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.     

Influence of pain and urinary symptoms on quality of life in young men with chronic prostatitis-like symptoms

BACKGROUND: Our aims in the present study were to estimate the influences of pain and urinary symptoms on quality of life, and to determine which of these two variables has the most predictive power with respect to quality of life in young men with chronic prostatitis-like symptoms. METHODS: Chronic prostatitis-like symptoms were measured by the National Institutes of Health-Chronic Prostatitis Symptom Index. Of the 28,841 men aged 20 years who lived in the study community, 18,495 men (a response rate 64.1%) agreed to participate in the study. A total of 1057 men who complained of symptoms indicative of chronic prostatitis were included in the study. The influences of pain and urinary symptoms on quality of life were determined using logistic regression analysis. The receiver operating characteristic (ROC) curve was used to estimate the predictive ability of each of these variables with respect to quality of life. RESULTS: Results from multivariate analysis showed that both pain and urinary symptoms were associated with an increased likelihood of impaired quality of life, although pain contributed more to a reduced quality of life than urinary symptoms. Relative to men who experienced mild pain, men who experienced moderate pain had a 3.9-fold risk of poor quality of life (odds ratio [OR], 3.87; 95% confidence interval [CI], 2.86-5.23; P < 0.001) and those who experienced severe pain had a 15.7-fold risk of reduced quality of life (OR, 15.68; 95% CI, 6.59-37.35; P < 0.001). Moderate urinary symptoms were associated with a 1.4-fold risk of bother (OR, 1.41; 95% CI, 1.01-1.99; P < 0.001) and severe urinary symptoms were associated with 2.4-fold risk (OR, 2.39; 95% CI, 1.37-4.12; P < 0.001), relative to mild urinary symptoms. Comparison of the effects of pain and urinary symptoms showed that pain severity had the most predictive power for bother, quality of life, and quality-of-life impact. The areas under the ROC curves for bother, quality of life, and quality-of-life impact were 71.3%, 69.3% and 72.5%, respectively. CONCLUSION: Urinary symptoms and pain might be associated with an increased likelihood of impaired quality of life in young men with chronic prostatitis-like symptoms. In addition, our findings suggest that pain severity is the most influential variable for determining quality of life in this population.     

Short-Ti inversion-recovery pulse sequence: analysis and initial experience in cancer imaging

Inversion recovery (IR), commonly considered a pulse sequence capable of producing T1-weighted images with excellent display of normal anatomy, is versatile: The null point and peak time provide a useful, succinct summary of the properties of IR and its capacity for producing both T1- and T2-weighted images. Shortening of the inversion time (TI) and creation of a short-TI inversion-recovery (STIR) pulse sequence increases sensitivity to malignancy and other abnormalities by making the effects of prolonged T1 and T2 on signal intensity additive and by nulling the signal from fat. The authors examined over 300 patients with various malignancies and compared STIR images with T1- and T2-weighted images obtained at 0.5 T. In 43 cases, signal-difference-to-noise ratios (SD/Ns) were calculated between tumor, fat, and muscle. In general, STIR images demonstrated tumor as a conspicuously high-intensity area in a background of muted, discernible anatomic detail. The good contrast achieved with STIR sequences between tumor and fat (SD/N = 18.1) and tumor and muscle (SD/N = 12.9) consolidated into a single image the information contained separately on T1- and T2-weighted images, which facilitates efficient detection and localization of malignancy.