Faculty Bibliography

BACKGROUND AND PURPOSE/OBJECTIVE:Conventional imaging frequently shows overlapping features between benign and malignant parotid neoplasms. We investigated dynamic contrast-enhanced MR imaging using golden-angle radial sparse parallel imaging in differentiating parotid neoplasms. MATERIALS AND METHODS/METHODS:= 32) combined semiquantitative time-intensity curve metrics with ADC values. RESULTS:< .001). CONCLUSIONS:Golden-angle radial sparse parallel MR imaging allows high spatial and temporal resolution permeability characterization of parotid neoplasms, with a high negative predictive value for malignancy prediction. Combining time-to-maximum and ADC improves pleomorphic adenoma prediction compared with either metric alone.

Cavernous sinuses are paired interconnected venous plexuses situated in the floor of the middle cranial fossa on either side of the sella turcica and sphenoid sinus. They are lined by dura mater and consist of multiple venous channels within. The cavernous sinuses are intimately related to the internal carotid artery and its associated sympathetic plexus, the oculomotor nerve, the trochlear nerve, the abducens nerve, and the ophthalmic nerve. Cavernous sinuses are connected to the orbit, the pterygopalatine fossa, the infratemporal fossa, the nasopharynx, and the posterior cranial fossa by various foramina, fissures, and canals in the skull base. A multitude of structures in close relation to the cavernous sinus give rise to a myriad of possible pathologic conditions that can be broadly classified into (a) neoplastic, (b) vascular, (c) infective or inflammatory, or (d) miscellaneous lesions. These pathologic conditions can have overlapping clinical manifestations. Hence, imaging plays a crucial role in identifying the disease, assessing its extent, providing a pertinent differential diagnosis to guide further management, and suggesting a site or route for biopsy. MRI is the modality of choice to depict the cavernous sinuses, with CT and digital subtraction angiography playing supplementary roles in certain situations. In this article, the cavernous sinus lesions encountered in our institution during a 10-year period are reviewed. The purpose of the article is to (a) describe the anatomy of the cavernous sinus; (b) demonstrate the multimodality imaging spectrum of a wide variety of pathologic conditions involving the cavernous sinus, correlating with the histopathologic findings; (c) highlight important imaging clues for differential diagnosis; and (d) help the reader overcome potential pitfalls in interpretation. Online supplemental material is available for this article. ^©RSNA, 2019.

OBJECTIVE:Machine learning has recently gained considerable attention because of promising results for a wide range of radiology applications. Here we review recent work using machine learning in brain tumor imaging, specifically segmentation and MRI radiomics of gliomas. CONCLUSION/CONCLUSIONS:We discuss available resources, state-of-the-art segmentation methods, and machine learning radiomics for glioma. We highlight the challenges of these techniques as well as the future potential in clinical diagnostics, prognostics, and decision making.

BACKGROUND:Hearing preservation is a goal for many patients with vestibular schwannoma. We examined pretreatment magnetic resonance imaging (MRI) and posttreatment hearing outcome after stereotactic radiosurgery. METHODS:From 2004 to 2014, a cohort of 125 consecutive patients with vestibular schwannoma (VS) treated via stereotactic radiosurgery (SRS) were retrospectively reviewed. MRIs containing three-dimensional constructive interference in steady state or equivalent within 1 year before treatment were classified by two radiologists for pretreatment characteristics. "Good" hearing was defined as American Academy of Otolaryngology-Head and Neck Surgery class A. Poor hearing outcome was defined as loss of good pretreatment hearing after stereotactic radiosurgery. RESULTS:Sixty-one patients met criteria for inclusion. Most had tumors in the distal internal auditory canal (55%), separated from the brainstem (63%), oval shape (64%) without cysts (86%), and median volume of 0.85 ± 0.55 cm. Pretreatment audiograms were performed a median of 108 ± 173 days before stereotactic radiosurgery; 38% had good pretreatment hearing. Smaller tumor volume (p < 0.005) was the only variable associated with good pretreatment hearing. 49 (80%) patients had posttreatment audiometry, with median follow-up of 197 ± 247 days. Asymmetrically decreased pretreatment cochlear CISS signal on the side of the VS was the only variable associated with poor hearing outcome (p = 0.001). Inter-rater agreement on cochlear three-dimensional constructive interference in steady state preservation was 91%. CONCLUSIONS:Decreased cochlear CISS signal may indicate a tumor's association with the cochlear neurovascular bundle, influencing endolymph protein concentration and creating an inability to preserve hearing. This important MRI characteristic can influence planning, counseling, and patient selection for vestibular schwannoma treatment.

OBJECTIVE Approximately 75%-92% of patients with trigeminal neuralgia (TN) achieve pain relief after Gamma Knife surgery (GKS), although a proportion of these patients will experience recurrence of their pain. To evaluate the reasons for durability or recurrence, this study determined the impact of trigeminal nerve length and volume, the nerve dose-volume relationship, and the presence of neurovascular compression (NVC) on pain outcomes after GKS for TN. METHODS Fifty-eight patients with 60 symptomatic nerves underwent GKS for TN between 2013 and 2015, including 15 symptomatic nerves secondary to multiple sclerosis (MS). High-resolution MRI was acquired the day of GKS. The median maximum dose was 80 Gy for initial GKS and 65 Gy for repeat GKS. NVC, length and volume of the trigeminal nerve within the subarachnoid space of the posterior fossa, and the ratio of dose to nerve volume were assessed as predictors of recurrence. RESULTS Follow-up was available on 55 patients. Forty-nine patients (89.1%) reported pain relief (Barrow Neurological Institute [BNI] Grades I-IIIb) after GKS at a median duration of 1.9 months. The probability of maintaining pain relief (BNI Grades I-IIIb) without requiring resumption or an increase in medication was 93% at 1 year and 84% at 2 years for patients without MS, and 68% at 1 year and 51% at 2 years for all patients. The nerve length, nerve volume, target distance from the brainstem, and presence of NVC were not predictive of pain recurrence. Patients with a smaller volume of nerve (< 35% of the total nerve volume) that received a high dose (>/= 80% isodose) were less likely to experience recurrence of their TN pain after 1 year (mean time to recurrence: < 35%, 32.2 +/- 4.0 months; > 35%, 17.9 +/- 2.8 months, log-rank test, chi2 = 4.3, p = 0.039). CONCLUSIONS The ratio of dose to nerve volume may predict recurrence of TN pain after GKS. Prospective studies are needed to determine the optimal dose to nerve volume ratio and whether this will result in longer pain-free outcomes.

Dynamic contrast-enhanced (DCE) MR imaging uses rapid sequential MR image acquisition before, during, and after intravenous contrast administration to elucidate information on the microvascular biologic function of tissues. The derived pharmacokinetic parameters provide useful information on tissue perfusion and permeability that may help to evaluate entities that otherwise appear similar by conventional imaging. When specifically applied to the evaluation of head and neck cancer, DCE-MR imaging may provide valuable information to help predict treatment response, discriminate between posttreatment changes and residual tumor, and discriminate between various head and neck neoplasms.

Perfusion and permeability computed tomography and MR imaging applied to head and neck cancer provide powerful diagnostic and prognostic tools for clinicians. Understanding the basics of these techniques allows the radiologist to make informed decisions regarding the use of modeling algorithms, acquisition parameters, and postprocessing techniques. This helps to ensure that studies are acquired, analyzed, and reported appropriately and erroneous results are avoided. These techniques are highly automated, widely available, and can be easily and safely incorporated into daily imaging workflow.

Background: Assessment of preoperative imaging is important for operative planning of meningioma resection. A hyperintense rim on T2-weighted (T2W) MRI is frequently thought to represent a CSF cleft between the tumor and the brain, suggesting a clean arachnoid plane. However, brain invasion (loss of arachnoid plane) is often encountered instead. We sought to further characterize this radiographic finding and identify correlates with intraoperative brain invasion and pathology in patients undergoing meningioma resection. Methods: Retrospective review of 42 patients (mean age: 54.2 years, SD: 13.0, 76% female) who underwent meningioma resection between 2013 and 2016 at a single institution. Demographic variables and pathology results were recorded. Radiographic variables on preoperative MRI included presence and size of a hyperintense rim on T2W MRI, a contrast-enhancing rim on fluid-attenuated inversion recovery (FLAIR), and adjacent edema. Operative reports were reviewed for identification of loss of arachnoid plane (brain invasion) noted during surgery. Radiographic findings were then correlated with brain invasion using nonparametric statistics. Results: Of 42 meningiomas resected, there were 29 (69%) WHO Grade I, 12 (29%) WHO Grade II, and 1 (2%) WHO Grade III. Twenty-three tumors (55%) were located at the skull base. On preoperative T2WI, 36 (86%) of meningiomas demonstrated a hyperintense cleft with a mean width of 2.86 mm (SD: 1.66) and 16 had adjacent edema. Twenty-six meningiomas demonstrated a contrast-enhancing rim on FLAIR with a mean width of 2.85 mm (SD: 1.06) and 28 meningiomas exhibited a rim that was both T2 hyperintense and enhancing on FLAIR. Intraoperatively, 24 (57%) of meningiomas were found to have partial or complete loss of an arachnoid plane between the tumor and adjacent brain parenchyma. Both a hyperintense T2 cleft and enhancing FLAIR rim were associated with loss of arachnoid plane (p=0.004 for T2, <0.001 for FLAIR, <0.001 for combined). Conclusion: Preoperative MRI of meningiomas often identifies a T2 hyperintense rim frequently thought to represent a CSF cleft. A correlation with enhancement on FLAIR and intraoperative loss of arachnoid plane suggests this may be a useful marker of brain invasion and could aid in operative planning and risk assessment