Spatial and geometric learning for classification of breast tumors from multi-center ultrasound images: a hybrid learning approach.

BACKGROUND: Breast cancer is the most common cancer among women, and ultrasound is a usual tool for early screening. Nowadays, deep learning technique is applied as an auxiliary tool to provide the predictive results for doctors to decide whether to make further examinations or treatments. This study aimed to develop a hybrid learning approach for breast ultrasound classification by extracting more potential features from local and multi-center ultrasound data. METHODS: We proposed a hybrid learning approach to classify the breast tumors into benign and malignant. Three multi-center datasets (BUSI, BUS, OASBUD) were used to pretrain a model by federated learning, then every dataset was fine-tuned at local. The proposed model consisted of a convolutional neural network (CNN) and a graph neural network (GNN), aiming to extract features from images at a spatial level and from graphs at a geometric level. The input images are small-sized and free from pixel-level labels, and the input graphs are generated automatically in an unsupervised manner, which saves the costs of labor and memory space. RESULTS: The classification AUCROC of our proposed method is 0.911, 0.871 and 0.767 for BUSI, BUS and OASBUD. The balanced accuracy is 87.6%, 85.2% and 61.4% respectively. The results show that our method outperforms conventional methods. CONCLUSIONS: Our hybrid approach can learn the inter-feature among multi-center data and the intra-feature of local data. It shows potential in aiding doctors for breast tumor classification in ultrasound at an early stage.

The interaction of global small vessel disease burden and Alzheimer's disease pathologies do not change the independent association of amyloid-beta with hippocampal volume: A longitudinal study on mild cognitive impairment subjects.

The purpose of this study was to investigate whether the co-existence of global small vessel disease (SVD) burdens and Alzheimer's disease (AD) pathologies change hippocampal volume (HV) and cognitive function of mild cognitive impairment (MCI) subjects. We obtained MRI images, cerebrospinal fluid biomarkers (Aß1-42 and p-tau), and neuropsychological tests of 310 MCI subjects from ADNI. The global SVD score was assessed. We used linear regression and linear mixing effect to analyze the effects of global SVD burdens, AD pathologies, and their interactions (SVD*AD) on baseline and longitudinal HV and cognition respectively. We used simple mediation effect to analyze the influencing pathways. After adjusting for global SVD and SVD*AD, Aß remained independently correlated with baseline and longitudinal HV (std ß = 0.294, p = .007; std ß = 0.292, p < .001), indicating that global SVD did not affect the correlation between Aß and HV. Global SVD score was correlated with longitudinal but not baseline HV (std ß = 0.470, p = .050), suggesting that global SVD may be more representative of long-term permanent impairment. Global SVD, AD pathologies, and SVD*AD were independently correlated with baseline and longitudinal cognitions, in which the association of Aß (B = 0.005, 95% CI: 0.005; 0.024) and p-tau (B = -0.002, 95% CI: -0.004; -0.000) with cognition were mediated by HV, suggesting that HV is more likely to explain the progression caused by AD pathology than SVD. The co-existence of global SVD and AD pathologies did not affect the individual association of Aß on HV; HV played a more important role in the influence of AD pathology on cognition than in SVD.

White Matter Tract Injury by MRI in CADASIL Patients is Associated With Iron Accumulation.

BACKGROUND: Widespread white matter (WM) injury is a hallmark feature of cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy (CADASIL). However, controversies about the mechanism of WM tract injury exist persistently. Excessive iron accumulation, frequently reported in CADASIL patients, might cause WM tract injury. PURPOSE: To test the association between iron accumulation and WM tract injury in CADASIL patients. STUDY TYPE: Retrospective. POPULATION: A total of 35 CADASIL patients (age = 50.4 ± 6.4, 62.9% female) and 48 healthy controls (age = 55.7 ± 8.0, 68.8% female). FIELD STRENGTH/SEQUENCE: Diffusion-weighted spin-echo echo-planar sequence; enhanced susceptibility-weighted angiography (ESWAN) gradient echo sequence on a 3 T scanner. ASSESSMENT: The phase images acquired by ESWAN were used to calculate quantitative susceptibility mapping (QSM). Iron accumulation was evaluated in deep gray matters using QSM. WM tract injury was quantified by diffusion metrics based on WM major tracts skeleton. We compared iron deposition between groups and analyzed the correlation between WM tract injury and iron deposition in regions showing significant differences from healthy controls. Exploratory analysis was carried out to investigate whether WM tract injury mediated the relationship between iron deposition and cognitive impairment evaluated by Mini-Mental State Examination (MMSE). STATISTICAL TESTS: General linear model (GLM), partial correlation, stepwise linear regression and mediation analysis were used. The threshold of statistical significance was set as p < 0.05. RESULTS: Compared with healthy controls, CADASIL patients had significantly increased iron deposition in the caudate and putamen. Aberrant iron deposition in these two regions was significantly associated with decreased WM fractional anisotropy (FA) (caudate, r = -0.373； putamen, r = - 0.421), and increased radial diffusivity (RD) (caudate, r = 0.372; putamen, r = 0.386). Furthermore, WM tract injury mediated the relationship between iron deposition and cognitive impairment. DATA CONCLUSION: Patients with CADASIL show increased iron deposition in the caudate and putamen that is correlated to WM tract injury, which may in turn mediate the association with cognitive impairment. EVIDENCE LEVEL: 2 TECHNICAL EFFICACY: Stage 2.

White Matter Free Water Outperforms Cerebral Small Vessel Disease Total Score in Predicting Cognitive Decline in Persons with Mild Cognitive Impairment.

BACKGROUND: Vascular pathology is an important partner of Alzheimer's disease (AD). Both total cerebral small vessel disease (CSVD) score and white matter free water (FW) are useful markers that could reflect cerebral vascular injury. OBJECTIVE: We aim to investigate the efficacy of these two metrics in predicting cognitive declines in patients with mild cognitive impairment (MCI). METHODS: We enrolled 126 MCI subjects with 3D T1-weighted images, fluid-attenuated inversion recovery images, T2* images, diffusion tensor imaging images, cerebrospinal fluid biomarkers and neuropsychological tests from the Alzheimer's Disease Neuroimaging Initiative database. The total CSVD score and FW values were calculated. Simple and multiple linear regression analyses were applied to explore the association between vascular and cognitive impairments. Linear mixed effect models were constructed to investigate the efficacy of total CSVD score and FW on predicting cognitive decline. RESULTS: FW was associated with baseline cognition and could predict the decline of executive and language functions in MCI subjects, while no association was found between total CSVD score and cognitive declines. CONCLUSION: FW is a promising imaging marker for investigating the effect of CSVD on AD progression.

Altered Spontaneous Brain Activity in Subjects With Different Cognitive States of Biologically Defined Alzheimer's Disease: A Surface-Based Functional Brain Imaging Study.

Background: Before the apparent cognitive decline, subjects on the course of Alzheimer's disease (AD) can have significantly altered spontaneous brain activity, which could be potentially used for early diagnosis. As previous studies investigating local brain activity may suffer from the problem of cortical signal aliasing during volume-based analysis, we aimed to investigate the cortical functional alterations in the AD continuum using a surface-based approach. Methods: Based on biomarker profile "A/T," we included 11 healthy controls (HC, A-T-), 22 preclinical AD (CU, A+T+), 33 prodromal AD (MCI, A+T+), and 20 AD with dementia (d-AD, A+T+) from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database. The amplitude of low-frequency fluctuation (ALFF) method was used to evaluate the changes of spontaneous brain activity, which was performed in the classic frequency band (0.01-0.08 Hz), slow-4 (0.027-0.073 Hz) band, and slow-5 (0.01-0.027 Hz) band. Results: Under classic frequency band and slow-4 band, analysis of covariance (ANCOVA) showed that there were significant differences of standardized ALFF (zALFF) in the left posterior cingulate cortex (PCC) among the four groups. The post-hoc analyses showed that under the classic frequency band, the AD group had significantly decreased zALFF compared with the other three groups, and the cognitively unimpaired (CU) group had decreased zALFF compared with the healthy control (HC) group. Under the slow-4 band, more group differences were detected (HC > CU/MCI > d-AD). The accuracy of classifying CU, mild cognitive impairment (MCI), and AD from HC by left PCC activity under the slow-4 band were 0.774, 0.744, and 0.920, respectively. Moreover, the zALFF values of the left PCC had significant correlations with cerebrospinal fluid (CSF) biomarkers and neuropsychological tests. Conclusions: Spontaneous brain activity in the left PCC may decrease in preclinical AD when cognitive functions were relatively normal. The combination of a surfaced-based approach and specific frequency band analysis may increase sensitivity for the identification of preclinical AD subjects.

Factors Associated With the Dilation of Perivascular Space in Healthy Elderly Subjects.

Background: The dilation of perivascular space (PVS) has been widely used to reflect brain degeneration in clinical brain imaging studies. However, PVS characteristics exhibit large differences in healthy subjects. Such variations need to be better addressed before PVS can be used to reflect pathological changes. In the present study, we aim to investigate the potential influence of several related factors on PVS dilation in healthy elderly subjects. Methods: One-hundred and three subjects (mean age = 59.5) were retrospectively included from a prospectively collected community cohort. Multi-modal high-resolution magnetic resonance imaging and cognitive assessments were performed on each subject. Machine-learning based segmentation methods were employed to quantify PVS volume and white matter hyperintensity (WMH) volume. Multiple regression analysis was performed to reveal the influence of demographic factors, vascular risk factors, intracranial volume (ICV), major brain artery diameters, and brain atrophy on PVS dilation. Results: Multiple regression analysis showed that age was positively associated with the basal ganglia (BG) (standardized beta = 0.227, p = 0.027) and deep white matter (standardized beta = 0.220, p = 0.029) PVS volume. Hypertension was positively associated with deep white matter PVS volume (standardized beta = 0.234, p = 0.017). Furthermore, we found that ICV was strongly associated with the deep white matter PVS volume (standardized beta = 0.354, p < 0.001) while the intracranial artery diameter was negatively associated with the deep white matter PVS volume (standardized beta = -0.213, p = 0.032). Conclusions: Intracranial volume has significant influence on deep white matter PVS volume. Future studies on PVS dilation should include ICV as an important covariate.

Author Correction: Plasma activity of Thioredoxin Reductase as a Novel Biomarker in Gastric Cancer.

An amendment to this paper has been published and can be accessed via a link at the top of the paper.

Dilated Perivascular Space in the Midbrain May Reflect Dopamine Neuronal Degeneration in Parkinson's Disease.

Background: The imbalance between the production and clearance of alpha-synuclein and its consequent accumulation plays a pivotal role in the pathogenesis of Parkinson's disease (PD). The diminished clearance of alpha-synuclein may be partly attributable to impaired interstitial fluid, which can be reflected by the extent of dilated perivascular space (dPVS). We studied the association between dPVS and dopamine neuronal degeneration. Method: We screened 71 healthy controls (HCs) and 88 patients from the Parkinson Progression Markers Initiative (PPMI) database. The dPVS was evaluated in different brain regions on axial T2-weighted images, and dopamine transporter (DAT) imaging data was used to elucidate the extent of dopaminergic neuronal degeneration. Patients with PD were further divided into two groups (SN + PD and SN - PD groups) according to whether dPVS was observed in the substantia nigra (SN). DAT uptake values and clinical scales were compared between the patients with PD and HCs and against dPVS scores. We also investigated the correlation between baseline dPVS scores and longitudinal DAT changes. Results: Relative to the HCs, patients with PD had more dPVS in the SN and basal ganglia (BG). PD patients with dPVS in the SN region exhibited greater expression of tau protein in cerebrospinal fluid (P = 0.038) and a trend towards decreased DAT binding (P = 0.086) relative to those without SN dPVS. No correlations were found between dPVS scores and DAT uptake values or between dPVS scores and longitudinal DAT changes. Conclusion: The dPVS in the SN of patients with PD may reflect the degeneration of dopaminergic neurons.

Plasma activity of Thioredoxin Reductase as a Novel Biomarker in Gastric Cancer.

Gastric cancer (GC) is one of the leading malignancies around the world. Identification of novel and efficient biomarkers for GC diagnosis and evaluation of therapeutic efficiency could improve the therapeutic strategy in future clinical application. This study aims to evaluate the levels of plasma thioredoxin reductase (TrxR) activity in GC patients to confirm its validity and efficacy in GC diagnosis and evaluation of therapeutic efficiency. 923 cases were enrolled in the current study. In the group of GC patients before clinical intervention, plasma TrxR activity [9.09 (7.96, 10.45) U/mL] was significantly higher than in healthy controls [3.69 (2.38, 5.32) U/mL]. The threshold of TrxR activity for GC diagnosis was set at 7.34 U/mL with a sensitivity of 85.5% and a specificity of 97.9%. In GC patients after chemotherapy, plasma TrxR activity was remarkably higher in patients with progressive disease or uncontrolled condition [10.07 (8.19, 11.02) U/mL] compared with patients with complete or partial response [7.12 (6.08, 8.37) U/mL] in response to chemotherapy. TrxR activity displayed the higher efficiency to distinguish between GC patients with two distinct clinical outcomes than carcinoembryonic antigen (CEA), cancer antigen 72-4 (CA72-4) and cancer antigen 19-9 (CA19-9). Moreover, combination of TrxR, CEA, CA72-4 and CA19-9 was demonstrated to be more effective in both GC diagnosis and evaluation of therapeutic efficiency than was each biomarker individually. Together, plasma TrxR activity was identified as a novel and efficient biomarker of GC, both in diagnosis and monitoring of therapeutic efficiency in response to chemotherapy.

Real options analysis for land use management: Methods, application, and implications for policy.

Discounted cash flow analysis, including net present value is an established way to value land use and management investments which accounts for the time-value of money. However, it provides a static view and assumes passive commitment to an investment strategy when real world land use and management investment decisions are characterised by uncertainty, irreversibility, change, and adaptation. Real options analysis has been proposed as a better valuation method under uncertainty and where the opportunity exists to delay investment decisions, pending more information. We briefly review the use of discounted cash flow methods in land use and management and discuss their benefits and limitations. We then provide an overview of real options analysis, describe the main analytical methods, and summarize its application to land use investment decisions. Real options analysis is largely underutilized in evaluating land use decisions, despite uncertainty in policy and economic drivers, the irreversibility and sunk costs involved. New simulation methods offer the potential for overcoming current technical challenges to implementation as demonstrated with a real options simulation model used to evaluate an agricultural land use decision in South Australia. We conclude that considering option values in future policy design will provide a more realistic assessment of landholder investment decision making and provide insights for improved policy performance.