (CSE) STIR sequences limit their routine use. Our results show that FSE STIR images of the musculoskeletal system can be obtained up to seven times more rapidly than CSE STIR images without compromising lesion detection or image quality.