Adversarial Defense Method Based on Latent Representation Guidance for Remote Sensing Image Scene Classification

Deep neural networks have made great achievements in remote sensing image analyses; however, previous studies have shown that deep neural networks exhibit incredible vulnerability to adversarial examples, which raises concerns about regional safety and production safety. In this paper, we propose an adversarial denoising method based on latent representation guidance for remote sensing image scene classification. In the training phase, we train a variational autoencoder to reconstruct the data using only the clean dataset. At test time, we first calculate the normalized mutual information between the reconstructed image using the variational autoencoder and the reference image as denoised by a discrete cosine transform. The reconstructed image is selectively utilized according to the result of the image quality assessment. Then, the latent representation of the current image is iteratively updated according to the reconstruction loss so as to gradually eliminate the influence of adversarial noise. Because the training of the denoiser only involves clean data, the proposed method is more robust against unknown adversarial noise. Experimental results on the scene classification dataset show the effectiveness of the proposed method. Furthermore, the method achieves better robust accuracy compared with state-of-the-art adversarial defense methods in image classification tasks.

A Novel Variable Selection Method Based on Binning-Normalized Mutual Information for Multivariate Calibration

Variable (wavelength) selection is essential in the multivariate analysis of near-infrared spectra to improve model performance and provide a more straightforward interpretation. This paper proposed a new variable selection method named binning-normalized mutual information (B-NMI) based on information entropy theory. "Data binning" was applied to reduce the effects of minor measurement errors and increase the features of near-infrared spectra. "Normalized mutual information" was employed to calculate the correlation between each wavelength and the reference values. The performance of B-NMI was evaluated by two experimental datasets (ideal ternary solvent mixture dataset, fluidized bed granulation dataset) and two public datasets (gasoline octane dataset, corn protein dataset). Compared with classic methods of backward and interval PLS (BIPLS), variable importance projection (VIP), correlation coefficient (CC), uninformative variables elimination (UVE), and competitive adaptive reweighted sampling (CARS), B-NMI not only selected the most featured wavelengths from the spectra of complex real-world samples but also improved the stability and robustness of variable selection results.

Altered global modular organization of intrinsic functional connectivity in autism arises from atypical node-level processing

Autism spectrum disorder (ASD) is a neurodevelopmental condition characterized by restricted interests and repetitive behaviors as well as social-communication deficits. These traits are associated with atypicality of functional brain networks. Modular organization in the brain plays a crucial role in network stability and adaptability for neurodevelopment. Previous neuroimaging research demonstrates discrepancies in studies of functional brain modular organization in ASD. These discrepancies result from the examination of mixed age groups. Furthermore, recent findings suggest that while much attention has been given to deriving atlases and measuring the connections between nodes, within node information may also be crucial in determining altered modular organization in ASD compared with typical development (TD). However, altered modular organization originating from systematic nodal changes are yet to be explored in younger children with ASD. Here, we used graph-theoretical measures to fill this knowledge gap. To this end, we utilized multicenter resting-state fMRI data collected from 5 to 10-year-old children-34 ASD and 40 TD obtained from the Autism Brain Image Data Exchange (ABIDE) I and II. We demonstrate that alterations in topological roles and modular cohesiveness are the two key properties of brain regions anchored in default mode, sensorimotor, and salience networks, and primarily relate to social and sensory deficits in children with ASD. These results demonstrate that atypical global network organization in children with ASD arises from nodal role changes, and contribute to the growing body of literature suggesting that there is interesting information within nodes providing critical markers of functional brain networks in autistic children.

Different Modular Organization Between Early Onset and Late Onset Depression: A Study Base on Granger Causality Analysis

Background: Modular organization reflects the activity patterns of our brain. Different disease states may lead to different activity patterns and clinical features. Early onset depression (EOD) and late onset depression (LOD) share the same clinical symptoms, but have different treatment strategies and prognosis. Thus, explored the modular organization of EOD and LOD might help us understand their pathogenesis.

Method: The study included 82 patients with late life depression (EOD 40, LOD 42) and 90 healthy controls. We evaluated the memory, executive function and processing speed and performed resting-stage functional MRI for all participants. We constructed a functional network based on Granger causality analysis and carried out modularity, normalized mutual information (NMI), Phi coefficient, within module degree z-score, and participation coefficient analyses for all the participants.

Result: The Granger function network analysis suggested that the functional modularity was different among the three groups (P_auc = 0.0300), and NMI analysis confirmed that the partition of EOD was different from that of LOD (P_auc = 0.0190). Rh.10d.ROI (polar frontal cortex) and Rh.IPS1.ROI (dorsal stream visual cortex) were shown to be the potential specific nodes in the modular assignment according to the Phi coefficient (P = 0.0002, P_fdr = 0.0744 & P = 0.0004, P_fdr = 0.0744).

Conclusion: This study reveal that the functional modularity and partition were different between EOD and LOD in Granger function network. These findings support the hypothesis that different pathological changes might exist in EOD and LOD.

Discriminant Analysis of Lung Cancer Using Nonlinear Clustering of Copy Numbers

Background: Patient survival is not optimal for non-small cell lung cancer (NSCLC) patients, recurrence rate is high, and hence, early detection is crucial to increase the patient's survival. Gene-cancer mapping intends to discover associated genes with cancers and due to advances in high-throughput genotyping, screening for disease loci on a genome-wide scale is now possible. DNA copy numbers can potentially be used to identify cancer from normal cells in early detection of cancer.Methods: We use a nonlinear clustering method, so-called kernel K-means to separate cancer from normal samples. Kernel K-means is applied to the copy numbers obtained for each chromosome to cluster 63 paired cancer-blood samples (total of 126 samples) into two groups. Clustering performance is evaluated using true and false-positive rates, true and false-negative rates, and a nonlinear criterion, normalized mutual information (NMI).Results: Copy numbers of paired cancer-blood samples for 63 NSCLC patients are used in this study. Kernel K-means was applied to cluster 126 samples in two groups using copy numbers on each chromosome separately. The clustering results for 22 chromosomes are evaluated and discriminant power of them in identifying cancer is computed. We identified the top five and bottom five chromosomes based on their discriminant power.Conclusions: The results reveal high discriminant power of chromosomes 8, 5, 1, 3, and 19 for identifying cancer with the highest sensitivity of 75% yielded by chromosome 5. Bottom 5 chromosomes 9, 6, 4, 13, and 21 show low discriminant power with the accuracy of below 54% where true cancer and normal samples are grouped into substantially overlapping groups using copy numbers. This indicates the similarities of copy numbers obtained for cancer and normal samples on these chromosomes.

An Information Theory Approach for the Analysis of Individual and Combined Evaluation Parameters of Multiple Age-Related Diseases

In view of the frequent presence of several aging-related diseases in geriatric patients, there is a need to develop analytical methodologies that would be able to perform diagnostic evaluation of several diseases at once by individual or combined evaluation parameters and select the most informative parameters or parameter combinations. So far there have been no established formal methods to enable such capabilities. We develop a new formal method for the evaluation of multiple age-related diseases by calculating the informative values (normalized mutual information) of particular parameters or parameter combinations on particular diseases, and then combine the ranks of informative values to provide an overall estimation (or correlation) on several diseases at once. Using this methodology, we evaluate a geriatric cohort, with several common age-related diseases, including cognitive and physical impairments (dementia, chronic obstructive pulmonary disease-COPD and ischemic heart disease), utilizing a set of evaluation parameters (such as demographic data and blood biomarkers) routinely available in geriatric clinical practice. This method permitted us to establish the most informative parameters and parameter combinations for several diseases at once. Combinations of evaluation parameters were shown to be more informative than individual parameters. This method, with additional clinical data, may help establish the most informative parameters and parameter combinations for the diagnostic evaluation of multiple age-related diseases and enhance specific assessment for older multi-morbid patients and treatments against old-age multimorbidity.

Nonrigid registration of medical image based on adaptive local structure tensor and normalized mutual information

Nonrigid registration of medical images is especially critical in clinical treatment. Mutual information is a popular similarity measure for medical image registration; however, only the intensity statistical characteristics of the global consistency of image are considered in MI, and the spatial information is ignored. In this paper, a novel intensity-based similarity measure combining normalized mutual information with spatial information for nonrigid medical image registration is proposed. The different parameters of Gaussian filtering are defined according to the regional variance, the adaptive Gaussian filtering is introduced into the local structure tensor. Then, the obtained adaptive local structure tensor is used to extract the spatial information and define the weighting function. Finally, normalized mutual information is distributed to each pixel, and the discrete normalized mutual information is multiplied with a weighting term to obtain a new measure. The novel measure fully considers the spatial information of the image neighborhood, gives the location of the strong spatial information a larger weight, and the registration of the strong gradient regions has a priority over the small gradient regions. The simulated brain image with single-modality and multimodality are used for registration validation experiments. The results show that the new similarity measure improves the registration accuracy and robustness compared with the classical registration algorithm, reduces the risk of falling into local extremes during the registration process.

[Evaluation of the Image Registration Program for Portal Images Using CR and DRR Images in Radiation Therapy]

In this study, computer simulations and experiments were used to verify the accuracy of a two-dimensional image registration program (program) for portal images that we previously developed. The program used a computed radiography cassette system and digitally reconstructed radiography images as planning images for external beam radiation therapy. Using this program, we also investigated the reason two-dimensional automatic image registration images experienced large misregistration in clinical practice using commercial image registration systems. Mutual information and normalized mutual information were used as the registration criteria. To investigate the influence of image background with or without a region of interest (ROI), results of image registrations were compared. Parameters of image registration were defined as translation in the horizontal and vertical directions (x and y, respectively) and rotation (θ) around the axis perpendicular to the x-y plane. There was no significant difference in image registration arising from the difference between mutual information and normalized mutual information. Image registration was improved with a ROI. Regardless of the registration criteria, errors in image registration with a ROI in the experimental study were ≤1.2 mm in directions x and y and ≤1.0 degree in rotation θ. We found that image registration required setting up as close to the planned position as possible.

Information theoretical analysis of aging as a risk factor for heart disease

We estimate the weight of various risk factors in heart disease, and the particular weight of age as a risk factor, individually and combined with other factors. To establish the weights we use the information theoretical measure of normalized mutual information that permits determining both individual and combined correlation of diagnostic parameters with the disease status. The present information theoretical methodology takes into account the non-linear correlations between the diagnostic parameters, as well as their non-linear changes with age. Thus it may be better suited to analyze complex biological aging systems than statistical measures that only estimate linear relations. We show that individual parameters, including age, often show little correlation with heart disease. Yet in combination, the correlation improves dramatically. For diagnostic parameters specific for heart disease the increase in the correlative capacity thanks to the combination of diagnostic parameters, is less pronounced than for the less specific parameters. Age shows the highest influence on the presence of disease among the non-specific parameters and the combination of age with other diagnostic parameters substantially improves the correlation with the disease status. Hence age is considered as a primary "metamarker" of aging-related heart disease, whose addition can improve diagnostic capabilities. In the future, this methodology may contribute to the development of a system of biomarkers for the assessment of biological/physiological age, its influence on disease status, and its modifications by therapeutic interventions.

Automatic detection method of hepatocellular carcinomas using the non-rigid registration method of multi-phase liver CT images

BACKGROUND

Multi-phase CT images are obtained sequentially after the injection of contrast agents so that there is a large amount of local deformation between images due to the respiratory and heart motion. Therefore, a non-rigid registration technique is required in order to establish the anatomical correspondence between the multi-phase CT images for liver CAD (computer-aided diagnosis).

OBJECTIVE

In this paper, we propose the automatic detection method of hepatocellular carcinomas using the non-rigid registration method of multi-phase CT images.

METHODS

Global movements between multi-phase CT images are aligned by rigid registration based on normalized mutual information. Local deformations between multi-phase CT images are modeled by non-rigid registration based on B-spline deformable model. After the registration of multi-phase CT images, hepatocellular carcinomas are automatically detected by analyzing the original and subtraction information of the registered multi-phase CT images.

RESULTS

We applied our method to twenty five multi-phase CT datasets. Experimental results showed that the multi-phase CT images were accurately aligned. All of the hepatocellular carcinomas including small size ones in our 25 subjects were accurately detected using our method.

CONCLUSION

We conclude that our method is useful for detecting hepatocellular carcinomas.

Accurate template-based correction of brain MRI intensity distortion with application to dementia and aging

This paper examines an alternative approach to separating magnetic resonance imaging (MRI) intensity inhomogeneity from underlying tissue-intensity structure using a direct template-based paradigm. This permits the explicit spatial modeling of subtle intensity variations present in normal anatomy which may confound common retrospective correction techniques using criteria derived from a global intensity model. A fine-scale entropy driven spatial normalisation procedure is employed to map intensity distorted MR images to a tissue reference template. This allows a direct estimation of the relative bias field between template and subject MR images, from the ratio of their low-pass filtered intensity values. A tissue template for an aging individual is constructed and used to correct distortion in a set of data acquired as part of a study on dementia. A careful validation based on manual segmentation and correction of nine datasets with a range of anatomies and distortion levels is carried out. This reveals a consistent improvement in the removal of global intensity variation in terms of the agreement with a global manual bias estimate, and in the reduction in the coefficient of intensity variation in manually delineated regions of white matter.