Machine-learning algorithm in acute stroke: real-world experience.

AIM: To assess the clinical performance of a commercially available machine learning (ML) algorithm in acute stroke. MATERIALS AND METHODS: CT and CT angiography (CTA) studies of 104 consecutive patients (43 females, age range 19-93, median age 62) performed for suspected acute stroke at a single tertiary institution with real-time ML software analysis (RAPID™ ASPECTS and CTA) were included. Studies were retrospectively reviewed independently by two neuroradiologists in a blinded manner. RESULTS: The cohort included 24 acute infarcts and 16 large vessel occlusions (LVO). RAPID™ ASPECTS interpretation demonstrated high sensitivity (87.5%) and NPV (87.5%) but very poor specificity (30.9%) and PPV (30.9%) for detection of acute ischaemic parenchymal changes. There was a high percentage of false positives (51.1%). In cases of proven LVO, RAPID™ ASPECTS showed good correlation with neuroradiologists' blinded independent interpretation, Pearson correlation coefficient = 0.96 (both readers), 0.63 (RAPID™ vs reader 1), 0.69 (RAPID™ vs reader 2). RAPID™ CTA interpretation demonstrated high sensitivity (92.3%), specificity (85.3%), and negative predictive (NPV) (98.5%) with moderate positive predictive value (PPV) (52.2%) for detection of LVO (N=13). False positives accounted for 12.5% of cases, of which 27.3% were attributed to arterial stenosis. CONCLUSION: RAPID™ CTA was robust and reliable in detection of LVO. Although demonstrating high sensitivity and NPV, RAPID™ ASPECTS interpretation was associated with a high number of false positives, which decreased clinicians' confidence in the algorithm. However, in cases of proven LVO, RAPID™ ASPECTS performed well and had good correlation with neuroradiologists' blinded interpretation.

Association between white matter ischaemia and carotid plaque morphology as defined by high-resolution in vivo MRI.

OBJECTIVES AND DESIGN: Both carotid plaque morphology and severity of white matter ischaemia (WMI) have been shown to be independent predictors of stroke risk. This study tests the hypothesis that there is an association between carotid plaque morphology as determined by high-resolution carotid MRI and WMI. MATERIALS AND METHODS: Forty patients (80 arteries) with at least 40% stenosis on screening Doppler ultrasound were recruited and underwent high-resolution axial carotid MRI at 1.5 T. In a blinded manner, plaque characteristics such as lipid core, fibrous cap, intraplaque haemorrhage, lumen area, plaque area, and American Heart Association (AHA) classification were qualitatively and quantitatively evaluated. The severity of WMI was independently quantified using a modified Scheltens score based on standard brain Fluid-Attenuated Inversion Recovery. Linear mixed effect models were used to test if carotid plaque characteristics could independently predict severity of WMI. RESULTS: Hypertension (p=0.005) and previous a history of transient ischaemic attack or stroke (p=0.038) were found to be significant predictors of severity of WMI. After accounting for confounding variables, no significant association was found between the modified Scheltens score and lipid core size (p=0.122), fibrous cap status (p=0.991), intraplaque haemorrhage (p=0.708), plaque area (0.835), lumen area (0.371) or an AHA Type VI complex plaque (p=0.195). CONCLUSIONS: Carotid plaque morphology as defined by MRI does not independently predict severity of WMI.

Temporal dependence of in vivo USPIO-enhanced MRI signal changes in human carotid atheromatous plaques.

INTRODUCTION: Ultrasmall superparamagnetic iron oxide (USPIO)-enhanced MRI has been shown to be a useful modality to image activated macrophages in vivo, which are principally responsible for plaque inflammation. This study determined the optimum imaging time-window to detect maximal signal change post-USPIO infusion using T1-weighted (T1w), T2*-weighted (T2*w) and quantitative T2* (qT2*) imaging. METHODS: Six patients with an asymptomatic carotid stenosis underwent high resolution T1w, T2*w and qT2* MR imaging of their carotid arteries at 1.5 T. Imaging was performed before and at 24, 36, 48, 72 and 96 h after USPIO (Sinerem, Guerbet, France) infusion. Each slice showing atherosclerotic plaque was manually segmented into quadrants and signal changes in each quadrant were fitted to an exponential power function to model the optimum time for post-infusion imaging. RESULTS: The power function determining the mean time to convergence for all patients was 46, 41 and 39 h for the T1w, T2*w and qT2* sequences, respectively. When modelling each patient individually, 90% of the maximum signal intensity change was observed at 36 h for three, four and six patients on T1w, T2*w and qT2*, respectively. The rates of signal change decrease after this period but signal change was still evident up to 96 h. CONCLUSION: This study showed that a suitable imaging window for T1w, T2*w and qT2* signal changes post-USPIO infusion was between 36 and 48 h. Logistically, this would be convenient in bringing patients back for one post-contrast MRI, but validation is required in a larger cohort of patients.

Invasive and non-invasive modalities of imaging carotid stenosis.

Despite recent therapeutic advances, acute ischemic complications of atherosclerosis remain the primary cause of morbidity and mortality in Western countries, with carotid atherosclerotic disease one of the major preventable causes of stroke. As the impact of this disease challenges our healthcare systems, we are becoming aware that factors influencing this disease are more complex than previously realized. In current clinical practice, risk stratification relies primarily on evaluation of the degree of luminal stenosis and patient symptomatology. Adequate investigation and optimal imaging are important factors that affect the quality of a carotid endarterectomy (CEA) service and are fundamental to patient selection. Digital subtraction angiography is still perceived as the most accurate imaging modality for carotid stenosis and historically has been the cornerstone of most of the major CEA trials but concerns regarding potential neurological complications have generated substantial interest in non-invasive modalities, such as contrast-enhanced magnetic resonance angiography. The purpose of this review is to give an overview to the vascular specialist of the current imaging modalities in clinical practice to identify patients with carotid stenosis. Advantages and disadvantages of each technique are outlined. Finally, limitations of assessing luminal stenosis in general are discussed. This article will not cover imaging of carotid atheroma morphology, function and other emerging imaging modalities of assessing plaque risk, which look beyond simple luminal measurements.

Characterisation of carotid atheroma in symptomatic and asymptomatic patients using high resolution MRI.

BACKGROUND AND PURPOSE: To prospectively evaluate differences in carotid plaque characteristics in symptomatic and asymptomatic patients using high resolution MRI. METHODS: 20 symptomatic and 20 asymptomatic patients, with at least 50% carotid stenosis as determined by Doppler ultrasound, underwent preoperative in vivo multispectral MRI of the carotid arteries. Studies were analysed both qualitatively and quantitatively in a randomised manner by two experienced readers in consensus, blinded to clinical status, and plaques were classified according to the modified American Heart Association (AHA) criteria. RESULTS: After exclusion of poor quality images, 109 MRI sections in 18 symptomatic and 19 asymptomatic patients were available for analysis. There were no significant differences in mean luminal stenosis severity (72.9% vs 67.6%; p = 0.09) or plaque burden (median plaque areas 50 mm(2) vs 50 mm(2); p = 0.858) between the symptomatic and asymptomatic groups. However, symptomatic lesions had a higher incidence of ruptured fibrous caps (36.5% vs 8.7%; p = 0.004), haemorrhage or thrombus (46.5% vs 14.0%; p<0.001), large necrotic lipid cores (63.8% vs 28.0%; p = 0.002) and complicated type VI AHA lesions (61.5% vs 28.1%; p = 0.001) compared with asymptomatic lesions. The MRI findings of plaque haemorrhage or thrombus had an odds ratio of 5.25 (95% CI 2.08 to 13.24) while thin or ruptured fibrous cap (as opposed to a thick fibrous cap) had an odds ratio of 7.94 (95% CI 2.93 to 21.51) for prediction of symptomatic clinical status. CONCLUSIONS: There are significant differences in plaque characteristics between symptomatic and asymptomatic carotid atheroma and these can be detected in vivo by high resolution MRI.

Comparison of the inflammatory burden of truly asymptomatic carotid atheroma with atherosclerotic plaques in patients with asymptomatic carotid stenosis undergoing coronary artery bypass grafting: an ultrasmall superparamagnetic iron oxide enhanced magnetic resonance study.

INTRODUCTION: Inflammation is a recognized risk factor for the vulnerable atherosclerotic plaque. The aim of this study was to explore whether there is a difference in the degree of Magnetic Resonance (MR) defined inflammation using Ultra Small Super-Paramagnetic Iron Oxide (USPIO) particles, within carotid atheroma in completely asymptomatic individuals and the asymptomatic carotid stenosis in a cohort of patients undergoing coronary artery bypass grafting (CABG). METHODS: 10 patients awaiting CABG with asymptomatic carotid disease and 10 completely asymptomatic individuals with no documented coronary artery disease underwent multi-sequence MR imaging before and 36 hours post USPIO infusion. Images were manually segmented into quadrants and signal change in each quadrant, normalised to adjacent muscle signal, was calculated following USPIO administration. RESULTS: The mean percentage of quadrants showing signal loss was 94% in the CABG group, compared to 24% in the completely asymptomatic individuals (p<0.001). The carotid plaques from the CABG patients showed a significant mean signal intensity decrease of 16.4% after USPIO infusion (95% CI 10.6% to 22.2%; p<0.001). The truly asymptomatic plaques showed a mean signal intensity increase (i.e. enhancement) after USPIO infusion of 8.4% (95% CI 2.6% to 14.2%; p=0.007). The mean signal difference between the two groups was 24.9% (95% CI 16.7% to 33.0%; p<0.001). CONCLUSIONS: These findings are consistent with the hypothesis that inflammatory atheroma is a systemic disease. The carotid territory is more likely to take up USPIO if another vascular territory is symptomatic.

A standardised pathway for the surveillance of stable vestibular schwannoma.

Introduction Conservative management of patients with a stable vestibular schwannoma (VS) places a significant burden on National Health Service (NHS) resources and yet patients' surveillance management is often inconsistent. Our unit has developed a standardised pathway to guide surveillance imaging of patients with stable VS. In this article, we provide the basis for our imaging protocol by reviewing the measurement, natural history and growth patterns of VS, and we present a cost analysis of implementing the pathway both regionally and nationally. Methods Patients with an extrameatal VS measuring ≤20mm in maximal diameter receive magnetic resonance imaging (MRI) six months after their index imaging, followed by three annual MRI scans, two two-year interval MRI scans, a single three-year interval MRI scan and then five-yearly MRI scans to be continued lifelong. Patients with purely intrameatal tumours follow the same protocol but the initial six-month imaging is omitted. A cost analysis of the new pathway was modelled on our unit's retrospective data for 2015 and extrapolated to reflect the cost of VS surveillance nationally. Results Based on an estimation that imaging surveillance would last approximately 25 years (+/- 10 years), the cost of implementing our regional surveillance programme would be £151,011 per year (for 99 new referrals per year) and it would cost the NHS £1,982,968 per year if implemented nationally. Conclusions A standardised surveillance pathway promotes safe practice in the conservative management of VS. The estimated cost of a national surveillance programme compares favourably with other tumour surveillance initiatives, and would enable the NHS to provide a safe and economical service to patients with VS.

Acute hyperammonemic encephalopathy in adults: imaging findings.

BACKGROUND AND PURPOSE: Acute hyperammonemic encephalopathy has significant morbidity and mortality unless promptly treated. We describe the MR imaging findings of acute hyperammonemic encephalopathy, which are not well-recognized in adult patients. MATERIALS AND METHODS: We retrospectively reviewed the clinical and imaging data and outcome of consecutive patients with documented hyperammonemic encephalopathy seen at our institution. All patients underwent cranial MR imaging at 1.5T. RESULTS: Four patients (2 women; mean age, 42 ± 13 years; range, 24-55 years) were included. Causes included acute fulminant hepatic failure, and sepsis with a background of chronic hepatic failure and post-heart-lung transplantation with various systemic complications. Plasma ammonia levels ranged from 55 to 168 µmol/L. Bilateral symmetric signal-intensity abnormalities, often with associated restricted diffusion involving the insular cortex and cingulate gyrus, were seen in all cases, with additional cortical involvement commonly seen elsewhere but much more variable and asymmetric. Involvement of the subcortical white matter was seen in 1 patient only. Another patient showed involvement of the basal ganglia, thalami, and midbrain. Two patients died (1 with fulminant cerebral edema), and 2 patients survived (1 neurologically intact and the other with significant intellectual impairment). CONCLUSIONS: The striking common imaging finding was symmetric involvement of the cingulate gyrus and insular cortex in all patients, with more variable and asymmetric additional cortical involvement. These specific imaging features should alert the radiologist to the possibility of acute hyperammonemic encephalopathy.

Correlation of carotid atheromatous plaque inflammation with biomechanical stress: utility of USPIO enhanced MR imaging and finite element analysis.

OBJECTIVE: The aim of this study was to explore whether there is a relationship between the degree of MR-defined inflammation using ultra small super-paramagnetic iron oxide (USPIO) particles, and biomechanical stress using finite element analysis (FEA) techniques, in carotid atheromatous plaques. METHODS AND RESULTS: 18 patients with angiographically proven carotid stenoses underwent multi-sequence MR imaging before and 36 h after USPIO infusion. T(2)(*) weighted images were manually segmented into quadrants and the signal change in each quadrant normalised to adjacent muscle was calculated after USPIO administration. Plaque geometry was obtained from the rest of the multi-sequence dataset and used within a FEA model to predict maximal stress concentration within each slice. Subsequently, a new statistical model was developed to explicitly investigate the form of the relationship between biomechanical stress and signal change. The Spearman's rank correlation coefficient for USPIO enhanced signal change and maximal biomechanical stress was -0.60 (p=0.009). CONCLUSIONS: There is an association between biomechanical stress and USPIO enhanced MR-defined inflammation within carotid atheroma, both known risk factors for plaque vulnerability. This underlines the complex interaction between physiological processes and biomechanical mechanisms in the development of carotid atheroma. However, this is preliminary data that will need validation in a larger cohort of patients.

Correlation of macrophage location and plaque stress distribution using USPIO-enhanced MRI in a patient with symptomatic severe carotid stenosis: a new insight into risk stratification.

High resolution, USPIO-enhanced MR imaging can be used to identify inflamed atherosclerotic plaque. We report a case of a 79-year-old man with a symptomatic carotid stenosis of 82%. The plaque was retrieved for histology and finite element analysis (FEA) based on the preoperative MR imaging was used to predict maximal Von Mises stress on the plaque. Macrophage location correlated with maximal predicted stresses on the plaque. This supports the hypothesis that macrophages thin the fibrous cap at points of highest stress, leading to an increased risk of plaque rupture and subsequent stroke.

Detection and evaluation of intracranial aneurysms with 16-row multislice CT angiography.

AIM: The aim of this study was to assess the usefulness of 16-row multislice CT angiography (CTA) in evaluating intracranial aneurysms, by comparison with conventional digital subtraction angiography (DSA) and intraoperative findings. METHODS: A consecutive series of 57 patients, scheduled for DSA for suspected intracranial aneurysm, was prospectively recruited to have CTA. This was performed with a 16-detector row machine, detector interval 0.75 mm, 0.5 rotation/s, table speed 10mm/rotation and reconstruction interval 0.40 mm. CTA studies were independently and randomly assessed by two neuroradiologists and a vascular neurosurgeon blinded to the DSA and surgical findings. Review of CTA was performed on workstations with an interactive 3D volume-rendered algorithm. RESULTS: DSA or intraoperative findings or both confirmed 53 aneurysms in 44 patients. For both independent readers, sensitivity and specificity per aneurysm of DSA were 96.2% and 100%, respectively. Sensitivity and specificity of CTA were also 96.2% and 100%, respectively. Mean diameter of aneurysms was 6.3mm (range 1.9 to 28.1 mm, SD 5.2 mm). For aneurysms of less than 3 mm, CTA had a sensitivity of 91.7% for each reader. Although the neurosurgeon would have been happy to proceed to surgery on the basis of CTA alone in all cases, he judged that DSA might have provided helpful additional anatomical information in 5 patients. CONCLUSION: The diagnostic accuracy of 16-slice CTA is promising and appears equivalent to that of DSA for detection and evaluation of intracranial aneurysms. A strategy of using CTA as the primary imaging method, with DSA reserved for cases of uncertainty, appears to be practical and safe.

Utility of an ultrafast magnetic resonance imaging protocol in recent and semi-recent strokes.

OBJECTIVE: To evaluate the technical feasibility of an integrated ultrafast head magnetic resonance (MR) protocol using a sensitivity encoding (SENSE) technique for depicting parenchymal ischaemia and vascular compromise in patients with suspected recent stroke. METHODS: 23 patients were evaluated with the ultrafast MR protocol using T2, T1, fluid attenuated inversion recovery (FLAIR), 3D time of flight magnetic resonance angiography (MRA), and diffusion weighted imaging (DWI) sequences. These were compared with routine conventional MR sequences. RESULTS: One patient could not tolerate conventional imaging, although imaging using the three minute head SENSE protocol was diagnostic. Both conventional and ultrafast protocols were of similar diagnostic yield in the remaining patients. There were no significant differences in clinical diagnostic quality for the T1, T2, FLAIR, and DWI sequences. One MRA examination was of better quality when SENSE was used, owing to reduced motion artefacts and shorter imaging time. CONCLUSIONS: It is possible to undertake a comprehensive MR examination in stroke patients in approximately three to five minutes. Ultrafast imaging may become a useful triage tool before thrombolytic therapy. It may be of particular benefit in patients unable to tolerate longer sequences. Further work is necessary to confirm these findings in hyperacute stroke.

Conventional digital subtraction x-ray angiography versus magnetic resonance angiography in the evaluation of carotid disease: patient satisfaction and preferences.

AIM: To compare conventional digital subtraction x-ray angiography (DSA) and contrast-enhanced magnetic resonance angiography (MRA) of the carotid arteries in terms of patient satisfaction and preferences. METHODS: One hundred and sixty-seven patients with symptomatic carotid artery disease, who underwent both DSA and MRA, were prospectively recruited in this study. Patients' perceptions of each method were assessed by the use of a questionnaire after each procedure. Main outcome measures were anxiety, pain, satisfaction rate and patient preferences. RESULTS: DSA generated more anxiety and pain during the procedure, but the severity of these ill-effects was mild. Satisfaction rates for each method were similar. More patients were, however, willing to have a repeat MRA compared with DSA (67 versus 41%). The majority of patients (62%) preferred MRA over DSA (31%). The shorter MRA imaging time was found to be a significant factor in patients' acceptance of the technique. The main reasons cited by patients for their dislike of a particular procedure was noise and claustrophobia for MRA and invasiveness, pain and post-procedural bed rest for DSA. CONCLUSIONS: MRA is the method that is preferred by the majority of patients, although the actual disutility of DSA may be small. Assuming equal diagnostic accuracy, our data supports replacement of DSA by MRA for routine carotid imaging.

Contrast-enhanced MR angiography for carotid disease: diagnostic and potential clinical impact.

OBJECTIVE: To compare contrast-enhanced MR angiography (CEMRA) with intra-arterial digital subtraction angiography (DSA) for evaluating carotid stenosis. METHODS: A total of 167 consecutive symptomatic patients, scheduled for DSA following screening duplex ultrasound (DUS), were prospectively recruited to have CEMRA. Three independent readers reported on each examination in a blinded and random manner. Agreement was assessed using the Bland-Altman method. Diagnostic and potential clinical impact of CEMRA was evaluated, singly and in combination with DUS. RESULTS: CEMRA tended to overestimate stenosis by a mean bias ranging from 2.4 to 3.8%. A significant part of the disagreement between CEMRA and DSA was directly caused by interobserver variability. For detection of severe stenosis, CEMRA alone had a sensitivity of 93.0% and specificity of 80.6%, with a diagnostic misclassification rate of 15.0% (n = 30). More importantly, clinical decision-making would, however, have been potentially altered only in 6.0% of cases (n = 12). The combination of concordant DUS and CEMRA reduced diagnostic misclassification rate to 10.1% (n = 19) at the expense of 47 (24.9%) discordant cases needing to proceed to DSA. An intermediate approach of selective DUS review resulted in a marginally worse diagnostic misclassification rate of 11.6% (n = 22) but with only 6.8% of discordant cases (n = 13). CONCLUSIONS: DSA remains the gold standard for carotid imaging. The clinical misclassification rate with CEMRA, however, is acceptably low to support its safe use instead of DSA. The appropriateness of combination strategies depends on institutional choice and cost-effectiveness issues.