Imaging Findings in MR Imaging–Guided Focused Ultrasound Treatment for Patients with Essential Tremor

BACKGROUND AND PURPOSE:MR imaging–guided focused sonography surgery is a new stereotactic technique that uses high-intensity focused sonography to heat and ablate tissue. The goal of this study was to describe MR imaging findings pre- and post-ventralis intermedius nucleus lesioning by MR imaging–guided focused sonography as a treatment for essential tremor and to determine whether there was an association between these imaging features and the clinical response to MR imaging–guided focused sonography.MATERIALS AND METHODS:Fifteen patients with medication-refractory essential tremor prospectively gave consent; were enrolled in a single-site, FDA-approved pilot clinical trial; and were treated with transcranial MR imaging–guided focused sonography. MR imaging studies were obtained on a 3T scanner before the procedure and 24 hours, 1 week, 1 month, and 3 months following the procedure.RESULTS:On T2-weighted imaging, 3 time-dependent concentric zones were seen at the site of the focal spot. The inner 2 zones showed reduced ADC values at 24 hours in all patients except one. Diffusion had pseudonormalized by 1 month in all patients, when the cavity collapsed. Very mild postcontrast enhancement was seen at 24 hours and again at 1 month after MR imaging–guided focused sonography. The total lesion size and clinical response evolved inversely compared with each other (coefficient of correlation = 0.29, P value = .02).CONCLUSIONS:MR imaging–guided focused sonography can accurately ablate a precisely delineated target, with typical imaging findings seen in the days, weeks, and months following the treatment. Tremor control was optimal early when the lesion size and perilesional edema were maximal and was less later when the perilesional edema had resolved.

MR imaging–guided focused sonography surgery is a new stereotactic technique that uses high-intensity focused sonography to heat and ablate tissue rapidly under closed-loop image guidance and control throughout all steps of the intervention process. MR imaging allows precise intraprocedural localization of the ablation target, verification of safety margins for the sonography treatment, and real-time monitoring of thermal ablation dynamics.^1–7 MR imaging–guided focused sonography is now accepted in the treatment of soft-tissue disorders, including prostate cancer and uterine fibroids. Intracranial applications for brain tumors^8,9 and neuropathic pain syndromes^10,11 are currently under investigation. More recently, MR imaging–guided focused sonography was tested in a clinical trial as a treatment for essential tremor.Essential tremor is a common and disabling movement disorder with an estimated prevalence of 0.3%–5.55%.^12–17 Patients with essential tremor may suffer more from the mental effects on quality of life, such as lower perceived health status,¹⁸ than from actual physical symptoms.¹⁹ Essential tremor may be medically refractory: up to 30% of patients do not respond to first-line therapy and may consider surgical options.²⁰ Improved imaging and refined electrophysiologic localization have demonstrated that the ventralis intermedius nucleus (Vim) of the thalamus is the most effective target. The ventralis intermedius nucleus was the target for the MR imaging–guided focused sonography treatment in the clinical trial mentioned above.The goal of this study was to describe findings on MR imaging both pre- and post-Vim lesioning by MR imaging–guided focused sonography as a treatment for essential tremor in the 15 patients enrolled in the trial and to determine whether there was an association between these imaging features, the number and/or energy of sonications, and the clinical response to MR imaging–guided focused sonography.     

Early-Stage Glioblastomas: MR Imaging–Based Classification and Imaging Evidence of Progressive Growth

BACKGROUND AND PURPOSE:The serial imaging changes describing the growth of glioblastomas from small to large tumors are seldom reported. Our aim was to classify the imaging patterns of early-stage glioblastomas and to define the order of appearance of different imaging patterns that occur during the growth of small glioblastomas.MATERIALS AND METHODS:Medical records and preoperative MR imaging studies of patients diagnosed with glioblastoma between 2006 and 2013 were reviewed. Patients were included if their MR imaging studies showed early-stage glioblastomas, defined as small MR imaging lesions detected early in the course of the disease, demonstrating abnormal signal intensity but the absence of classic imaging findings of glioblastoma. Each lesion was reviewed by 2 neuroradiologists independently for location, signal intensity, involvement of GM and/or WM, and contrast-enhancement pattern on MR imaging.RESULTS:Twenty-six patients with 31 preoperative MR imaging studies met the inclusion criteria. Early-stage glioblastomas were classified into 3 types and were all hyperintense on FLAIR/T2-weighted images. Type I lesions predominantly involved cortical GM (n = 3). Type II (n = 12) and III (n = 16) lesions involved both cortical GM and subcortical WM. Focal contrast enhancement was present only in type III lesions at the gray-white junction. Interobserver agreement was excellent (κ = 0.95; P < .001) for lesion-type classification. Transformations of lesions from type I to type II and type II to type III were observed on follow-up MR imaging studies. The early-stage glioblastomas of 16 patients were pathologically confirmed after imaging progression to classic glioblastoma.CONCLUSIONS:Cortical lesions may be the earliest MR imaging–detectable abnormality in some human glioblastomas. These cortical tumors may progress to involve WM.

Glioblastoma (GB) is the most common primary malignant brain tumor. It typically appears as a large mass with necrosis, prominent edema, mass effect, and strong heterogeneous contrast enhancement when diagnosed. MR imaging, a noninvasive diagnostic tool with excellent tissue contrast, has the potential to detect small GBs. However, it is uncommon to detect small GBs clinically, probably due to nonspecific or absent symptoms. The serial imaging changes depicting the growth of GBs from small to large tumors are seldom reported.Some reports described small MR imaging lesions that subsequently progressed to GBs.^1–11 These are often described as ill-defined, FLAIR or T2-weighted hyperintensities without discernable mass effect that typically involve both the cortex and subcortical WM, but occasionally appear as only cortical lesions.^2,4,8 Contrast enhancement is not a consistent feature and tends to be focal and nodular when present.^6–8 The commonly affected brain areas include frontal (n = 4),^2,3,6,8 parietal (n = 2),^7,10 occipital (n = 1),¹¹ temporal (n = 5),^2,3,6,7,11 hippocampal (n = 3),^1,2,9 and insular (n = 1)⁹ regions. Because these MR imaging lesions were detected early in the course of the disease, they were frequently referred to as early-stage GBs.^3,5–8,11We have noticed different imaging patterns in early-stage GBs. An imaging classification for early-stage GB, however, is not available because most previous studies included only a few such cases. It is important for radiologists to be familiar with early imaging findings and growth patterns of GBs because familiarity may help diagnose small tumors that are symptomatic or incidentally found. Early diagnosis of GB may lead to a higher extent of tumor resection, which has been demonstrated to correlate with patient survival.¹² In this study, we aimed to classify the imaging patterns of early-stage GBs and to the define the order of appearance of different imaging patterns that occur during the growth of these tumors.     

Diffusion-Weighted Imaging–Detected Ischemic Lesions following Endovascular Treatment of Cerebral Aneurysms: A Systematic Review and Meta-Analysis

BACKGROUND AND PURPOSE:Endovascular treatment of intracranial aneurysms is associated with the risk of thromboembolic ischemic complications. Many of these events are asymptomatic and identified only on diffusion-weighted imaging. We performed a systematic review and meta-analysis to study the incidence of DWI positive for thromboembolic events following endovascular treatment of intracranial aneurysms.MATERIALS AND METHODS:A comprehensive literature search identified studies published between 2000 and April 2016 that reported postprocedural DWI findings in patients undergoing endovascular treatment of intracranial aneurysms. The primary outcome was the incidence of DWI positive for thromboembolic events. We examined outcomes by treatment type, sex, and aneurysm characteristics. Meta-analyses were performed by using a random-effects model.RESULTS:Twenty-two studies with 2148 patients and 2268 aneurysms were included. The overall incidence of DWI positive for thromboembolic events following endovascular treatment was 49% (95% CI, 42%–56%). Treatment with flow diversion trended toward a higher rate of DWI positive for lesions than coiling alone (67%; 95% CI, 46%–85%; versus 45%; 95% CI, 33%–56%; P = .07). There was no difference between patients treated with coiling alone and those treated with balloon-assisted (44%; 95% CI, 29%–60%; P = .99) or stent-assisted (43%; 95% CI, 24%–63%; P = .89) coiling. Sex, aneurysm rupture status, location, and size were not associated with the rate of DWI positive for lesions.CONCLUSIONS:One in 2 patients may have infarcts on DWI following endovascular treatment of intracranial aneurysms. There is a trend toward a higher incidence of DWI-positive lesions following treatment with flow diversion compared with coiling. Patient demographics and aneurysm characteristics were not associated with DWI-positive thromboembolic events.

Coil embolization and flow diversion have proved highly efficacious options for the endovascular treatment of intracranial aneurysms. However, both techniques are associated with potential periprocedural complications, including aneurysm rupture, transient ischemic attacks, and ischemic stroke. Small, silent infarcts caused by thromboemboli are often seen on postprocedural diffusion-weighted imaging. While many of these lesions remain ostensibly asymptomatic, the long-term effects of such tiny infarcts remain unclear.^1–3Previous studies have reported that the rate of ischemic lesions on postoperative DWI ranges from 10% to 77% following coil embolization^4–15 and 51% to 63% following therapy with flow diversion.^16–19 However, baseline clinical and angiographic risk factors for postoperative DWI lesions, to our knowledge, have not been fully elucidated previously. We performed a systematic review and meta-analysis for the following: 1) to determine the overall incidence of perioperative infarcts on DWI in patients undergoing endovascular treatment of intracranial aneurysms; and 2) to demonstrate the relationship between treatment type, patient demographics, and aneurysm characteristics with postoperative infarcts on DWI.     

Optimal MR Plaque Imaging for Cervical Carotid Artery Stenosis in Predicting the Development of Microembolic Signals during Exposure of Carotid Arteries in Endarterectomy: Comparison of 4 T1-Weighted Imaging Techniques

BACKGROUND AND PURPOSE:Preoperative identification of plaque vulnerability may allow improved risk stratification for patients considered for carotid endarterectomy. The present study aimed to determine which plaque imaging technique, cardiac-gated black-blood fast spin-echo, magnetization-prepared rapid acquisition of gradient echo, source image of 3D time-of-flight MR angiography, or noncardiac-gated spin-echo, most accurately predicts development of microembolic signals during exposure of carotid arteries in carotid endarterectomy.MATERIALS AND METHODS:Eighty patients with ICA stenosis (≥70%) underwent the 4 sequences of preoperative MR plaque imaging of the affected carotid bifurcation and then carotid endarterectomy under transcranial Doppler monitoring of microembolic signals in the ipsilateral middle cerebral artery. The contrast ratio of the carotid plaque was calculated by dividing plaque signal intensity by sternocleidomastoid muscle signal intensity.RESULTS:Microembolic signals during exposure of carotid arteries were detected in 23 patients (29%), 3 of whom developed new neurologic deficits postoperatively. Those deficits remained at 24 hours after surgery in only 1 patient. The area under the receiver operating characteristic curve to discriminate between the presence and absence of microembolic signals during exposure of the carotid arteries was significantly greater with nongated spin-echo than with black-blood fast spin-echo (difference between areas, 0.258; P < .0001), MPRAGE (difference between areas, 0.106; P = .0023), or source image of 3D time-of-flight MR angiography (difference between areas, 0.128; P = .0010). Negative binomial regression showed that in the 23 patients with microembolic signals, the contrast ratio was associated with the number of microembolic signals only in nongated spin-echo (risk ratio, 1.36; 95% confidence interval, 1.01–1.97; P < .001).CONCLUSIONS:Nongated spin-echo may predict the development of microembolic signals during exposure of the carotid arteries in carotid endarterectomy more accurately than other MR plaque imaging techniques.

For appropriately selected patients, carotid endarterectomy (CEA) can effectively prevent stroke,^1–3 with few neurologic deficits observed immediately following the procedure. Surgical site embolism represents >70% of intraoperative procedure-related strokes.⁴ When one monitors the middle cerebral artery by using intraoperative transcranial Doppler (TCD), microembolic signals (MES) are detected in >90% of patients undergoing CEA^4–6; however, the quality and quantity of MES detected depends on the stage of CEA.^5–7 During exposure procedures for the carotid arteries, plaque that represents a source of emboli and has not been removed remains exposed to blood flow. Under such conditions, manipulation of the carotid arteries can dislodge emboli from the surgical site into the intracranial arteries.⁸ Furthermore, because the target vessel remains closed during the exposure procedure, detectable MES are thought to represent solid masses, such as thrombi, necrosis, or lipid.⁶ In contrast, once the walls of the carotid arteries are cut for endarterectomy, a high number of harmless gaseous MES may develop during carotid declamping due to air entering the lumen of the arteries.^6,9 Detection of MES during the exposure procedure has been shown to correlate with postoperative neurologic deficits immediately after CEA.^5–7,9–11Several investigators have compared MES during the exposure procedure for the carotid arteries in CEA with histopathologic findings of excised carotid plaque and have demonstrated that development of the MES was strongly associated with vulnerable carotid plaques consisting primarily of intraplaque hemorrhage and/or intraluminal thrombus.^12,13 Intraplaque hemorrhage might cause formation of intraluminal thrombus likely due to chemical mediators, increased stenosis, or changes in eddy currents, though the associations among these remain unclear. Other research has shown that more cerebrovascular adverse events related to CEA occurred in patients with atheromatous plaques compared with patients with fibrous plaques.⁹ Preoperative identification of plaque vulnerability may thus allow improved risk stratification for patients considered for CEA.Intraplaque characteristics are generally assessed by using MR imaging based on T1-weighted sequences,¹⁴ and the detection of intraplaque hemorrhage on preoperative MR imaging is associated with the development of MES during the procedure for exposure of the carotid arteries.¹² However, there has been inconsistency among published findings on vulnerable plaques.¹⁵ This could be due to interinstitutional differences in the methodology for such imaging techniques as cardiac-gated black-blood fast spin-echo (BB-FSE),^16–19 magnetization-prepared rapid acquisition of gradient echo,^12,20–22 source image of 3D time-of-flight MR angiography (SI-MRA),²³ and noncardiac-gated spin-echo (SE).^15,24,25 Although the cardiac-gated BB-FSE method is most commonly used for T1-weighted MR plaque imaging,^17,18 the TR is dependent on a single R-R interval from electrocardiography, which occasionally results in an overly long TR to diminish proton density–weighted contrast and to enhance T1-weighted contrast.²⁵In addition to cardiac gating, proton density–weighted contrast is preserved when using T1-weighted spoiled gradient-echo techniques, which are generally used for MRA.²³ The use of T1-weighted spoiled gradient-echo techniques on SI-MRA could result in insufficient contrast between fibrous and lipid/necrotic plaques.¹⁵ Originally developed for direct thrombus imaging, MPRAGE is a modified sequence in which the TI is set to permit black-blood effects.²¹ Because the signal intensity of the lipid/necrotic component tends to show T1 values similar to those of blood, the intensity can theoretically be attenuated.¹⁵ The substantial influence of the proton density and inversion recovery pulse can be avoided in nongated SE; however, this sequence requires a relatively long acquisition time and is known to be susceptible to patient motion even when motion correction is used.¹⁵ Among these 4 kinds of imaging techniques, substantial variation is observed in the contrast provided by T1-weighted MR plaque imaging and its ability to characterize intraplaque components. Furthermore, quantitative color-coded MR plaque imaging performed by using the nongated SE sequence has recently been shown to provide accurate evaluation of the composition (ie, fibrous tissue, lipid/necrosis, or hemorrhage) of excised carotid plaques compared with histopathologic findings in patients undergoing CEA.²⁶The purpose of the present study was thus to determine which plaque imaging technique, BB-FSE, MPRAGE, SI-MRA, or nongated SE, all of which are variations of T1-weighted imaging, can most accurately predict development of MES during exposure of the carotid arteries in CEA.     

Beyond Isolated and Associated: A Novel Fetal MR Imaging–Based Scoring System Helps in the Prenatal Prognostication of Callosal Agenesis

BACKGROUND AND PURPOSE:Although “corpus callosum agenesis” is an umbrella term for multiple entities, prenatal counseling is based reductively on the presence (associated) or absence (isolated) of additional abnormalities. Our aim was to test the applicability of a fetal MR neuroimaging score in a cohort of fetuses with prenatally diagnosed isolated corpus callosum agenesis and associated corpus callosum agenesis and correlate it with neurodevelopmental outcomes.MATERIALS AND METHODS:We performed a single-center retrospective analysis of a cohort of cases of consecutive corpus callosum agenesis collected between January 2011 and July 2019. Cases were scored by 2 raters, and interater agreement was calculated. Outcome was assessed by standardized testing (Bayley Scales of Infant and Toddler Development, Kaufman Assessment Battery for Children) or a structured telephone interview and correlated with scores using 2-way ANOVA.RESULTS:We included 137 cases (74 cases of isolated corpus callosum agenesis), imaged at a mean of 27 gestational weeks. Interrater agreement was excellent (0.98). Scores were higher in associated corpus callosum agenesis (P < .0001) without a significant score difference between complete and partial corpus callosum agenesis (P = .38). Outcome was assessed in 42 children with isolated corpus callosum agenesis and 9 with associated corpus callosum agenesis (mean age, 3.1 years). MR imaging scores correctly predicted developmental outcome in 90.7% of patients with isolated corpus callosum agenesis, improving neurodevelopmental risk stratification in corpus callosum agenesis.CONCLUSIONS:The scoring system is very reproducible and can differentiate isolated corpus callosum agenesis and associated isolated corpus callosum agenesis (significantly higher scores) but not between partial and complete corpus callosum agenesis. Scores correlated with outcome in isolated corpus callosum agenesis, but there were too few associated postnatal cases of isolated corpus callosum agenesis to draw conclusions in this group.

Corpus callosum (CC) agenesis (CCA) is one of the most common malformations of the CNS.¹ Rather than a single entity, CCA is an umbrella term defined by anatomy, independent of etiology or outcome. To further complicate matters, CCA includes several subtypes, including complete (when the entire CC is missing) and partial (when part but not all of the CC is absent), but CCA often also includes several degrees of hypoplasia (CC present but of reduced dimensions) and dysgenesis (CC present yet malformed). Each option may be seen in isolation (no other fetal brain or body malformations) or in the context of a polymalformative or genetic condition.^1-3In terms of outcome risk stratification, patients with associated CCA (aCCA) are at a high risk of neurodevelopmental delay,⁴ while isolated CCA (iCCA) is associated with development within the normal range in up to 88% of children.^5-9 Other features, such as the presence of Probst bundles or sigmoid bundles, have been used inconsistently in an attempt to predict outcome.^10,11 In an attempt to improve risk stratification in iCCA, we developed and tested a score based on anatomic features evaluated on fetal brain MR imaging in patients with detailed postnatal neuropsychological outcomes.¹²This study aimed to test the validity of an MR imaging score initially developed for isolated congenital CCA in a heterogeneous group of complete and partial CCA and to correlate it with neurodevelopmental outcomes.     

Vulnerable Carotid Atherosclerotic Plaque Creation in a Swine Model: Evaluation of Stenosis Creation Using Absorbable and Permanent Suture in a Diabetic Dyslipidemic Model

PurposeTo compare the creation of carotid atherosclerotic plaque after stenosis creation with absorbable or permanent suture in a diabetic dyslipidemic swine model.Materials and MethodsA high-cholesterol diet was fed to 15 Sinclair pigs. Diabetes was induced by intraarterial injection of streptozotocin. Stenosis creation in carotid arteries was performed with an absorbable or a permanent suture assigned randomly on both sides. After 20 weeks, Doppler ultrasound (US), angiography, and intravascular US examinations were performed before sacrifice. Carotid, coronary, and femoral arteries were analyzed by histology according to the American Heart Association (AHA) classification.ResultsThree animals died during the perioperative period, and three others died during follow-up. Diabetes was successfully induced in all surviving animals (9 of 15). On angiography, stenoses were estimated at 80.4%±12.4 in carotid arteries with permanent sutures and at 48.8%±39 with absorbable sutures (P = .03). With permanent suturing, carotid plaques were observed in all animals with five of nine manifesting an AHA stage IV or more. With absorbable suture, atherosclerosis developed in seven of nine carotid arteries including three animals with an AHA stage IV or more. Advanced coronary and femoral plaques were observed in four and one of the nine animals. A correlation between AHA classes of coronary plaques and cholesterol level was observed (P = .01), whereas for carotid arteries, AHA class correlated with the degree of stenosis (P = .045).ConclusionsCreation of atheromatous lesions in carotid and coronary arteries was successful with this model despite a high mortality rate. Less severe carotid stenoses and advanced plaques were observed with absorbable sutures.     

Usefulness of Non–Contrast-Enhanced MR Angiography Using a Silent Scan for Follow-Up after Y-Configuration Stent-Assisted Coil Embolization for Basilar Tip Aneurysms

BACKGROUND AND PURPOSE:Y-configuration stent-assisted coil embolization is used for treating wide-neck aneurysms. Noninvasive alternatives to x-ray DSA for follow-up after Y-configuration stent-assisted coil embolization treatment are required. This study aimed to assess the usefulness of non–contrast-enhanced MRA by using a Silent Scan (silent MRA) for follow-up after Y-configuration stent-assisted coil embolization for basilar tip aneurysms.MATERIALS AND METHODS:Seven patients treated with Y-configuration stent-assisted coil embolization for basilar tip aneurysms underwent silent MRA, 3D TOF-MRA, and DSA. Silent MRA and 3D TOF-MRA images were obtained during the same scan session on a 3T MR imaging system. Two neuroradiologists independently reviewed both types of MRA images and subjectively scored the flow in the stents on a scale of 1 (not visible) to 5 (nearly equal to DSA) by referring to the latest DSA image as a criterion standard. Furthermore, we evaluated the visualization of the neck remnant.RESULTS:In all patients, the 2 observers gave a higher score for the flow in the stents on silent MRA than on 3D TOF-MRA. The average score ± standard deviation was 4.07 ± 0.70 for silent MRA and 1.93 ± 0.80 (P < .05) for 3D TOF-MRA. Neck remnants were depicted by DSA in 5 patients. In silent MRA, neck remnants were depicted in 5 patients, and visualization was similar to DSA; however, in 3D TOF-MRA, neck remnants were depicted in only 1 patient.CONCLUSIONS:Silent MRA might be useful for follow-up after Y-configuration stent-assisted coil embolization.

In recent years, intracranial stents have been used for the treatment of wide-neck aneurysms. The Y-configuration stent-assisted coil embolization technique has been generally used for wide-neck bifurcation aneurysms such as those at the tip of the basilar artery (BA).^1–6 The Y stent deploys 2 stents from the BA to the bilateral posterior cerebral artery (PCA). The 2 stents overlap in the distal BA trunk, with 1 stent penetrating the mesh of the other. The use of 2 stents in a “Y” configuration to assist with coil embolization for bifurcation aneurysms has been accepted for broad-neck aneurysms.X-ray DSA is the standard technique used for follow-up after an intracranial stent. However, DSA is an invasive technique that carries a risk of neurologic complications, contrast materials, and x-ray radiation.^7–10On the other hand, 3D TOF-MRA is widely used as a noninvasive substitute for DSA for the follow-up of coiled aneurysms.^11–14 Although there have been reports of 3D TOF-MRA being used after stent-assisted coil embolization,^12,13 it remains difficult to visualize flow in an intracranial stent when using this method because of magnetic susceptibility and radiofrequency shielding. Therefore, contrast-enhanced MRA (CE-MRA) is used for follow-up after stent-assisted coil embolization. However, the use of contrast materials in CE-MRA is associated with nephrogenic systemic fibrosis and anaphylactic shock; therefore, this technique might not be appropriate for repeated examinations.^15–17 Furthermore, it has been reported that gadolinium-based contrast material accumulates in the dentate nucleus and globus pallidus.¹⁸Silent MRA uses a Silenz pulse sequence (GE Healthcare, Milwaukee, Wisconsin) containing an ultrashort echo time (UTE) combined with arterial spin-labeling. Data acquisition is based on 3D radial sampling, and the arterial spin-labeling technique is used as a preparation pulse for visualization of the blood flow.^19,20 It is a non–contrast-enhanced MRA technique; therefore, it is better for the patient and suitable for repeated follow-ups.UTE of silent MRA minimizes phase dispersion of the labeled blood flow signal and decreases magnetic susceptibility to coils and stents. Thus, silent MRA can evaluate the blood flow in an intracranial stent.²⁰To the best of our knowledge, there have been no studies of the use of non–contrast-enhanced MRA for follow-up after Y-configuration stent-assisted coil embolization for basilar tip aneurysms. Therefore, in the present study, we evaluated the usefulness of silent MRA compared with 3D TOF-MRA for follow-up after Y-configuration stent-assisted coil embolization for basilar tip aneurysms.     

Development and Validation of a Deep Learning–Based Model to Distinguish Glioblastoma from Solitary Brain Metastasis Using Conventional MR Images

BACKGROUND AND PURPOSE:Differentiating glioblastoma from solitary brain metastasis preoperatively using conventional MR images is challenging. Deep learning models have shown promise in performing classification tasks. The diagnostic performance of a deep learning–based model in discriminating glioblastoma from solitary brain metastasis using preoperative conventional MR images was evaluated.MATERIALS AND METHODS:Records of 598 patients with histologically confirmed glioblastoma or solitary brain metastasis at our institution between February 2006 and December 2017 were retrospectively reviewed. Preoperative contrast-enhanced T1WI and T2WI were preprocessed and roughly segmented with rectangular regions of interest. A deep neural network was trained and validated using MR images from 498 patients. The MR images of the remaining 100 were used as an internal test set. An additional 143 patients from another tertiary hospital were used as an external test set. The classifications of ResNet-50 and 2 neuroradiologists were compared for their accuracy, precision, recall, F1 score, and area under the curve.RESULTS:The areas under the curve of ResNet-50 were 0.889 and 0.835 in the internal and external test sets, respectively. The area under the curve of neuroradiologists 1 and 2 were 0.889 and 0.768 in the internal test set and 0.857 and 0.708 in the external test set, respectively.CONCLUSIONS:A deep learning–based model may be a supportive tool for preoperative discrimination between glioblastoma and solitary brain metastasis using conventional MR images.

Glioblastoma (GBM) and brain metastases are the most common malignant tumors in adults.¹ These 2 entities have different treatment options, and it is therefore essential to distinguish them promptly to determine the proper treatment strategy. In patients with a history of underlying malignancy and conventional MR imaging findings of multiple enhancing lesions, a diagnosis can be made easily. However, approximately 25%–30% of brain metastases present as single lesions, and in lung cancer—the most common cancer to metastasize to the brain—approximately 50% of patients are thought to have brain metastases as the initial presentation.^2,3 In addition, GBM and solitary brain metastasis have overlapping MR imaging features, including rim enhancement with perilesional T2 hyperintensity, and are thus difficult to differentiate preoperatively.⁴ However, GBM has an infiltrative growth pattern; therefore, tumor cells diffusely infiltrate beyond the enhancing portion, manifesting as a perilesional T2 hyperintense region. Brain metastases have similar MR imaging features; however, this perilesional T2 hyperintensity is primarily due to vasogenic edema caused by the leaky capillary vessels of the enhancing tumor.^5,6 In an effort to detect these microstructural differences, various advanced MR imaging techniques, such as perfusion MR imaging, MR spectroscopy, and diffusion tensor imaging, have been applied to distinguish GBM from solitary brain metastasis, with particular emphasis on the aforementioned perilesional T2 hyperintense region.^7-10 Collectively, these studies have shown promising results indicating that the perilesional T2 hyperintense region, along with the enhancing portion itself, carries valuable information that may preoperatively distinguish these 2 entities. However, advanced imaging techniques require additional scanning time, and their quantitative values can vary depending on the imaging parameters, posing difficult challenges for practical application.Recently, radiomics have been used to analyze various textural and handcrafted features to classify or predict prognosis of disease through medical images that are beyond the perception of human eye.^11,12 However, radiomics needs careful preprocessing steps, including delicate segmentation. Deep learning—a subfield in machine learning—extracts information directly from the data, omitting the step of manual feature extraction in decision making.¹³ In the field of neuro-oncology, specifically glioma imaging, previous studies have shown the potential of deep learning for classifying gliomas based on genetic mutations or clinical outcomes.^14-16In this study, we hypothesized that deep learning may differentiate GBM from solitary brain metastasis without extraction of predefined features. Thus, we aimed to develop a deep learning–based model to differentiate GBM from solitary brain metastasis using preoperative T2-weighted and contrast-enhanced (CE) T1-weighted MR images and further validate its diagnostic performance.     

Development of the Fetal Vermis: New Biometry Reference Data and Comparison of 3 Diagnostic Modalities–3D Ultrasound, 2D Ultrasound,and MR Imaging

BACKGROUND AND PURPOSE:Normal biometry of the fetal posterior fossa rules out most major anomalies of the cerebellum and vermis. Our aim was to provide new reference data of the fetal vermis in 4 biometric parameters by using 3 imaging modalities, 2D ultrasound, 3D ultrasound, and MR imaging, and to assess the relation among these modalities.MATERIALS AND METHODS:A retrospective study was conducted between June 2011 and June 2013. Three different imaging modalities were used to measure vermis biometry: 2D ultrasound, 3D ultrasound, and MR imaging. The vermian parameters evaluated were the maximum superoinferior diameter, maximum anteroposterior diameter, the perimeter, and the surface area. Statistical analysis was performed to calculate centiles for gestational age and to assess the agreement among the 3 imaging modalities.RESULTS:The number of fetuses in the study group was 193, 172, and 151 for 2D ultrasound, 3D ultrasound, and MR imaging, respectively. The mean and median gestational ages were 29.1 weeks, 29.5 weeks (range, 21–35 weeks); 28.2 weeks, 29.05 weeks (range, 21–35 weeks); and 32.1 weeks, 32.6 weeks (range, 27–35 weeks) for 2D ultrasound, 3D ultrasound, and MR imaging, respectively. In all 3 modalities, the biometric measurements of the vermis have shown a linear growth with gestational age. For all 4 biometric parameters, the lowest results were those measured by MR imaging, while the highest results were measured by 3D ultrasound. The inter- and intraobserver agreement was excellent for all measures and all imaging modalities. Limits of agreement were considered acceptable for clinical purposes for all parameters, with excellent or substantial agreement defined by the intraclass correlation coefficient.CONCLUSIONS:Imaging technique–specific reference data should be used for the assessment of the fetal vermis in pregnancy.

Imaging of the fetal posterior fossa is considered a routine part of the fetal sonographic examination. Normal sonographic biometry and normal morphology of the posterior fossa rule out most major anomalies of the fetal cerebellum and vermis.¹ However, in case of an abnormal posterior fossa, evaluation of the vermian biometry and morphology is of paramount importance, considering the wide clinical spectrum of this imaging finding.²Fetal posterior fossa anomalies range from benign asymptomatic conditions to severe abnormalities associated with neurologic impairment.^3–6 The most frequent of these anomalies, Blakes pouch cyst, vermian hypoplasia, and Dandy-Walker malformation, have a similar imaging appearance^7,8 but different vermian biometry and, therefore, different prognoses.⁹Many anomalies of the posterior fossa can be depicted with sonography alone.¹⁰ Although the standard axial imaging planes may detect most anomalies of the posterior fossa, the diagnosis of the exact type of abnormality might be challenging because a clear visualization of the midsagittal plane is essential. Subtle changes in the morphology of the vermis are hidden by this axial view, and this feature can lead to false-positive diagnoses of vermian pathologies.^11,12Our group has proposed using the transabdominal sagittal plane for visualization of the fetal vermis,¹³ while Malinger et al¹⁴ reported their experience with the transvaginal approach. Vinals et al¹⁵ used volume contrast imaging (VCI) on plane C to construct nomograms for the normal fetal vermis. Our group used this same VCI on plane C technique to compare normal and abnormal fetal vermis measurements, and we concluded that the 3D sonographic technique has many advantages in the detection of posterior fossa anomalies.¹⁶ It allows off-line evaluation and reconstruction of images, even with abnormal angles when the midsagittal plane is difficult to obtain.MR imaging is a well-known complementary tool in the prenatal diagnosis of fetal brain abnormalities. The challenges described above in achieving a good visualization of the midsagittal plane in prenatal sonography led to frequent use of this tool to assess, more accurately, the structures of the posterior fossa and improve prenatal diagnosis.Various nomograms have been developed to establish normal biometric measures of the fetal vermis by using ultrasound (US) or MR imaging.^17–19 None of these nomograms provided data regarding all 4 vermian biometric parameters. Moreover, there were no comparisons among all 3 imaging modalities.²⁰The aims of our study were the following: to provide normal reference biometric data of the fetal vermis in 4 biometric parameters for 3 imaging modalities, to evaluate the reproducibility of the vermian biometry, and to compare the measurements obtained by 2D sonography, 3D sonography, and MR imaging.     

Treatment of Supratentorial Spontaneous Intracerebral Hemorrhage Using Image-Guided Minimally Invasive Surgery: Initial Experiences of a Flat Detector CT–Based Puncture Planning and Navigation System in the Angiographic Suite

BACKGROUND AND PURPOSE:The intracerebral hemorrhage drainage through minimally invasive approach is emerging as an alternative for traditional craniotomy, due to its improved survival rate and reduced complication rate. In this study, we investigated the feasibility and safety of a flat detector CT–based puncture planning and navigation system for minimally invasive hematoma drainage on patients with intracerebral hemorrhage.MATERIALS AND METHODS:The minimally invasive hematoma drainage was performed on 21 hypertensive patients with intracerebral hemorrhage in the angiographic suite with the guidance of a flat detector CT–based puncture planning and navigation system. This system is integrated in the angiographic machine, and was used for 1) planning the needle path based on a preprocedural flat detector CT scan, 2) advancing the catheter with real-time fluoroscopic guidance, and 3) confirming the procedure outcome based on an immediate postprocedural flat detector CT. The surgery efficiency, accuracy, and the treatment outcome were measured and compared with the published data.RESULTS:All procedures were successfully completed with the catheter placed 4 ± 1 mm from the planned position. The average surgery time was 40 ± 7 minutes. The volume of the hematoma was reduced to 28 ± 4% of the original volume. The Glasgow Coma Scale score was significantly improved from 10 ± 1 at the admission to 14 ± 1 at the discharge. The Extended Glasgow Coma Scale score also improved from 5 ± 1 at the discharge to 6 ± 1 at the 6-month follow-up. No major complication, rebleeding, and mortality were observed in this study.CONCLUSIONS:This flat detector CT–based needle guidance system provided a feasible, convenient, and safe way to perform the puncture and drainage of brain hematoma in the angiographic suite.

Spontaneous intracerebral hemorrhage (ICH) affects more than 2 million people each year around the world.¹ Timely medical intervention is critical for ICH, as it is a medical emergency with a mortality rate range from 35% to 52% within 1 month.² However, the current treatments for patients with ICH continue to be controversial, with several large-scale randomized trials reporting no significant benefit of early surgery, primarily craniotomy, over conservative treatment.^3–5 In recent years, the surgical removal of hematoma through burr-holes is emerging as a minimally invasive alternative for conventional craniotomy because of its improved complication rate and survival rate.⁶ These new techniques usually involve stereotactic positioning and intraprocedural image guidance, such as endoscopic, traditional CT, intraoperative CT, and sonography-guided aspiration.^6–8 These techniques have not yet been widely used because of expensive clinical costs such as the demand for specialized equipment, difficulties scheduling procedures on the diagnostic machines (eg, CT scan suite), and a lack of confirmed clinical benefit from large-scale clinical trials. Until now, the implementation of minimally invasive surgery for treating intracranial hemorrhage remains investigational, with only 2 ongoing large-scale randomized trials (Minimally Invasive Surgery and Thrombolytic Evacuation in ICH [MISTIE] and Clot Lysis Evaluating Accelerated Resolution III [CLEAR III]).^9,10 MISTIE and CLEAR III are currently phase III trials aimed at comparing minimally invasive surgery combined with a thrombolysis agent with the best critical management alone for treating ICH, and with minimally invasive surgery combined with placebo for treating intraventricular hemorrhage, respectively. Both of their early results indicated that the catheter location is one of the most important criteria for good surgery outcome, and careful planning and accurate execution is necessary for optimal catheter placement.¹⁰In this study, we attempted an affordable clinical solution for image-guided hematoma drainage that could be performed in the traditional angiographic suite without the need for specialized equipment. We implemented an innovative puncture planning and navigation system that is integrated into the angiographic machine for hematoma evacuation. This guided evacuation procedure based on an intraprocedural flat detector CT (FDCT), in which a 3D CT-like image was reconstructed from a rotational scan¹¹using a C-arm angiographic machine. Commercial software (syngo iGuide needle guidance; Siemens, Erlangen, Germany) then facilitates the needle path planning that allows physicians to choose optimum entry and target points of the puncture, and automatically overlays the path onto the fluoroscopic images during the procedure for real-time guidance.The feasibilities of this puncture planning and guidance system had been previously demonstrated in selective cervical nerve root block,¹² percutaneous kidney puncture,¹³ and hepatic tumor ablation¹³ based on ex vivo animal models and real patients. Its application in the brain was only recently demonstrated on 6 patients with cerebrovascular ischemia for external ventricular drainage.¹⁴ In this study, we further expanded this technique on a more complicated brain procedure, intracranial hemorrhage evacuation in 21 hypertensive patients with ICH. Unlike the external ventricular drainage, the location of hemorrhage in patients with ICH could vary from patient to patient, and could not be performed freehand without any image guidance. Moreover, we assessed the feasibility and the safety of this technique in terms of procedure length, puncture accuracy, drainage volume, and treatment outcome at the discharge and 6-month follow-up. This is also the first attempt to evaluate the short-term treatment outcome of FDCT-guided hemorrhage drainage using Glasgow Coma Scale (GCS) and Extended Glasgow Coma Scale (GOSE) scores.     

Motion and flow insensitive adiabatic T2‐preparation module for cardiac MR imaging at 3 tesla

A versatile method for generating T₂‐weighting is a T₂‐preparation module, which has been used successfully for cardiac imaging at 1.5T. Although it has been applied at 3T, higher fields (B₀ ≥ 3T) can degrade B₀ and B₁ homogeneity and result in nonuniform magnetization preparation. For cardiac imaging, blood flow and cardiac motion may further impair magnetization preparation. In this study, a novel T₂‐preparation module containing multiple adiabatic B₁‐insensitive refocusing pulses is introduced and compared with three previously described modules [(a) composite MLEV4, (b) modified BIR‐4 (mBIR‐4), and (c) Silver‐Hoult–pair]. In the static phantom, the proposed module provided similar or better B₀ and B₁ insensitivity than the other modules. In human subjects (n = 21), quantitative measurement of image signal coefficient of variation, reflecting overall image inhomogeneity, was lower for the proposed module (0.10) than for MLEV4 (0.15, P < 0.0001), mBIR‐4 (0.27, P < 0.0001), and Silver‐Hoult–pair (0.14, P = 0.001) modules. Similarly, qualitative analysis revealed that the proposed module had the best image quality scores and ranking (both, P < 0.0001). In conclusion, we present a new T₂‐preparation module, which is shown to be robust for cardiac imaging at 3T in comparison with existing methods. Magn Reson Med 70:1360–1368, 2013. © 2012 Wiley Periodicals, Inc.