A Stacked Generalization of 3D Orthogonal Deep Learning Convolutional Neural Networks for Improved Detection of White Matter Hyperintensities in 3D FLAIR Images

BACKGROUND AND PURPOSE:Accurate and reliable detection of white matter hyperintensities and their volume quantification can provide valuable clinical information to assess neurologic disease progression. In this work, a stacked generalization ensemble of orthogonal 3D convolutional neural networks, StackGen-Net, is explored for improving automated detection of white matter hyperintensities in 3D T2-FLAIR images.MATERIALS AND METHODS:Individual convolutional neural networks in StackGen-Net were trained on 2.5D patches from orthogonal reformatting of 3D-FLAIR (n = 21) to yield white matter hyperintensity posteriors. A meta convolutional neural network was trained to learn the functional mapping from orthogonal white matter hyperintensity posteriors to the final white matter hyperintensity prediction. The impact of training data and architecture choices on white matter hyperintensity segmentation performance was systematically evaluated on a test cohort (n = 9). The segmentation performance of StackGen-Net was compared with state-of-the-art convolutional neural network techniques on an independent test cohort from the Alzheimer’s Disease Neuroimaging Initiative-3 (n = 20).RESULTS:StackGen-Net outperformed individual convolutional neural networks in the ensemble and their combination using averaging or majority voting. In a comparison with state-of-the-art white matter hyperintensity segmentation techniques, StackGen-Net achieved a significantly higher Dice score (0.76 [SD, 0.08], F1-lesion (0.74 [SD, 0.13]), and area under precision-recall curve (0.84 [SD, 0.09]), and the lowest absolute volume difference (13.3% [SD, 9.1%]). StackGen-Net performance in Dice scores (median = 0.74) did not significantly differ (P = .22) from interobserver (median = 0.73) variability between 2 experienced neuroradiologists. We found no significant difference (P = .15) in white matter hyperintensity lesion volumes from StackGen-Net predictions and ground truth annotations.CONCLUSIONS:A stacked generalization of convolutional neural networks, utilizing multiplanar lesion information using 2.5D spatial context, greatly improved the segmentation performance of StackGen-Net compared with traditional ensemble techniques and some state-of-the-art deep learning models for 3D-FLAIR.

White matter hyperintensities (WMHs) correspond to pathologic features of axonal degeneration, demyelination, and gliosis observed within cerebral white matter.¹ Clinically, the extent of WMHs in the brain has been associated with cognitive impairment, Alzheimer’s disease and vascular dementia, and increased risk of stroke.^2,3 The detection and quantification of WMH volumes to monitor lesion burden evolution and its correlation with clinical outcomes have been of interest in clinical research.^4,5 Although the extent of WMHs can be visually scored,⁶ the categoric nature of such scoring systems makes quantitative evaluation of disease progression difficult. Manually segmenting WMHs is tedious, prone to inter- and intraobserver variability, and is, in most cases, impractical. Thus, there is an increased interest in developing fast, accurate, and reliable computer-aided automated techniques for WMH segmentation.Convolutional neural network (CNN)-based approaches have been successful in several semantic segmentation tasks in medical imaging.⁷ Recent works have proposed using deep learning–based methods for segmenting WMHs using 2D-FLAIR images.^8-11 More recently, a WMH segmentation challenge¹² was also organized (http://wmh.isi.uu.nl/) to facilitate comparison of automated segmentation of WMHs of presumed vascular origin in 2D multislice T2-FLAIR images. Architectures that used an ensemble of separately trained CNNs showed promising results in this challenge, with 3 of the top 5 winners using ensemble-based techniques.¹²Conventional 2D-FLAIR images are typically acquired with thick slices (3–4 mm) and possible slice gaps. Partial volume effects from a thick slice are likely to affect the detection of smaller lesions, both in-plane and out-of-plane. 3D-FLAIR images, with isotropic resolution, have been shown to achieve higher resolution and contrast-to-noise ratio¹³ and have shown promising results in MS lesion detection using 3D CNNs.¹⁴ Additionally, the isotropic resolution enables viewing and evaluation of the images in multiple planes. This multiplanar reformatting of 3D-FLAIR without the use of interpolating kernels is only possible due to the isotropic nature of the acquisition. Network architectures that use information from the 3 orthogonal views have been explored in recent works for CNN-based segmentation of 3D MR imaging data.¹⁵ The use of data from multiple planes allows more spatial context during training without the computational burden associated with full 3D training.¹⁶ The use of 3 orthogonal views simultaneously mirrors how humans approach this segmentation task.Ensembles of CNNs have been shown to average away the variances in the solution and the choice of model- and configuration-specific behaviors of CNNs.¹⁷ Traditionally, the solutions from these separately trained CNNs are combined by averaging or using a majority consensus. In this work, we propose the use of a stacked generalization framework (StackGen-Net) for combining multiplanar lesion information from 3D CNN ensembles to improve the detection of WMH lesions in 3D-FLAIR. A stacked generalization¹⁸ framework learns to combine solutions from individual CNNs in the ensemble. We systematically evaluated the performance of this framework and compared it with traditional ensemble techniques, such as averaging or majority voting, and state-of-the-art deep learning techniques.     

Attachment to Caregivers and Type 1 Diabetes in Children

Attachment is a behavioral and physiological system, which enables individual’s dynamic adaptation to its environment. Attachment develops in close interaction between an infant and his/her mother, plays an important role in the development of the infant’s brain, and influences the quality of interpersonal relationships throughout life.Security of attachment is believed to influence individual response to stress, exposing insecurely organized individuals to deregulated autonomic nervous system and exaggerated hypothalamic-pituitary-adrenal activity, which, in turn, produces increased and prolonged exposure to stress-hormones. Such stress responses may have considerable implications for the development of diverse health-risk conditions, such as insulin resistance and hyperlipidemia, shown by numerous studies.Although the mechanisms are not yet fully understood, there is compelling evidence highlighting the role of psychological stress in the development of type 1 diabetes (T1D). One of the possible contributing factors for the development of T1D may be the influence of attachment security on individual stress reactivity. Thus, the suggestion is that insecurely attached individuals are more prone to experience increased and prolonged influence of stress hormones and other mechanisms causing pancreatic beta-cell destruction.The present paper opens with a short overview of the field of attachment in children, the principal attachment classifications and their historic development, describes the influence of attachment security on individual stress-reactivity and the role of the latter in the development of T1D. Following is a review of recent literature on the attachment in patients with T1D with a conclusion of a proposed role of attachment organization in the etiology of T1D.     

Secondary haemophagocytic lymphohistiocytosis: Experience from the Uppsala University Hospital

Background. Haemophagocytic lymphohistiocytosis (HLH) is a rare clinical syndrome characterized by fever, hepatosplenomegaly, cytopenia, and progressive multiple-organ failure. HLH in adults is often secondary to autoimmune diseases, cancer, or infections in contrast to familial HLH. Treatment of secondary HLH is directed against the triggering disease in addition to immunosuppressive therapy, the latter commonly according to the HLH-2004 protocol.Methods. We conducted a retrospective study to identify triggering diseases, disease-specific and immunosuppressive therapy administered, and prognosis in adult patients with secondary HLH. Patient data were collected from October 2010 to January 2015.Results. Ten adult patients with secondary HLH were identified. Seven were men, and the median age at diagnosis was 62 years. Five cases were triggered by malignant disease and five by infection. The median patient fulfilled five of the eight HLH-2004 diagnostic criteria. All patients fulfilled the criteria fever, cytopenia, and ferritin >500 µg/L. Median time from hospital admission to HLH diagnosis was 20 days. Four patients received immunosuppressive therapy according to the HLH-2004 protocol. The prognosis was dismal, especially for the patients with malignancy-associated HLH, of whom all died.Conclusion. HLH should be suspected in patients who present with fever, cytopenia, and ferritin >500 µg/L. Secondary HLH has a dismal prognosis. None of the patients with HLH triggered by malignancy survived. Achieving remission of the triggering disease seems to be important for a favourable outcome as, in all surviving patients, the haemophagocytic syndrome resolved after remission of the underlying infection.     

Pathophysiology and management of sensitive skin: position paper from the special interest group on sensitive skin of the International Forum for the Study of Itch (IFSI)

The special interest group on sensitive skin of the International Forum for the Study of Itch previously defined sensitive skin as a syndrome defined by the occurrence of unpleasant sensations (stinging, burning, pain, pruritus and tingling sensations) in response to stimuli that normally should not provoke such sensations. This additional paper focuses on the pathophysiology and the management of sensitive skin. Sensitive skin is not an immunological disorder but is related to alterations of the skin nervous system. Skin barrier abnormalities are frequently associated, but there is no cause and direct relationship. Further studies are needed to better understand the pathophysiology of sensitive skin – as well as the inducing factors. Avoidance of possible triggering factors and the use of well-tolerated cosmetics, especially those containing inhibitors of unpleasant sensations, might be suggested for patients with sensitive skin. The role of psychosocial factors, such as stress or negative expectations, might be relevant for subgroups of patients. To date, there is no clinical trial supporting the use of topical or systemic drugs in sensitive skin. The published data are not sufficient to reach a consensus on sensitive skin management. In general, patients with sensitive skin require a personalized approach, taking into account various biomedical, neural and psychosocial factors affecting sensitive skin.     

Prevalence and Predictors of Quality of Recovery at Home After Day Surgery

Traditionally, major complications and unanticipated admission/readmission rates were used to assess outcome after day surgery. However, in view of the relative absence of major complications the quality of recovery (QOR) should be considered one of the principal endpoints after day surgery. In our study, the level of QOR is defined by a combination of the Global Surgical Recovery (GSR) Index and the Quality of Life (QOL).The aim of this study was to analyze prevalence and predictors of QOR after day surgery on the fourth postoperative day.Elective patients scheduled for day surgery from November 2008 to April 2010 were enrolled in a prospective cohort study. Outcome parameters were measured by using questionnaire packages at 2 time points: 1 week preoperatively and 4 days postoperatively. Primary outcome parameter is the QOR and is defined as good if the GSR index >80% as well as the postoperative QOL is unchanged or improved as compared with baseline. QOR is defined as poor if both the GSR index ≤80% and if the postoperative QOL is decreased as compared with baseline. QOR is defined as intermediate in all other cases. Three logistic regression analyses were performed to determine predictors for poor QOR after day surgery.A total of 1118 patients were included. A good QOR was noted in 17.3% of patients, an intermediate QOR in 34.8%, and a poor QOR in 47.8% 4 days after day surgery. The best predictor for poor QOR after day surgery was type of surgery. Other predictors were younger age, work status, and longer duration of surgery. A history of previous surgery, expected pain (by the patient) and high long-term surgical fear were significant predictors of poor QOR in only 1 of 3 prediction models.The QOR at home 4 days after day surgery was poor in the majority of patients and showed a significant procedure-specific variation. Patients at risk for poor QOR can be identified during the preoperative period based on type of surgery, age, work status, and the duration of the surgery.