Texture-Guided Transfer Learning for Low-Quality Face Recognition.

Although many advanced works have achieved significant progress for face recognition with deep learning and large-scale face datasets, low-quality face recognition remains a challenging problem in real-word applications, especially for unconstrained surveillance scenes. We propose a texture-guided (TG) transfer learning approach under the knowledge distillation scheme to improve low-quality face recognition performance. Unlike existing methods in which distillation loss is built on forward propagation; e.g., the output logits and intermediate features, in this study, the backward propagation gradient texture is used. More specifically, the gradient texture of low-quality images is forced to be aligned to that of its high-quality counterpart to reduce the feature discrepancy between the high- and low-quality images. Moreover, attention is introduced to derive a soft-attention (SA) version of transfer learning, termed as SA-TG, to focus on informative regions. Experiments on the benchmark low-quality face DB's TinyFace and QMUL-SurFace confirmed the superiority of the proposed method, especially more than 6.6% Rank1 accuracy improvement is achieved on TinyFace.

Experimental stability analysis of neural networks in classification problems with confidence sets for persistence diagrams.

We investigate classification performance of neural networks (NNs) based on topological insight in an attempt to guarantee stability of their inference. NNs which can accurately classify a dataset map it into a hidden space while disentangling intertwined data. NNs sometimes acquire forcible mapping to disentangle the data, and this forcible mapping generates outliers. The mapping around the outliers is unstable because the outputs change drastically. Hence, we define stable NNs to mean that they do not generate outliers. To investigate the possibility of the existence of outliers, we use persistent homology and a method to estimate the confidence set for persistence diagrams. The combined use enables us to test whether the focused geometry is topologically simple, that is, no outliers. In this work, we use the MNIST and CIFAR-10 datasets and investigate the relationship between the classification performance and the topological characteristics with several NNs. Investigation results with the MNIST dataset show that the test accuracy of all the networks is superior, exceeding 98%, even though the transformed dataset is not topologically simple. Results with the CIFAR-10 dataset also show that the possibility of the existence of outliers is shown in the mapping by the accurate convolutional NNs. Therefore, we conclude that the presented investigation is necessary to guarantee that the NNs, in particular deep NNs, do not acquire unstable mapping for forcible classification.

Validation of Antiobesity Effects of Black Soybean Seed Coat Powder Suitable as a Food Material: Comparisons with Conventional Yellow Soybean Seed Coat Powder.

In this study, we fed obese model mice black soybean seed coat powder (BSCP) and evaluated the antiobesity effects. As a control, normal yellow soybean seed coat powder (YSCP) was used. C57BL/6J, a high-fat diet-induced obesity model mouse, was fed a high-fat diet containing BSCP or YSCP (20% fat) to induce obesity. The results showed that in the BSCP group, it caused significant suppression of body weight gain and suppression of white adipose tissue weight compared with the YSCP group. Moreover, it significantly decreased serum leptin levels, which correlated with visceral fat mass, and increased antidiabetic adipocytokine and adiponectin levels. Therefore, this suggests the pigmented components contained in BSCP have an antiobesity effect in obese model mice. It is suggested that this material, which can be prepared without extraction with an organic solvent and is suitable for use as a food material, could be a functional food material with a practicable antiobesity effect.

The Seed Coat Extract of Black Soybean Decreases Nicotine-Induced Vascular Fiber Degradation by Suppressing Matrix Metalloproteinase 2 Expression.

Abdominal aortic aneurysm (AAA) is a vascular disease characterized by weakening of vascular walls and progressive dilation of the abdominal aorta. Nicotine, the main component of tobacco, is reportedly associated with the development and rupture of AAA. It is desirable to attenuate the destructive effect of nicotine on vascular walls, using dietary food components. However, effective methods for preventing AAA progression using dietary food components remain unestablished. This study focuses on proanthocyanidins, well known for their potent antioxidant activity. We speculated that proanthocyanidins can suppress nicotine-induced weakening of vascular walls. To estimate the effect of black soybean seed coat extract (BSSCE), rich in proanthocyanidins, on nicotine-induced weakening of the aortic wall, mice were divided into four groups: the control diet and distilled water group (named C), BSSCE solution diet and distilled water group (named B), control diet and 0.5 mg/mL nicotine solution group (named CN), and BSSCE solution diet and 0.5 mg/mL nicotine solution group (named BN). Nicotine-induced degradation of elastin and collagen fibers were significantly suppressed in BN group. The positive areas for matrix metalloproteinase (MMP)-2 and oxidative stress in BN group were significantly decreased compared to those in CN group. These results suggest that proanthocyanidins-rich BSSCE can prevent the weakening of the aortic wall via inhibiting MMP-2 upregulation.

Reconstructing Face Image from the Thermal Infrared Spectrum to the Visible Spectrum.

During the night or in poorly lit areas, thermal cameras are a better choice instead of normal cameras for security surveillance because they do not rely on illumination. A thermal camera is able to detect a person within its view, but identification from only thermal information is not an easy task. The purpose of this paper is to reconstruct the face image of a person from the thermal spectrum to the visible spectrum. After the reconstruction, further image processing can be employed, including identification/recognition. Concretely, we propose a two-step thermal-to-visible-spectrum reconstruction method based on Canonical Correlation Analysis (CCA). The reconstruction is done by utilizing the relationship between images in both thermal infrared and visible spectra obtained by CCA. The whole image is processed in the first step while the second step processes patches in an image. Results show that the proposed method gives satisfying results with the two-step approach and outperforms comparative methods in both quality and recognition evaluations.

Elevated plasma B-type natriuretic peptide concentration and resistive index, but not decreased aortic distensibility, associate with impaired blood flow at popliteal artery in type 2 diabetic patients.

Blood flow in lower extremity arteries is frequently impaired in diabetic patients even though they have a normal ankle-brachial index (ABI 1.0-1.4). Risk factors contributing to this lower extremity arterial disease have not been fully elucidated. We enrolled 52 type 2 diabetic patients with normal ABI and 30 age-matched nondiabetic subjects consecutively admitted to our hospital. Plasma B-type natriuretic peptide (BNP) concentrations were measured. Distensibility in ascending thoracic and abdominal aortas as well as total flow volume and resistive index at popliteal artery were evaluated by gated magnetic resonance imaging. An automatic device was used to measure ABI and brachial-ankle pulse-wave velocity (baPWV). Diabetic patients showed lower distensibility in ascending thoracic aorta (p<0.001) and total flow volume (p<0.001) and higher baPWV (p<0.001) and resistive index (p=0.005) and similar BNP and distensibility in abdominal aorta compared to nondiabetic subjects. Simple linear regression analyses revealed that distensibility in ascending thoracic (p=0.019) and abdominal (p=0.030) aortas positively as well as baPWV (p=0.020), resistive index (p<0.001) and BNP (p<0.001) negatively correlated with total flow volume. Stepwise multiple regression analysis demonstrated that increased BNP and resistive index were independent risk factors for total flow volume in diabetic patients (r(2)=0.639, p<0.001). These results indicate that increased plasma BNP levels and peripheral vascular resistance, but not decreased aortic distensibility, associate with impaired blood flow in lower extremity arteries in diabetic patients.

Associations of plasma von Willebrand factor ristocetin cofactor activity and 5-hydroxyindole acetic acid concentrations with blood flow in lower-leg arteries in Japanese type 2 diabetic patients with normal ankle-brachial index.

AIMS: To evaluate the associations of circulating levels of proinflammatory molecules and endothelial factors with blood flow in lower-leg arteries in diabetic patients with normal ankle-brachial index (ABI>0.9). METHODS: We enrolled 123 type 2 diabetic patients with normal ABI and 30 age-matched nondiabetic subjects consecutively admitted to our hospital. Flow volume and resistive index, an index of peripheral vascular resistance, at the popliteal artery were evaluated using gated two-dimensional cine-mode phase-contrast magnetic resonance imaging. An automatic device was used to measure ABI and brachial-ankle pulse-wave velocity (baPWV) for evaluation of arterial stiffness. Plasma soluble intercellular adhesion molecule-1 (sICAM-1) and monocyte chemoattractant protein-1 (MCP-1) concentrations, serum high-sensitivity C-reactive protein (hsCRP) levels, plasma von Willebrand factor ristocetin cofactor activity (VWF), and plasma vasoconstrictor serotonin metabolite 5-hydroxyindole acetic acid (5-HIAA) concentrations were measured. RESULTS: Diabetic patients had higher baPWV (P<.0001), resistive index (P<.0001), sICAM-1 (P<.0001), MCP-1 (P=.0224), log hsCRP (P<.0001), VWF (P<.0001), 5-HIAA (P=.0015), and lower blood flow (P<.0001) than nondiabetic subjects. VWF (P=.0019) or 5-HIAA (P=.0011), but not sICAM-1, MCP-1, and log hsCRP, was negatively correlated with blood flow in diabetic patients. A multivariate analysis revealed that the significant independent determinants of blood flow were hypertension, use of renin-angiotensin system inhibitors, VWF and 5-HIAA (r(2)=0.198, P<.0001) in diabetic patients. CONCLUSIONS: Plasma VWF and 5-HIAA concentrations are associated with blood flow and are involved in the pathogenesis of impaired peripheral circulation due to higher arterial stiffness and greater vascular resistance in lower-leg arteries in diabetic patients with normal ABI.

Real-time marker-free patient registration for electromagnetic navigated bronchoscopy: a phantom study.

PURPOSE: To perform bronchoscopy safely and smoothly, it is very important to develop a bronchoscopic guidance system. Transbronchial lung biopsy (TBLB) with a bronchoscopic guidance system especially should permit safe image-guided procedures. Recently, electromagnetic tracking (EMT) is utilized to track the tip of the bronchoscope camera in real time. For most tracking methods using position sensors, registration between tracking data and previously acquired reference image data, such as CT image, is performed using natural landmarks of the patient or fiducial markers attached to the patient, whose positions need to be measured manually by the physician before the actual bronchoscopy. Therefore, this paper proposes a marker-free CT-to-patient registration method utilizing bronchoscope's position and orientation obtained by the EMT. METHODS: We developed a guidance system that is able to track the tip of the bronchoscope camera in real time. In the case of a guidance system that uses position sensors, natural landmarks of the patient or fiducial markers attached to the patient are needed to obtain the correspondence between EMT outputs and previously acquired reference image data, such as CT image. This paper proposes a registration method without landmarks or fiducials by estimating the transformation matrix between the patient and the CT image taken prior to the bronchoscopic examination. This estimation is performed by computing correspondences between the outputs of the EMT sensor and airways extracted from the CT image. As ambiguities between EMT measurements and their corresponding airway branches may arise at airway bifurcations, we introduce a stable airway branch selection mechanism for improving the robustness of the estimation of the transformation matrix. To evaluate the performance of the proposed method, we applied the method to a rubber bronchial phantom and added virtual breathing motion to the sensor output. RESULTS: Experimental results show that the accuracy of our proposed method is within 2.0-3.0 mm (without breathing motion) and 2.5-3.5 mm (with breathing motion). The proposed method could also track a bronchoscope camera in real time. CONCLUSIONS: We developed a method for CT-to-patient registration using a position sensor without fiducial markers and natural landmarks. Endoscopic guided biopsy of lung lesions is feasible using a marker-free CT-to-patient registration method.

Associations of coronary artery calcification and carotid intima-media thickness with plasma concentrations of vascular calcification inhibitors in type 2 diabetic patients.

Vascular calcification is frequently accompanied by intima-media thickening, but the associations among these atherosclerotic features and bone-related peptides in diabetic patients are unclear. We enrolled 168 type 2 diabetic patients and 40 non-diabetic subjects consecutively admitted to our hospital. Mean intima-media thickness (mean-IMT) in common carotid arteries was measured by B-mode ultrasonography. Agatston coronary artery calcium score (CACS) was obtained using multidetector-row computed tomography (MDCT). Plasma bone-related peptides osteopontin and osteoprotegerin levels were measured. Diabetic patients had higher mean-IMT (p=0.0002) and log(CACS+1) (p<0.0001) and similar bone-related peptides compared to non-diabetic subjects. In diabetic patients classified into tertiles according to their CACS levels, those with the highest scores showed the highest mean-IMT (p=0.0004) and bone-related peptides (p<0.05) among the groups. log(CACS+1) and mean-IMT were associated (p<0.0001) and were positively correlated with osteopontin (p<0.01) and osteoprotegerin (p<0.01) in diabetic patients. Multivariate analyses revealed that the significant independent determinants of log(CACS+1) were age, duration of diabetes and osteopontin (p<0.0001) and those of mean-IMT were age, hypertension, osteopontin and osteoprotegerin (p<0.0001), respectively. We have demonstrated that vascular calcification in type 2 diabetic patients is frequently accompanied by intima-media thickening, and osteopontin may act as a vascular calcification inhibitor by increasing intima-media thickness.

[Analysis of basic activity of electroencephalogram in patients with carbon monoxide intoxication for monitoring efficacy of treatment].

Electroencephalography (EEG) should be used in carbon monoxide intoxication on the basis of its clinical significance. The aim of this study was to analyze basic activity of EEG in acute carbon monoxide intoxication and clarify its significance for monitoring the efficacy of treatment. Thirty-six cases with acute carbon monoxide intoxication were treated with either hyperbaric or normobaric oxygenation. EEG was performed every 2 weeks for a total of three times in each case. Basic activity in the first EEG revealed slow waves (slow alpha, theta or delta) and normal waves (9-11 Hz) in 11, and 25 cases, respectively. When changes of the basic activity of EEG were analyzed, they were classified into 3 patterns: pattern A (faster over 2 Hz), B (stable) and C (slower over 2 Hz) in 7, 27 and 2 cases, respectively. Three cases with the pattern B turned to be an intermittent type of carbon monoxide intoxication. Two cases of pattern C were related with poor prognosis and death. In conclusion, the patterns in the basic activity change of EEG at the interval of 2 weeks in acute carbon monoxide intoxication are useful in monitoring efficacy of treatment.

Automated nomenclature of bronchial branches extracted from CT images and its application to biopsy path planning in virtual bronchoscopy.

We propose a novel anatomical labeling algorithm for bronchial branches extracted from CT images. This method utilizes multiple branching models for anatomical labeling. In the actual labeling process, the method selects the best candidate models at each branching point. Also a special labeling procedure is proposed for the right upper lobe. As an application of the automated nomenclature of bronchial branches, we utilized anatomical labeling results for assisting biopsy planning. When a user inputs a target point around suspicious regions on the display of a virtual bronchoscopy (VB) system, the path to the desired position is displayed as a sequence of anatomical names of branches. We applied the proposed method to 25 cases of CT images. The labeling accuracy was about 90%. Also the paths to desired positions were generated by using anatomical names in VB.

Two lysosomal membrane proteins, LGP85 and LGP107, are delivered to late endosomes/lysosomes through different intracellular routes after exiting from the trans-Golgi network.

Lysosomal membrane proteins are delivered from their synthesis site, the endoplasmic reticulum (ER) to late endosomes/lysosomes through the Golgi complex. It has been proposed that after leaving the Golgi they are transported either directly or indirectly (via the cell surface) to late endosomes/lysosomes. In the present study, we examined the transport routes taken by two structurally different lysosomal membrane proteins, LGP85 and LGP107, in rat 3Y1-B cells. Here we show that newly synthesized LGP85 and LGP107 are delivered to late endosomes/lysosomes via a direct route without passing through the cell surface. Interestingly, although LGP107 is delivered from the Golgi to early endosomes containing internalized horseradish peroxidase-conjugated transferrin (HRP-Tfn) en route to lysosomes, LGP85 does not pass through the HRP-Tfn-positive early endosomes. These results suggest, therefore, that LGP85 and LGP107 are sorted into distinct transport vesicles at the post-Golgi, presumably the trans-Golgi network (TGN), after which LGP85 is delivered directly to late endosomes/lysosomes, but significant fractions of LGP107 are targeted to early endosomes before transport to late endosomes/lysosomes. This study provides the first evidence that after exiting from the Golgi, LGP85 and LGP107 are targeted to late endosomes/lysosomes via a different pathway.

Eutopic overexpression of vasopressin v1a receptor in adrenocorticotropin-independent macronodular adrenal hyperplasia.

Arginine vasopressin (AVP) stimulates cortisol secretion through its vascular type V(1a) receptor in the adrenal glands, in addition to stimulating ACTH secretion through pituitary V(3) receptor. Because hyper-response of plasma cortisol to vasopressin is documented in some patients with Cushing's syndrome due to adrenal adenoma (CS) or ACTH-independent macronodular adrenocortical hyperplasia (AIMAH), we analyzed the expression of V(1a), V(2), V(3) receptor and AVP mRNA in human adrenal tissues by quantitative competitive RT-PCR or real-time PCRs. V(1a) receptor mRNA levels (ratio against glyceraldehyde 3-phosphate dehydrogenase) were 0.378 +/- 0.143 (mean +/- SE) in preclinical CS (n = 5) and 0.630 +/- 0.072 in AIMAH (n = 4), which were significantly higher than those (0.046 +/- 0.012; n = 9) in control adrenals, whereas those in overt CS (0.143 +/- 0.048; n = 10) or aldosterone-producing adenomas (0.069 +/- 0.018; n = 12) were similar to control adrenals. Although ectopic expression of V(2) or V(3) receptor was detected in half of AIMAH cases, the absolute levels were low. Furthermore, V(1a) receptor mRNA levels in the adjacent adrenal glands (0.190 +/- 0.039, n = 9) of aldosterone-producing adenomas were higher than those in control adrenals and in the corresponding tumor portions (0.079 +/- 0.024). In contrast, there were no significant differences in AVP mRNA levels among these groups. These results suggest that eutopic V(1a) receptor overexpression is involved in the etiology of AIMAH and a subset of adrenal adenomas causing overt or preclinical Cushing's syndrome. Our results imply a possible association of V(1a) receptor expression with adrenal hyperplasia.