Search

This study trained and evaluated algorithms to detect, segment, and classify simple and complex pleural effusions on computed tomography (CT) scans.For detection and segmentation, we randomly selected 160 chest CT scans out of all consecutive patients (January 2016-January 2021, n = 2659) with reported pleural effusion. Effusions were manually segmented and a negative cohort of chest CTs from 160 patients without effusions was added. A deep convolutional neural network (nnU-Net) was trained and cross-validated (n = 224; 70%) for segmentation and tested on a separate subset (n = 96; 30%) with the same distribution of reported pleural complexity features as in the training cohort (eg, hyperdense fluid, gas, pleural thickening and loculation). On a separate consecutive cohort with a high prevalence of pleural complexity features (n = 335), a random forest model was implemented for classification of segmented effusions with Hounsfield unit thresholds, density distribution, and radiomics-based features as input. As performance measures, sensitivity, specificity, and area under the curves (AUCs) for detection/classifier evaluation (per-case level) and Dice coefficient and volume analysis for the segmentation task were used.Sensitivity and specificity for detection of effusion were excellent at 0.99 and 0.98, respectively (n = 96; AUC, 0.996, test data). Segmentation was robust (median Dice, 0.89; median absolute volume difference, 13 mL), irrespective of size, complexity, or contrast phase. The sensitivity, specificity, and AUC for classification in simple versus complex effusions were 0.67, 0.75, and 0.77, respectively.Using a dataset with different degrees of complexity, a robust model was developed for the detection, segmentation, and classification of effusion subtypes. The algorithms are openly available at https://github.com/usb-radiology/pleuraleffusion.git.

Background Artificial intelligence can assist in cardiac image interpretation. Here, we achieved a substantial reduction in time required to read a cardiovascular magnetic resonance (CMR) study to estimate left atrial volume without compromising accuracy or reliability. Rather than deploying a fully automatic black-box, we propose to incorporate the automated LA volumetry into a human-centric interactive image-analysis process. Methods and results Atri-U, an automated data analysis pipeline for long-axis cardiac cine images, computes the atrial volume by: (i) detecting the end-systolic frame, (ii) outlining the endocardial borders of the LA, (iii) localizing the mitral annular hinge points and constructing the longitudinal atrial diameters, equivalent to the usual workup done by clinicians. In every step human interaction is possible, such that the results provided by the algorithm can be accepted, corrected, or re-done from scratch. Atri-U was trained and evaluated retrospectively on a sample of 300 patients and then applied to a consecutive clinical sample of 150 patients with various heart conditions. The agreement of the indexed LA volume between Atri-U and two experts was similar to the inter-rater agreement between clinicians (average overestimation of 0.8 mL/m2 with upper and lower limits of agreement of − 7.5 and 5.8 mL/m2, respectively). An expert cardiologist blinded to the origin of the annotations rated the outputs produced by Atri-U as acceptable in 97% of cases for step (i), 94% for step (ii) and 95% for step (iii), which was slightly lower than the acceptance rate of the outputs produced by a human expert radiologist in the same cases (92%, 100% and 100%, respectively). The assistance of Atri-U lead to an expected reduction in reading time of 66%—from 105 to 34 s, in our in-house clinical setting. Conclusions Our proposal enables automated calculation of the maximum LA volume approaching human accuracy and precision. The optional user interaction is possible at each processing step. As such, the assisted process sped up the routine CMR workflow by providing accurate, precise, and validated measurement results.

CT patterns of viral pneumonia are usually only qualitatively described in radiology reports. Artificial intelligence enables automated and reliable segmentation of lungs with chest CT. Based on this, the purpose of this study was to derive meaningful imaging biomarkers reflecting CT patterns of viral pneumonia and assess their potential to discriminate between healthy lungs and lungs with viral pneumonia. This study used non-enhanced and CT pulmonary angiograms (CTPAs) of healthy lungs and viral pneumonia (SARS-CoV-2, influenza A/B) identified by radiology reports and RT-PCR results. After deep learning segmentation of the lungs, histogram-based and threshold-based analyses of lung attenuation were performed and compared. The derived imaging biomarkers were correlated with parameters of clinical and biochemical severity (modified WHO severity scale; c-reactive protein). For non-enhanced CTs (n = 526), all imaging biomarkers significantly differed between healthy lungs and lungs with viral pneumonia (all p < 0.001), a finding that was not reproduced for CTPAs (n = 504). Standard deviation (histogram-derived) and relative high attenuation area [600–0 HU] (HU-thresholding) differed most. The strongest correlation with disease severity was found for absolute high attenuation area [600–0 HU] (r = 0.56, 95% CI = 0.46–0.64). Deep-learning segmentation-based histogram and HU threshold analysis could be deployed in chest CT evaluation for the differentiating of healthy lungs from AP lungs.

Background CT has emerged as an important diagnostic tool in COVID-19, but the underlying pathological changes behind CT findings are not yet fully elucidated. Purpose The purpose of this retrospective study was to correlate CT patterns of fatal cases of COVID-19 with post-mortem pathology observations. Material and Methods The study included 70 lung lobes of 14 patients who died from RT-PCR confirmed COVID-19. All patients underwent ante-mortem CT and autopsy between March 9 and April 30, 2020. Board-certified radiologists and pathologists performed lobe-wise correlations of pulmonary observations. In a consensus reading, 267 radiological and 257 histopathological observations of the lungs were recorded and systematically graded according to severity. These observations were matched and evaluated. Results Predominant CT observations were ground glass opacities (GGO; 59 of 70 lobes examined) and areas of consolidation (33/70). The histopathological observations were consistent with diffuse alveolar damage (70/70), capillary dilatation and congestion (70/70), often accompanied by microthrombi (27/70), superimposed acute bronchopneumonia (17/70) and leukocytoclastic vasculitis (7/70). Four patients had pulmonary emboli. Bronchial wall thickening on CT histologically corresponded with acute bronchopneumonia. GGOs and consolidations corresponded with mixed histopathological observations including capillary dilation and congestion, interstitial edema, diffuse alveolar damage and microthrombosis. Vascular alterations were prominent observations in both CT and histopathology. Conclusion A significant proportion of GGO correlated with the pathologic processes of diffuse alveolar damage, capillary dilatation and congestion and microthrombosis. Our results confirm the presence and underline the importance of vascular alterations as a key pathophysiological driver in lethal COVID-19.

Highlights • It is feasible to develop clinically useful AI-based software for quantification of pulmonary opacities in COVID-19 in just 10 days. • An established pipeline for fast transition of prototypes to full clinical implementation is an important key to success. • Human-level performance, even in the presence of advanced disease, was achieved with less than 200 chest CT scans for training of the AI algorithm., Purpose During the emerging COVID-19 pandemic, radiology departments faced a substantial increase in chest CT admissions coupled with the novel demand for quantification of pulmonary opacities. This article describes how our clinic implemented an automated software solution for this purpose into an established software platform in 10 days. The underlying hypothesis was that modern academic centers in radiology are capable of developing and implementing such tools by their own efforts and fast enough to meet the rapidly increasing clinical needs in the wake of a pandemic. Method Deep convolutional neural network algorithms for lung segmentation and opacity quantification on chest CTs were trained using semi-automatically and manually created ground-truth (Ntotal = 172). The performance of the in-house method was compared to an externally developed algorithm on a separate test subset (N = 66). Results The final algorithm was available at day 10 and achieved human-like performance (Dice coefficient = 0.97). For opacity quantification, a slight underestimation was seen both for the in-house (1.8 %) and for the external algorithm (0.9 %). In contrast to the external reference, the underestimation for the in-house algorithm showed no dependency on total opacity load, making it more suitable for follow-up. Conclusions The combination of machine learning and a clinically embedded software development platform enabled time-efficient development, instant deployment, and rapid adoption in clinical routine. The algorithm for fully automated lung segmentation and opacity quantification that we developed in the midst of the COVID-19 pandemic was ready for clinical use within just 10 days and achieved human-level performance even in complex cases.

OBJECTIVE: To extract pulmonary and cardiovascular metrics from chest CTs of patients with coronavirus disease 2019 (COVID-19) using a fully automated deep learning-based approach and assess their potential to predict patient management. MATERIALS AND METHODS: All initial chest CTs of patients who tested positive for severe acute respiratory syndrome coronavirus 2 at our emergency department between March 25 and April 25, 2020, were identified (n = 120). Three patient management groups were defined: group 1 (outpatient), group 2 (general ward), and group 3 (intensive care unit [ICU]). Multiple pulmonary and cardiovascular metrics were extracted from the chest CT images using deep learning. Additionally, six laboratory findings indicating inflammation and cellular damage were considered. Differences in CT metrics, laboratory findings, and demographics between the patient management groups were assessed. The potential of these parameters to predict patients' needs for intensive care (yes/no) was analyzed using logistic regression and receiver operating characteristic curves. Internal and external validity were assessed using 109 independent chest CT scans. RESULTS: While demographic parameters alone (sex and age) were not sufficient to predict ICU management status, both CT metrics alone (including both pulmonary and cardiovascular metrics; area under the curve [AUC] = 0.88; 95% confidence interval [CI] = 0.79-0.97) and laboratory findings alone (C-reactive protein, lactate dehydrogenase, white blood cell count, and albumin; AUC = 0.86; 95% CI = 0.77-0.94) were good classifiers. Excellent performance was achieved by a combination of demographic parameters, CT metrics, and laboratory findings (AUC = 0.91; 95% CI = 0.85-0.98). Application of a model that combined both pulmonary CT metrics and demographic parameters on a dataset from another hospital indicated its external validity (AUC = 0.77; 95% CI = 0.66-0.88). CONCLUSION: Chest CT of patients with COVID-19 contains valuable information that can be accessed using automated image analysis. These metrics are useful for the prediction of patient management.

Background Although frontal lobe resections account for one-third of intralobar resections in pediatric epilepsy surgery, there is a dearth of information regarding long-term seizure freedom, overall cognitive and adaptive functioning. Objective To identify outcome predictors and define the appropriate timing for surgery. Methods We retrospectively analyzed the data of 75 consecutive patients aged 10.0 ± 4.9 yr at surgery that had an 8.1 yr mean follow-up. Results Etiology comprised focal cortical dysplasia (FCD) in 71% and benign tumors in 16% cases. All patients but one had a magnetic resonance imaging-visible lesion. At last follow-up, 63% patients remained seizure-free and 37% had discontinued antiepileptic drugs. Presurgical predictors of seizure freedom were a shorter epilepsy duration, strictly regional epileptic discharges in electroencephalography (EEG), and an epileptogenic zone and/or lesion distant from eloquent cortex. Postsurgical predictors were the completeness of resection and the lack of early postoperative seizures or epileptic discharges in EEG. Higher presurgical overall cognitive and adaptive functioning was related to later epilepsy onset and to a sublobar epileptogenic zone and/or lesion. Following surgery, scores remained stable in the majority of patients. Postsurgical gains were determined by higher presurgical performance and tumors vs FCD. Conclusion Our findings highlight the favorable long-term outcomes following frontal lobe epilepsy surgery in childhood and adolescence and underline the importance of early surgical intervention in selected candidates. Early postsurgical relapses and epileptic discharges in EEG constitute key markers of treatment failure and should prompt timely reevaluation. Postsurgical overall cognitive and adaptive functioning is stable in most patients, whereas those with benign tumors have higher chances of improvement.

Background In contrast to the abundance of seizure outcome reports in epilepsy surgery for glioneuronal tumors in childhood and adolescence, there is a dearth of information regarding cognitive outcomes. Objective To investigate the seizure and cognitive outcome of children and adolescents that underwent resective surgery for glioneuronal tumor-associated refractory epilepsy and determine their predictive factors. Methods We retrospectively analyzed the presurgical findings, resection types, and outcomes over 1.3 to 12.3 years (mean, 7.3) of 29 consecutive patients, who underwent resection in 2000 to 2011. The mean age at epilepsy onset was 7.9 years (range, 0-15.4), the mean age at surgery was 11.7 years (range, 2.6-17.3), and the mean epilepsy duration to surgery was 3.8 years (range, 0.3-15.3). Etiology comprised 13 dysembryoplastic neuroepithelial tumors and 16 gangliogliomas, with additional focal cortical dysplasia in 5 cases. Results Eighty-six percent of children were seizure free 12 months after surgery; at final follow-up, 76% remained seizure free and 62% had discontinued antiepileptic drugs. Gross total resection was related to significantly higher rates of seizure freedom. Higher presurgical cognitive functioning (full-scale IQ, verbal IQ) was related to shorter epilepsy duration to surgery independent of age at epilepsy onset, thus determining postsurgical functioning. Improvements in verbal IQ, performance IQ, and visual memory as well as a trend toward improvement in full-scale IQ were established after surgery. Despite individual losses in full-scale IQ, verbal or visual memory, no deterioration was noted in any cognitive variable on a group level. Conclusion Completeness of resection predisposes to favorable outcomes regarding seizure alleviation. Whereas cognitive functioning deteriorates with time in glioneuronal tumor-related refractory epilepsy, surgery is linked to improvement rather than to deterioration on a group level.

BACKGROUND AND PURPOSE: Fiber tractography is increasingly used in the preoperative evaluation of endangered fiber bundles. From a clinical point of view, an accurate and methodologically transparent procedure is desired. Our aim was to evaluate the recently described global tracking algorithm compared with other established methods, such as deterministic and probabilistic tractography. MATERIALS AND METHODS: Twenty patients, candidates for excision of epileptogenic lesions, were subjected to higher-angular resolution diffusion imaging–based fiber tractography. Seed points were created without manual bias, predominantly by FreeSurfer and voxel-based atlases. We focused on 2 important fiber bundles, namely the descending motor pathways and the optic radiation. Postoperatively, the accuracy of the predicted fiber route was controlled by structural MR imaging and by inflicted functional deficits. RESULTS: Among the 3 evaluated methods, global tracking was the only method capable of reconstructing the full extent of the descending motor pathways, including corticobulbar fibers from the area of face representation. Still, probabilistic tractography depicted the optic radiation better, especially the Meyer loop. The deterministic algorithm performed less adequately. CONCLUSIONS: The probabilistic method seems to be the best balance between computational time and effectiveness and seems to be the best choice in most cases, particularly for the optic radiation. If, however, a detailed depiction of the fiber anatomy is intended and tract crossings are implicated, then the computationally time-consuming global tracking should be preferred.

To visualize and quantify physiological blood flow of intracranial veins in vivo using time-resolved, 3D phase-contrast MRI (4D flow MRI), and to test measurement accuracy. Fifteen healthy volunteers underwent repeated ECG-triggered 4D flow MRI (3 Tesla, 32-channel head coil). Intracranial venous blood flow was analysed using dedicated software allowing for blood flow visualization and quantification in analysis planes at the superior sagittal, straight, and transverse sinuses. MRI was evaluated for intra- and inter-observer agreement and scan-rescan reproducibility. Measurements of the transverse sinuses were compared with transcranial two-dimensional duplex ultrasound. Visualization of 3D blood flow within cerebral sinuses was feasible in 100 % and within at least one deep cerebral vein in 87 % of the volunteers. Blood flow velocity/volume increased along the superior sagittal sinus and was lower in the left compared to the right transverse sinus. Intra- and inter-observer reliability and reproducibility of blood flow velocity (mean difference 0.01/0.02/0.02 m/s) and volume (mean difference 0.0002/-0.0003/0.00003 l/s) were good to excellent. High/low velocities were more pronounced (8 % overestimation/9 % underestimation) in MRI compared to ultrasound. Four-dimensional flow MRI reliably visualizes and quantifies three-dimensional cerebral venous blood flow in vivo and is promising for studies in patients with sinus thrombosis and related diseases. • 4D flow MRI can be used to visualize and quantify physiological cerebral venous haemodynamics • Flow quantification within cerebral sinuses reveals high reliability and accuracy of 4D flow MRI • Blood flow volume and velocity increase along the superior sagittal sinus • Limited spatial resolution currently precludes flow quantification in small cerebral veins

Decompressive craniectomy (DC) is a controversially discussed neurosurgical procedure to reduce elevated intracranial pressure after severe traumatic brain injury (TBI). In contrast to adults, several studies could show a benefit for the pediatric population, but still DC is considered as an emergency procedure only. The aim of our study was to identify secondary complications and long-term sequelae of the procedure. All children presenting to the University Medical Center Freiburg between 2005 and 2013 who underwent DC after severe TBI were retrospectively reviewed with respect to complications and outcome. Twelve children were included with a mean Glasgow Coma Scale of 4.5 ± 1.7. The most frequent complications after TBI and DC were formation of hygroma (83%), aseptic bone resorption of the reimplanted bone flap (50%), posttraumatic hydrocephalus (42%), secondary infection or dysfunction of ventriculoperitoneal shunt (25%) or cranioplasty (33%), and epilepsy (33%). Because of these complications, 75% of patients required further surgery in addition to cranioplasty with up to eight interventions. At follow-up, mean Glasgow Outcome Scale was 3.3 ± 1.2. Within our patient population, we demonstrated high incidence of complications after DC, leading to further surgical procedures and longer hospitalization. These potential complications have to be considered in any decision about DC as an emergency procedure.

Primary deficiency of laminin alpha-2 due to mutations in the LAMA2 gene accounts for 30% of all patients with congenital muscular dystrophy. Here, we present seven patients with partial or total laminin alpha-2 deficiency (MDC1A) with a wide clinical spectrum, ranging from ambulant patients to patients who were never able to stand or sit. We identified two pathogenic mutations in the LAMA2 gene in all patients except for one patient in whom only one mutation was found. Six of the mutations were previously undescribed. In some of the milder cases, laminin alpha-2 expression in the muscle biopsy was only slightly reduced. These findings emphasize that analysis of the LAMA2 gene might be necessary in patients with muscle weakness, cerebral white matter changes and high creatine kinase levels, even in the presence of laminin alpha-2 in the muscle biopsy.

MR Images of the brain in Multiple Sclerosis (MS) show regions of signal abnormalities that can provide information for the diagnosis and for the pathogenesis of the disease. Two very commonly used MRI contrasts in this context are the T1 weighted (T1-w) and the FLAIR. This study shows that additional information can be extracted from the Susceptibility Weighted MRI (SWI) contrast. In particular, the signal and the contrast of white matter lesions in SWI are examined and compared to T1-w and FLAIR contrasts. The lesions are analysed into hypo- and hyper-intense. Additionally, the spatial distributions for the two lesion types are computed and summarised with their expected distance from the ventricles. The data from 19 MS patients and 23 controls have been acquired and examined. The results show the presence of two lesion classes in SWI for MS patients, while T1-w and FLAIR contrast mechanisms present only a single class each. The hypo-intense SWI lesions appear closer to the ventricles and are more correlated to the T1-w signal rather than the FLAIR signal.

Global fiber reconstruction aims at providing a consistent view of the fiber architecture in the whole volume of cerebral white matter on the basis of diffusion-sensitized magnetic resonance imaging. A new realization of this principle is presented. The method utilizes data acquired with high angular resolution diffusion imaging (HARDI), a measurement method that fulfills clinical requirements. For the first time among global reconstruction methods, the computation time is acceptable for a broad class of practical applications. The method does not involve any boundary conditions that prescribe the location of the ends of reconstructed fibers. This helps to minimize necessary user interaction and operator dependence. Results obtained in a physical phantom demonstrate a high reconstruction quality. In vivo results have been obtained in several volunteers. The algorithm found a number of prominent fascicles including those in the limbic system, which had been problematic for a previously published version of global tracking.

We present a 49-year-old male who suffered a cerebellar infarction due to a vertebral artery dissection. The patient had taken sildenafil daily for at least 2 years for sexual enhancement. There was no sexual intercourse or traumatic event prior to symptom onset. Sildenafil intake has been associated with aortic dissection and, in the light of this report, we suggest that chronic sildenafil intake could be a risk factor for arterial dissection.