SummaryBackgroundMenarche and menopause mark the onset and cessation, respectively, of ovarian activity associated with reproduction, and affect breast cancer risk. Our aim was to assess the strengths of their effects and determine whether they depend on characteristics of the tumours or the affected women.MethodsIndividual data from 117 epidemiological studies, including 118 964 women with invasive breast cancer and 306 091 without the disease, none of whom had used menopausal hormone therapy, were included in the analyses. We calculated adjusted relative risks (RRs) associated with menarche and menopause for breast cancer overall, and by tumour histology and by oestrogen receptor expression.FindingsBreast cancer risk increased by a factor of 1·050 (95% CI 1·044–1·057; p<0·0001) for every year younger at menarche, and independently by a smaller amount (1·029, 1·025–1·032; p<0·0001), for every year older at menopause. Premenopausal women had a greater risk of breast cancer than postmenopausal women of an identical age (RR at age 45–54 years 1·43, 1·33–1·52, p<0·001). All three of these associations were attenuated by increasing adiposity among postmenopausal women, but did not vary materially by women's year of birth, ethnic origin, childbearing history, smoking, alcohol consumption, or hormonal contraceptive use. All three associations were stronger for lobular than for ductal tumours (p<0·006 for each comparison). The effect of menopause in women of an identical age and trends by age at menopause were stronger for oestrogen receptor-positive disease than for oestrogen receptor-negative disease (p<0·01 for both comparisons).InterpretationThe effects of menarche and menopause on breast cancer risk might not be acting merely by lengthening women's total number of reproductive years. Endogenous ovarian hormones are more relevant for oestrogen receptor-positive disease than for oestrogen receptor-negative disease and for lobular than for ductal tumours.FundingCancer Research UK.

Purpose:To investigate whether the variable forms of putative iron deposition seen with susceptibility weighted imaging (SWI) will lead to a set of multiple sclerosis (MS) lesion characteristics different than that seen in conventional MR imaging. Materials and Methods:Twenty-seven clinically definite MS patients underwent brain scans using magnetic resonance imaging including: pre-and postcontrast T1-weighted imaging, T2-weighted imaging, FLAIR, and SWI at 1.5 T, 3 T, and 4 T. MS lesions were identified separately in each imaging sequence. Lesions identified in SWI were reevaluated for their iron content using the SWI filtered phase images. Results:There were a variety of new lesion characteristics identified by SWI, and these were classified into six types. A total of 75 lesions were seen only with conventional imaging, 143 only with SWI, and 204 by both. From the iron quantification measurements, a moderate linear correlation between signal intensity and iron content (phase) was established. Conclusion:The amount of iron deposition in the brain may serve as a surrogate biomarker for different MS lesion characteristics. SWI showed many lesions missed by conventional methods and six different lesion characteristics. SWI was particularly effective at recognizing the presence of iron in MS lesions and in the basal ganglia and pulvinar thalamus. MULTIPLE SCLEROSIS (MS) is an inflammatory demyelinating and neurodegenerative disease of the central nervous system (1,2). Most patients start with a relapsing-remitting course, which has a clearly defined episode of neurologic disability and recovery. The pathologic hallmark of multiple sclerosis is the demyelinated plaque, a well-demarcated hypocellular area characterized by the loss of myelin, along with axonal loss due to (3,4), and the formation of astrocytic scars. The etiologic mechanism underlying MS is generally believed to be autoimmune inflammation (5). Nevertheless, what initiates the disease and the sequence of events underlying the development of MS is not yet well established (6).Conventional magnetic resonance imaging (MRI) has been used routinely to diagnose and monitor the disease spatially and temporally. The use of conventional MRI to measure disease activity and assess effects of therapy is now standard in clinical practice and drug trials (7). T2-weighted imaging (T2WI) is highly sensitive in the detection of hyperintensities in white matter. However, hyperintensities on T2WI can correspond to a wide spectrum of pathology, ranging from edema and mild demyelination to lesions in which the neurons and supporting glial cells are replaced by glial scars or liquid necrosis (8 -14). In addition to T2WI, Gadolinium enhancement on T1-weighted imaging (T1WI) can suggest acute inflammation, which is a marker of disease It is becoming a consensus among many studies that iron is enriched within oligodendrocytes and myelin in both normal and diseased tissue (20 -23). One explanation for such findings proposes that iron is associated with the biosynthetic enzymes ...

No abstract

Racial bias is an important issue in biometric, but has not been thoroughly studied in deep face recognition. In this paper, we first contribute a dedicated dataset called Racial Faces in-the-Wild (RFW) database, on which we firmly validated the racial bias of four commercial APIs and four state-of-the-art (SOTA) algorithms. Then, we further present the solution using deep unsupervised domain adaptation and propose a deep information maximization adaptation network (IMAN) to alleviate this bias by using Caucasian as source domain and other races as target domains. This unsupervised method simultaneously aligns global distribution to decrease race gap at domain-level, and learns the discriminative target representations at cluster level. A novel mutual information loss is proposed to further enhance the discriminative ability of network output without label information. Extensive experiments on RFW, GBU, and IJB-A databases show that IMAN successfully learns features that generalize well across different races and across different databases.

Attention has become more attractive in person reidentification (ReID) as it is capable of biasing the allocation of available resources towards the most informative parts of an input signal. However, state-of-the-art works concentrate only on coarse or first-order attention design, e.g. spatial and channels attention, while rarely exploring higher-order attention mechanism. We take a step towards addressing this problem. In this paper, we first propose the High-Order Attention (HOA) module to model and utilize the complex and high-order statistics information in attention mechanism, so as to capture the subtle differences among pedestrians and to produce the discriminative attention proposals. Then, rethinking person ReID as a zero-shot learning problem, we propose the Mixed High-Order Attention Network (MHN) to further enhance the discrimination and richness of attention knowledge in an explicit manner. Extensive experiments have been conducted to validate the superiority of our MHN for person ReID over a wide variety of state-of-the-art methods on three large-scale datasets, including Market-1501, DukeMTMC-ReID and CUHK03-NP. Code is available at