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.     

Elevated serotonin transporter binding in major depressive disorder assessed using positron emission tomography and [11C]DASB; comparison with bipolar disorder.

BACKGROUND: Altered serotonergic function is thought to play a role in the pathophysiology of major depressive episodes based upon evidence from neuroimaging, pharmacological, postmortem and genetic studies. It remains unclear, however, whether depressed samples that differ with respect to having shown a unipolar versus a bipolar illness course also would show distinct patterns of abnormalities within the serotonergic system. The current study compared serotonin transporter (5-HTT) binding between unipolar-depressives (MDD), bipolar-depressives (BD) and healthy-controls (HC) to assess whether the abnormalities in 5-HTT binding recently found in depressed subjects with BD extend to depressed subjects with MDD. METHODS: The 5-HTT binding-potential (BP) measured using positron emission tomography (PET) and [(11)C]DASB was compared between unmedicated, depressed subjects with MDD (n = 18) or BD (n = 18) and HC (n = 34). RESULTS: Relative to the healthy group both MDD and BD groups showed significantly increased 5-HTT BP in the thalamus (24%, 14%, respectively), insula (15%) and striatum (12%). The unipolar-depressives had elevated 5-HTT BP relative to both BD and HC groups in the vicinity of the periaqueductal gray (PAG, 20%, 22%, respectively). The bipolar-depressives had reduced 5-HTT BP relative to both HC and MDD groups in the vicinity of the pontine raphe nuclei. Depression-severity correlated negatively with 5-HTT BP in the thalamus in MDD-subjects. CONCLUSIONS: The depressed phases of MDD and BD both were associated with elevated 5-HTT binding in the insula, thalamus and striatum, but showed distinct abnormalities in the brainstem. The latter findings conceivably could underlie differences in the patterns of illness symptoms and pharmacological sensitivity observed between MDD and BD.     

Temperature Dependent Adsorption–Desorption Behaviour of Pendimethalin in Punjab Soils

Bulletin of Environmental Contamination and Toxicology - This study shows the effect of soil type and temperature on the adsorption and desorption behaviour of pendimethalin using a batch...     

Granulomatous and lymphocytic hypophysitis – are they immunologically distinct?

Hypophysitis includes three histopathologically distinct entities – granulomatous, lymphocytic and xanthomatous forms. Etiopathogenesis and the immunological differences among these is not well characterized. This study aims to explore the immunopathogenesis of granulomatous and lymphocytic forms of hypophysitis. Demographic, clinical, endocrine function and radiological features of 33 histologically confirmed cases of hypophysitis were reviewed. Immunophenotyping of inflammatory component was performed in 13/33 cases. Visual disturbances (46%), headache (36%), polyuria/polydipsia (6%), menstrual disturbance (6%) and galactorrhoea (6%) were the frequent presenting symptoms. Endocrine abnormalities were noted in 11/18 cases evaluated (61%). Hypothyroidism was the most common endocrine abnormality (33.33%) followed by hyperprolactinaemia (22%) and hypocortisolism (16.66%). On neuroimaging, sellar mass with variable contrast enhancement was observed. On histology, granulomatous hypophysitis (GH) was more common (84.84%) than lymphocytic hypophysitis (LH) (15.15%). In GH, the infiltrate had almost equal proportions of CD3+ T cells and CD68+ histiocytes. Cytotoxic T cells (CD8+) predominated [CD4:CD8 < 1]. CD20+ B cell component ranged from <5% to 50%. Fibrosis, necrosis and giant cells accompanied GH. LH in contrast, had CD4+ T‐helper cell predominance [CD4: CD8 > 1]. CD68+ histiocytes constituted <20% and CD20+ B cells, 5–40% of the infiltrates. In conclusion, GH revealed cytotoxic T cell and histiocyte rich infiltrate in contrast to CD4+ T‐cell predominance in LH suggesting that the two forms have distinct immunological mechanisms in evolution, an autoimmune process in LH and type IV hypersensitivity response in GH.     

Transradial Approach to Cardiovascular Interventions: An Update

Background Since the first cardiac catheterization in 1929, the procedure has continually evolved with advances in understanding, capabilities, and ease of operation. Though historically performed by cut down of the brachial artery, cardiologists soon learned that transfemoral access was both easier to perform and more efficacious with regard to patient outcome. In the last 20 years, the transradial approach has been adopted, and is being utilized with increasing frequency. Methods We conducted a survey of literature published concerning safety, efficacy, cost-effectiveness, and global uptake of transradial catheterization with specific attention to how transradial interventions compare with transfemoral interventions. Results This review of literature indicates that when performed by an experienced interventionalist, radial catheterization is as effective as femoral catheterization and has additional benefits of shorter length of hospital stay and reduced patient costs. Transradial access is superior to transfemoral access in some, but not all, clinical scenarios; in addition, it is an effective alternative for catheterization in patients contraindicated for transfemoral procedures. Adoption of radial access in the United States is at a faster rate than previously expected, though rate of use varies drastically worldwide. Conclusion The transradial approach is an excellent option for carrying out cardiovascular interventions, and will be adopted by more cardiologists in the upcoming years.