Timothy J. Amrhein

Myelographic Techniques for the Detection of Spinal CSF Leaks in Spontaneous Intracranial Hypotension

OBJECTIVE Spinal leakage of CSF causes almost all cases of spontaneous intracranial hypotension. Leak detection and localization are important for both diagnosis and treatment. The myelographic appearance of the leaks may vary, however, depending on the cause of the leak, rate of leakage, and imaging modality used. CONCLUSION The purpose of this article is to review the imaging of spinal CSF leaks and to assist in the selection of appropriate imaging modalities in this condition.

Imaging Signs in Spontaneous Intracranial Hypotension: Prevalence and Relationship to CSF Pressure

BACKGROUND AND PURPOSE: Patients with spontaneous intracranial hypotension often exhibit low CSF pressure and changes on brain MR imaging and/or evidence of CSF leak on myelography. We investigated whether individual imaging signs of spontaneous intracranial hypotension correlate with measured CSF pressure and how frequently these 2 markers of spontaneous intracranial hypotension were concordant. MATERIALS AND METHODS: We performed a retrospective, cross-sectional study of 99 subjects with spontaneous intracranial hypotension. Prevalence of brain and myelographic imaging signs of spontaneous intracranial hypotension was recorded. CSF pressure among subjects with or without individual imaging signs was compared by using a 2-tailed t test and ANOVA. Concordance between low CSF pressure (≤6 cm H2O) and imaging was defined as the presence of the sign in a subject with low CSF pressure or absence of the sign when pressure was not low. RESULTS: Dural enhancement, brain sagging, and venous distension sign were present in 83%, 61%, and 75% of subjects, respectively, and myelographic evidence of CSF leak was seen in 55%. Marginal correlations between CSF pressure and brain sagging (P = .046) and the venous distension sign (P = .047) were found. Dural enhancement and myelographic evidence of leak were not significantly correlated with CSF pressure. Rates of concordance between imaging signs and low CSF pressure were generally low, ranging from 39% to 55%. CONCLUSIONS: Brain and myelographic signs of spontaneous intracranial hypotension correlate poorly with CSF pressure. These findings reinforce the need to base the diagnosis of spontaneous intracranial hypotension on multiple diagnostic criteria and suggest the presence of patient-specific variables that influence CSF pressure in these individuals.

How common is normal cerebrospinal fluid pressure in spontaneous intracranial hypotension

Objectives To determine the proportion of patients with spontaneous intracranial hypotension (SIH) who had a cerebrospinal fluid (CSF) pressure >6 cm H2O and to investigate the clinical and imaging variables associated with CSF pressure (PCSF) in this condition. Methods We retrospectively reviewed 106 patients with SIH. PCSF was measured by lumbar puncture prior to treatment. Clinical and imaging variables – including demographic data, brain imaging results, symptom duration, and abdominal circumference – were collected. Univariate and multivariate analyses were performed to determine the correlation of these variables with PCSF. Results Sixty-one percent of patients had a PCSF between 6 and 20 cm H2O; only 34% had a PCSF ≤6 cm H2O. The factors associated with increased PCSF included abdominal circumference (p < 0.001), symptom duration (p = 0.015), and the absence of brain magnetic resonance imaging findings of SIH (p = 0.003). A wide variability in PCSF was observed among all patients, which was not completely accounted for by the variables included in the model. Conclusions Normal CSF pressure is common in patients with SIH; the absence of a low opening pressure should not exclude this condition. Body habitus, symptom duration, and brain imaging are correlated with PCSF measurements, but these factors alone do not entirely explain the wide variability in observed pressures in this condition and this suggests the influence of other factors.

Incidence of Inadvertent Intravascular Injection during CT Fluoroscopy-Guided Epidural Steroid Injections.

Retrospective analysis of 575 CT fluoroscopy-guided epidural injections was done to detect inadvertent arterial or venous injections. Intravascular injections occurred in 26% of cervical transforaminal injections, 9% of cervical interlaminar injections, 8% of lumbar transforaminal injections, and 2% of lumbar interlaminar injections. These findings are similar to previous reports using conventional fluoroscopy and demonstrate that intravascular injections may be detected with CT fluoroscopy. BACKGROUND AND PURPOSE: Inadvertent intravascular injection during epidural steroid injection can result in complications and has been investigated previously with conventional fluoroscopy, but not CT fluoroscopy. The purpose of this study was to determine the incidence of intravascular injections recognized during CT fluoroscopy–guided epidural steroid injection. MATERIALS AND METHODS: We retrospectively reviewed 575 consecutive CT fluoroscopy–guided epidural steroid injections. Procedures were assessed to determine the incidence of intravascular injection. Cases positive for intravascular injection were classified on the basis of anatomic location, distance from the needle tip, washout pattern, and presence of combined epidural and vascular injection. Cases were also graded as either venous or arterial by using a 5-point scale. RESULTS: Intravascular injection was observed in 26% of cervical transforaminal epidural steroid injections (7/27), 9% of cervical interlaminar epidural steroid injections (4/47), 8% of lumbar transforaminal epidural steroid injections (22/275), and 2% of lumbar interlaminar epidural steroid injections (4/222). Vessels were most commonly identified close to the needle, but in 30% of cases, they were visualized in the anterior paraspinal soft tissues remote from the needle. Washout was most commonly delayed (86%), though rapid washout occurred in 14% of cases. Simultaneous epidural and vascular injections occurred in 32% of cases. Most visualized vessels were venous, but 2 cases were classified as probably arterial. CONCLUSIONS: Intravascular injections can be detected with CT fluoroscopy. The incidence in our study was similar to that in previous reports using conventional fluoroscopy. Technical factors such as the “double-tap” on CT fluoroscopy following contrast injection, assessment for discordance between injected and visualized contrast volume, and maintenance of an appropriate FOV facilitate the detection of such events.

Mn Porphyrins as Novel Molecular Magnetic Resonance Imaging Contrast Agents

BACKGROUND AND PURPOSE In this study, we investigated the potential of a new class of therapeutic Mn porphyrins as molecular MRI probes for prostate cancer imaging. Two compounds of different bioavailibility were investigated: Mn(III) meso-tetrakis(N-ethylpyridinium-2-yl)porphyrin (MnTE-2-PyP(5+)) and Mn(III) meso-tetrakis(N-n-hexylpyridinium-2-yl)porphyrin (MnTnHex-2-PyP(5+)). These compounds have previously been shown to have adjunctive antineoplastic activity through their actions as powerful superoxide dismutase mimics, peroxynitrite scavengers, and modulators of cellular redox-based signaling pathways. Strong paramagnetic MRI contrast properties and affinity for cancer cells suggest their potential application as novel diagnostic imaging agents. MATERIALS AND METHODS MRI experiments were performed at 7.0T on a Bruker Biospec horizontal bore scanner. All in-vivo experiments were performed on 12 C57 black mice implanted with RM-9 prostate cancer cells on the hind limb. Two mg/kg of MnTnHex-2-PyP(5+) (n=6) and 8 mg/kg MnTE-2-PyP(5+) (n=6) were administered intraperitoneally 90 minutes before imaging. All the images were collected using a volume coil and processed using Paravision 4.0. RESULTS Phantom studies reveal remarkably high T1 relaxivity changes for both metalloporphyrins, which are twofold to threefold higher than commercially available gadolinium chelates. Observable detection limits using conventional T1-weighted MRI are in the low micromolar range for both compounds. In vivo, MR relaxation changes in prostate tumor xenografts were readily observed after a single injection of either MnTE-2-PyP(5+)or MnTnHex-2-PyP(5+), with tumor contrast to background ratio greatest after MnTE-2-PyP(5+) administration. CONCLUSION After a single dose of MnTE-2-PyP(5+), contrast changes in prostate tumors are up to sixfold greater than in surrounding, noncancerous tissues, suggesting the potential use of this metalloporphyrin as a novel diagnostic probe for detecting prostate malignancy using MRI.

Rebound Intracranial Hypertension: A Complication of Epidural Blood Patching for Intracranial Hypotension

SUMMARY: Rebound intracranial hypertension is a complication of epidural blood patching for treatment of intracranial hypotension characterized by increased intracranial pressure, resulting in potentially severe headache, nausea, and vomiting. Because the symptoms of rebound intracranial hypertension may bear some similarity to those of intracranial hypotension and literature reports of rebound intracranial hypertension are limited, it may be mistaken for refractory intracranial hypotension, leading to inappropriate management. This clinical report of 9 patients with confirmed rebound intracranial hypertension reviews the clinical characteristics of patients with this condition, emphasizing factors that can be helpful in discriminating rebound intracranial hypertension from refractory spontaneous intracranial hypotension, and discusses treatment.

Physician Self-Referral: Frequency of Negative Findings at MR Imaging of the Knee as a Marker of Appropriate Utilization

PURPOSE To determine whether ownership of magnetic resonance (MR) imaging equipment by ordering physicians affects the likelihood of positive findings at MR imaging of the knee and to evaluate rates of knee abnormalities seen at MR imaging as a metric for comparison of utilization. MATERIALS AND METHODS The institutional review board approved this retrospective HIPAA-compliant study and waived the need for informed consent. A retrospective review was performed of consecutive diagnostic MR images of the knee interpreted by one radiology practice between January and April 2009 for patients who had been referred by two separate physician groups serving the same geographic community: one with financial interest (FI) in the MR imaging equipment used and one with no FI (NFI) in the MR imaging equipment used. The percentage of examinations with negative results was tabulated for both groups, and the relative frequency of each abnormality subtype was calculated among the studies with positive findings in each group. To examine frequency differences among groups, χ(2) tests were used, and to examine mean differences among groups, t tests were used. RESULTS Of 700 examinations, 205 had negative results (117 of 350 in the FI group and 88 of 350 in the NFI group, P = .016). Among the examinations with positive results, the mean total number of positive abnormality subtypes per image did not significantly differ between groups: 1.52 for the FI group and 1.53 for the NFI group (P = .96). CONCLUSION MR images of the knee among patients referred by the FI group were significantly more likely to be negative than those among patients referred by the NFI group. Frequency of abnormality subtype and distribution among examinations with positive results suggests a highly similar distribution and severity of abnormalities between the two patient groups.

Deep Learning to Classify Radiology Free-Text Reports

Purpose To evaluate the performance of a deep learning convolutional neural network (CNN) model compared with a traditional natural language processing (NLP) model in extracting pulmonary embolism (PE) findings from thoracic computed tomography (CT) reports from two institutions. Materials and Methods Contrast material-enhanced CT examinations of the chest performed between January 1, 1998, and January 1, 2016, were selected. Annotations by two human radiologists were made for three categories: the presence, chronicity, and location of PE. Classification of performance of a CNN model with an unsupervised learning algorithm for obtaining vector representations of words was compared with the open-source application PeFinder. Sensitivity, specificity, accuracy, and F1 scores for both the CNN model and PeFinder in the internal and external validation sets were determined. Results The CNN model demonstrated an accuracy of 99% and an area under the curve value of 0.97. For internal validation report data, the CNN model had a statistically significant larger F1 score (0.938) than did PeFinder (0.867) when classifying findings as either PE positive or PE negative, but no significant difference in sensitivity, specificity, or accuracy was found. For external validation report data, no statistical difference between the performance of the CNN model and PeFinder was found. Conclusion A deep learning CNN model can classify radiology free-text reports with accuracy equivalent to or beyond that of an existing traditional NLP model.

The “Hyperdense Paraspinal Vein” Sign: A Marker of CSF-Venous Fistula

SUMMARY: CSF-venous fistula is a recently reported cause of spontaneous intracranial hypotension that may occur in the absence of myelographic evidence of CSF leak. Information about this entity is currently very limited, but it is of potential importance given the large percentage of cases of spontaneous intracranial hypotension associated with negative myelography findings. We report 3 additional cases of CSF-venous fistula and describe the “hyperdense paraspinal vein” sign, which may aid in its detection.

JOURNAL CLUB: Shoulder MRI Utilization: Relationship of Physician MRI Equipment Ownership to Negative Study Frequency

OBJECTIVE The purpose of this article is to determine whether ownership of MRI equipment by ordering physicians influences the frequency of negative shoulder MRI scans. MATERIALS AND METHODS A retrospective review was performed of 1140 consecutive shoulder MRI scans ordered by two separate referring physician groups serving the same geographic community. The first group (financially incentivized) owned the scanners used and received technical fees for their use. The second group (non-financially incentivized) did not own the scanners used and had no direct financial interest. All examinations were performed with identical protocols and were interpreted by a single radiologist group without financial interest in the imaging equipment used. The frequency of negative examinations and the number of abnormalities in each positive study was tabulated for each group. RESULTS A total of 1140 shoulder MRI scans met inclusion criteria; 255 were negative (142 for the financially incentivized group and 113 for the non-financially incentivized group). There were 25.6% more negative scans in the financially incentivized group (p=0.047). There was no statistically significant difference in the average number of lesions per positive scan (1.67 for the financially incentivized group and 1.71 for the non-financially incentivized group; p=0.34). No statistically significant difference was found in the frequency of 19 of 20 examined lesions. CONCLUSION Shoulder MRI examinations referred by physicians with a financial interest in the imaging equipment used were significantly more likely to be negative. Positive examinations exhibited no statistically significant difference in the number of lesions per scan or in the frequency of 19 of 20 lesion subtypes. This finding suggests a highly similar distribution and severity of disease among the two patient groups.