MR Interpretations Research Articles

Analysis of Discordant Findings between 3T Magnetic Resonance Imaging and Arthroscopic Evaluation of the Knee Meniscus.

Numerous studies have assessed the performance of magnetic resonance imaging (MRI) in detecting tears of the knee menisci using arthroscopy results as the gold standard, but few have concentrated on the nature of discordant findings. The purpose of this study was to analyze the discordances between 3T MRI and arthroscopic evaluation of the knee meniscus. Medical records of 112 patients who underwent 3T MRI and subsequent arthroscopy of the knee were retrospectively analyzed to determine the accuracy of diagnoses of meniscal tear. Compared with arthroscopy, there were 22 false-negative and 14 false-positive MR interpretations of meniscal tear occurring in 32 patients. Images with errors in diagnosis were retrospectively reviewed by two musculoskeletal radiologists in consensus and all errors were categorized as either unavoidable, equivocal or as interpretation error. Of 36 MR diagnostic errors, there were 16 (44%) unavoidable, 5 (14%) interpretation errors and 15 (42%) equivocal for meniscal tear. The largest categories of errors were unavoidable false-positive MRI diagnoses (71%) and equivocal false-negative MRI diagnoses (50%). All meniscal tears missed by MRI were treated with partial meniscectomy (n = 14) or meniscal repair (n = 8). Discordant findings between 3T MRI and arthroscopic evaluation of the knee meniscus remain a concern and primarily occur due to unavoidable and equivocal errors. Clinicians involved in the diagnosis and treatment of patients with meniscal tears should understand why and how the findings seen on knee MRI and arthroscopy may sometimes differ.

Utility of an Automated Radiology-Pathology Feedback Tool

PurposeTo determine the utility of an automated radiology-pathology feedback tool. MethodsWe previously developed a tool that automatically provides radiologists with pathology results related to imaging examinations they interpreted. The tool also allows radiologists to mark the results as concordant or discordant. Five abdominal radiologists prospectively scored their own discordant results related to their previously interpreted abdominal ultrasound, CT, and MR interpretations between August 2017 and June 2018. Radiologists recorded whether they would have followed up on the case if there was no automated alert, reason for the discordance, whether the result required further action, prompted imaging rereview, influenced future interpretations, enhanced teaching files, or inspired a research idea. ResultsThere were 234 total discordances (range 30-66 per radiologist), and 70.5% (165 of 234) of discordances would not have been manually followed up in the absence of the automated tool. Reasons for discordances included missed findings (10.7%; 25 of 234), misinterpreted findings (29.1%; 68 of 234), possible biopsy sampling error (13.3%; 31 of 234), and limitations of imaging techniques (32.1%; 75/234). In addition, 4.7% (11 of 234) required further radiologist action, including report addenda or discussion with referrer or pathologist, and 93.2% (218 of 234) prompted radiologists to rereview the images. Radiologists reported that they learned from 88% (206 of 234) of discordances, 38.6% (90 of 233) of discordances probably or definitely influenced future interpretations, 55.6% (130 of 234) of discordances prompted the radiologist to add the case to his or her teaching files, and 13.7% (32 of 233) inspired a research idea. ConclusionAutomated pathology feedback provides a valuable opportunity for radiologists across experience levels to learn, increase their skill, and improve patient care.

Clinical experience with two-point mDixon turbo spin echo as an alternative to conventional turbo spin echo for magnetic resonance imaging of the pediatric knee.

Two-point modified Dixon (mDixon) turbo spin-echo (TSE) sequence provides an efficient, robust method of fat suppression. In one mDixon acquisition, four image types can be generated: water-only, fat-only, in-phase and opposed-phase images. To determine whether PD mDixon TSE water-only and, by proxy, PD in-phase images generated by one acquisition can replace two conventional PD TSE sequences with and without fat suppression in routine clinical MR examination of the knee. This is a retrospective study of 50 consecutive pediatric knee MR examinations. PD mDixon TSE water-only and PD fat-saturated TSE sequences (acquired in the sagittal plane with identical spatial resolution) were reviewed independently by two pediatric radiologists for homogeneity of fat suppression and detection of intra-articular pathology. Thirteen of the 50 patients underwent arthroscopy, and we used the arthroscopic results as a reference standard for the proton-density fat-saturated and proton-density mDixon results. We used the Kruskal-Wallis rank test to assess difference in fat suppression between the proton-density mDixon and proton-density fat-saturated techniques. We used kappa statistics to compare the agreement of detection of intra-articular pathology between readers and techniques. We also calculated sensitivity, specificity and accuracy between arthroscopy and MR interpretations. Proton-density mDixon water-only imaging showed significant improvement with the fat suppression compared with proton-density fat-saturated sequence (P=0.02). Each observer demonstrated near-perfect agreement between both techniques for detecting meniscal and ligamentous pathology and fair to substantial agreement for bone contusions, and chondral and osteochondral lesions. Two-point mDixon water-only imaging can replace conventional proton-density fat-saturated sequence. When same-plane proton-density fat-saturated and non-fat-saturated sequences are required, proton-density water-only and proton-density in-phase image types acquired in the same acquisition shorten the overall examination time while maintaining excellent intra-articular lesion conspicuity.

Cross-Sectional Examination Interpretation Discrepancies Between On-Call Diagnostic Radiology Residents and Subspecialty Faculty Radiologists: Analysis by Imaging Modality and Subspecialty

The aim of this study was to identify the rate of discrepancies between radiology residents and faculty radiologists at an academic hospital and to determine the distribution across subspecialties and modalities, specifically CT, MR, and ultrasound. Consecutive CT, MR, and ultrasound preliminary interpretations rendered by on-call second-year through fourth-year radiology residents for 9 months on emergency department patients, inpatients, and urgent outpatients formed the study population. All preliminary interpretations were graded using a modified RADPEER(®) system (scores 0 and 1 = concordance; score 2 = minor, clinically insignificant discordance; scores 3 and 4 = clinically significant discordance) by the subspecialty faculty members who rendered the final interpretation. There were 158 clinically significant discrepancies out of 21,482 preliminary interpretations, for a discrepancy rate of 0.7%. There was no statistically significant difference in rates across subspecialties or between adult and pediatric examinations (cardiothoracic, 1%; abdominal, 0.7%; neuroradiology, 0.6%; musculoskeletal, 0.7%; pediatrics, 0.8%). MR and CT interpretations had significantly higher rates than ultrasound (MR, 1.4%; CT, 0.9%; ultrasound, 0.2%; P < .001). Within neuroradiology, there was a significantly higher rate for MR than CT (1.5% vs 0.6%, P < .01), and within abdominal radiology, there was a significantly higher rate for CT than ultrasound (1.1% vs 0.2%, P < .01). Discrepancy rates in this study were less than or comparable with those reported previously and within or lower than rates for practicing radiologists. Discrepancy rates varied among subspecialties and modalities, suggesting the use of a RADPEER system with attention to modality and subspecialty as a methodology for identifying areas for targeted resident education.

Acute abdominal and pelvic pain in pregnancy: MR imaging as a valuable adjunct to ultrasound?

To investigate the usefulness of magnetic resonance imaging (MRI) in the evaluation of pregnant women with acute abdominal and pelvic pain after sonographically (US) indeterminate findings. Forty pregnant patients with acute abdominal and/or pelvic pain, in whom ultrasound was indeterminate, were included in this study. Multiplanar MR images of the abdomen and pelvis were obtained and independently reviewed by two radiologists with discrepancies were resolved by consensus. Correlation of prospective clinical MR interpretations with surgical and obstetric records and clinical follow-up showed correct identification of disease entities in all 40/40 patients. The following disease processes were correctly identified using MRI in 21/40 patients: appendicitis (n = 5), ileal volvulus (n = 1), common bile duct lithiasis (n = 2), pubic chondrosarcoma (n = 1), uterine dehiscence (n = 1), placental hematoma (n = 3), iliac venous thrombosis (n = 2), ulcerative colitis (n = 1), acute pancreatitis (n = 1), hydronephrosis (n = 1), ovarian dermoid (n = 1), and ovarian torsion (n = 2). 19 of the 40 patients had normal findings on MR examinations and unremarkable follow-up. Interobserver agreement for lesion detection was excellent (0.95 k). Magnetic resonance imaging (MRI) is an accurate investigation in detecting the cause of acute abdominal and pelvic pain during pregnancy and should be considered after US indeterminate findings.

3.0-T MRI of Meniscal Tears

The sensitivity and specificity of 3.0-T MRI of the knee for meniscal tears have not been specifically assessed. We undertook a retrospective review of 100 consecutive MR examinations of the knee followed by arthroscopy to evaluate the sensitivity and specificity of 3.0-T MRI compared with arthroscopy in the detection of meniscal tears. Two experienced musculoskeletal radiologists retrospectively reviewed MR scans of the knees of 100 patients who underwent consecutive knee arthroscopic procedures after MRI examinations. Interpretations were performed by consensus review with the reviewers blinded to arthroscopic results at the time of consensus review. All patients underwent complete MRI of the knee on a 3.0-T unit. All patients underwent fast spin-echo proton density sagittal imaging at 2-mm slice thickness. Fat-saturated fast spin-echo T2 axial, coronal, and sagittal imaging and T1 coronal imaging also were performed. All 100 patients underwent arthroscopy, and the results were compared with MR interpretations. Consensus retrospective review of the 100 knee MR examinations revealed 111 meniscal tears. Four meniscal tears seen on arthroscopy were not seen on MR examination. There were three false-positive MR interpretations of meniscal tear compared with arthroscopy. In this study, the sensitivity of MRI in the detection of meniscal tears was 96%, and the specificity was 97%. MRI of the knee at 3.0 T is sensitive and specific compared with arthroscopy in the detection of meniscal tears. Findings at 3.0 T compare favorably with results at 1.5-T or lower field strength.

MRI of Acute Abdominal and Pelvic Pain in Pregnant Patients

The purpose of this study was to show the usefulness of MRI in the evaluation of pregnant women with acute abdominal or pelvic pain. All MRI studies of pregnant patients who were referred for examination because of acute abdominal or pelvic pain between June 2002 and May 2004 were included in this study (n=29). The use of MRI was at the discretion of the clinician. Fetal sonography was performed in all patients before any other imaging. A complete abdominal sonographic examination was performed in six patients before MRI. In the remaining 23 patients, MRI was the choice for primary imaging. Multiplanar multisequence MR images of the abdomen and pelvis were obtained in each patient. Unenhanced images were reviewed by an experienced radiologist to determine whether a diagnosis could be made without the administration of gadolinium. In 22 of 29 studies, gadolinium was not administered. The prospective clinical MR interpretations were compared with follow-up medical, surgical, and obstetric records to determine the correctness of the interpretation. All patients were followed up until the date of article submission or until the date of final chart entry. Correlation of prospective clinical MR interpretations with follow-up medical records showed correct identification of disease entities in all but one patient. In one patient, torsion of the ovary was neither described prospectively nor seen in retrospect. The following disease processes were correctly identified using MRI: appendiceal abscess (n=1), appendicitis (n=2), intraabdominal and rectus muscle abscess (n=1), intussusception (n=1), pancreatitis (n=1), ulcerative colitis (n=1), Crohn's disease with diffuse peritoneal inflammation (n=1), bilateral adrenal hemorrhage (n=1), pyelonephritis (n=2), hydronephrosis (n=1), uterine fibroid degeneration (n=2), degeneration and torsion of a submucosal uterine fibroid (n=1), simple ovarian cysts (n=1), and ovarian torsion (n=1). Twelve of the 29 patients had normal findings on MR examinations and unremarkable follow-up. The intrinsic safety of MRI and its ability to accurately show abdominal and pelvic disease in pregnant patients make it highly useful in the evaluation of these patients.

Reassessment of the MR criteria for stability of osteochondritis dissecans in the knee and ankle

T2-weighted MR images has been reported to be an accurate method for assessing osteochondritis dissecans. We reviewed our MR experience to confirm the accuracy of the published criteria of instability. We also assessed the value of each of four MR signs of instability. We reviewed the original MR interpretations, arthroscopic reports, and MR examinations of 40 patients with osteochondritis dissecans of the talar dome or femoral condyles. Arthroscopy was used as the gold standard for stability. The MR examinations were reviewed retrospectively for a high-signal-intensity line or cystic area beneath the lesion, a high-signal-intensity line through the articular cartilage, or a focal articular defect. All patients who had undergone MR imaging for osteochondritis dissecans from 1990 to 1993 were reviewed. Forty patients were identified who had arthroscopy after the MR examination. There were 30 male and 10 female patients with an average age of 25.7 years. Thirty-one lesions were in femoral condyle and nine were in the talar dome. The original MR interpretations correctly identified 35 of the 36 unstable lesions and all 4 stable lesions, giving a sensitivity of 0.97 and specificity of 1.0. There was a 98% agreement between the original and retrospective diagnoses. A high-signal-intensity line was seen beneath 72% of the 36 unstable lesions. The other three signs were noted in 22-31% of the unstable lesions. Fifty-six percent of the unstable lesions showed only one sign of instability. MR imaging is a highly sensitive method for detection of unstable osteochondritis dissecans. The presence of any one sign indicates instability, the most frequent sign being an underlying high-signal-intensity line. Because we examined only four stable lesions, our 95% confidence interval of 0.40-1.0 for a specificity of 1.0 gives only a limited estimate of the specificity of MR.

Epidural fibrosis and recurrent disk herniation in the lumbar spine: MR imaging assessment.

Twenty patients were enrolled in a prospective study to evaluate MR imaging in the differentiation of epidural scar and herniated disk material. Fourteen patients had surgical verification of imaging findings. In 12 (86%) of these patients, the MR interpretations fully agreed with the observations at surgery. Careful integration of the findings on sagittal and axial T1-weighted images with more T2-weighted axial images was important for analysis. Anterior and lateral recess scars were hypo- or isointense on T1-weighted sequences and hyperintense on more T2-weighted sequences relative to the "parent" anulus intensity. Free fragments demonstrated a slightly hyperintense signal intensity on T1-weighted images relative to epidural fibrosis but had a similar hyperintense signal intensity on T2-weighted sequences. Prolapsed or extruded disk fragments were hypo- or isointense relative to the parent anulus on all sequences. Morphology, epidural location, mass effect, and often signal intensity were the important parameters by which scar and herniated disk could be differentiated with MR.