We present a deep convolutional neural network for breast cancer screening exam classification, trained and evaluated on over 200,000 exams (over 1,000,000 images). Our network achieves an AUC of 0.895 in predicting whether there is a cancer in the breast, when tested on the screening population. We attribute the high accuracy of our model to a two-stage training procedure, which allows us to use a very high-capacity patch-level network to learn from pixel-level labels alongside a network learning from macroscopic breast-level labels. To validate our model, we conducted a reader study with 14 readers, each reading 720 screening mammogram exams, and find our model to be as accurate as experienced radiologists when presented with the same data. Finally, we show that a hybrid model, averaging probability of malignancy predicted by a radiologist with a prediction of our neural network, is more accurate than either of the two separately. To better understand our results, we conduct a thorough analysis of our network's performance on different subpopulations of the screening population, model design, training procedure, errors, and properties of its internal representations.deep learning | deep convolutional neural networks | breast cancer screening | mammography B reast cancer is the second leading cancer-related cause of death among women in the US. In 2014, over 39 million screening and diagnostic mammography exams were performed in the US. It is estimated that in 2015 232,000 women were diagnosed with breast cancer and approximately 40,000 died from it (1). Although mammography is the only imaging test that has reduced breast cancer mortality (2-4), there has been discussion regarding the potential harms of screening, including false positive recalls and associated false positive biopsies. The vast majority of the 10-15% of women asked to return following an inconclusive screening mammogram undergo another mammogram and/or ultrasound for clarification. After the additional imaging exams, many of these findings are determined as benign and only 10-20% are recommended to undergo a needle biopsy for further work-up. Among these, only 20-40% yield a diagnosis of cancer (5). Evidently, there is an unmet need to shift the balance of routine breast cancer screening towards more benefit and less harm.Traditional computer-aided detection (CAD) in mammography is routinely used by radiologists to assist with image interpretation, despite multicenter studies showing these CAD programs do not improve their diagnostic performance (6).These CAD programs typically use handcrafted features to mark sites on a mammogram that appear distinct from normal tissue structures. The radiologist decides whether to recall these findings, determining clinical significance and actionability. Recent developments in deep learning (7)-in particular, deep convolutional neural networks (CNNs) (8-12)-open possibilities for creating a new generation of CAD-like tools.This paper makes several contributions. Primarily, we train and evaluate a set of stro...
Lithium is used in the clinical treatment of bipolar disorder, a disease where patients suffer mood swings between mania and depression. Although the mode of action of lithium remains elusive, a putative primary target is thought to be inositol monophosphatase (IMPase) activity. Two IMPase genes have been identified in mammals, the well characterized myo-inositol monophosphatase 1 (IMPA1) and myo-inositol monophosphatase 2 (IMPA2). Several lines of genetic evidence have implicated IMPA2 in the pathogenesis of not only bipolar disorder but also schizophrenia and febrile seizures. However, little is known about the protein, although it is predicted to have lithium-inhibitable IMPase activity based on its homology to IMPA1. Here we present the first biochemical study comparing the enzyme activity of IMPA2 to that of IMPA1. We demonstrate that in vivo, IMPA2 forms homodimers but no heterodimers with IMPA1. Recombinant IMPA2 exhibits IMPase activity, although maximal activity requires higher concentrations of magnesium and a higher pH. IMPA2 shows significantly lower activity toward myo-inositol monophosphate than IMPA1. We therefore screened for additional substrates that could be more efficiently dephosphorylated by IMPA2, but failed to find any. Importantly, when using myo-inositol monophosphate as a substrate, the IMPase activity of IMPA2 was inhibited at high lithium and restricted magnesium concentrations. This kinetics distinguishes it from IMPA1. We also observed a characteristic pattern of differential expression between IMPA1 and IMPA2 in a selection of tissues including the brain, small intestine, and kidney. These data suggest that IMPA2 has a separate function in vivo from that of IMPA1.Inositol monophosphatase (IMPase 2 ; EC 3.1.3.25) is an enzyme that dephosphorylates myo-inositol monophosphate to generate free myo-inositol. This enzymatic pathway is important in cellular functions because myo-inositol is a precursor of the membrane phospholipid, phosphatidylinositol (PI). PI and its phosphorylated derivatives (phosphatidylinositol phosphate; PIP) play crucial roles in intracellular signal transduction via the production of second messengers, myoinositol 1,4,5-trisphosphate, and diacylglycerol. Inositol 1,4,5-trisphosphate triggers release of Ca 2ϩ from intracellular stores and undergoes a multi-step dephosphorylation by multiple enzymes including IMPase to generate free myo-inositol. Cells can generate myo-inositol by another biochemical pathway in which glucose 6-phosphate is isomerized by myo-inositol 1-phosphate synthase to produce myo-inositol 1-phosphate (1) and then dephosphorylated by IMPase leading to free myo-inositol. The dephosphorylation of myo-inositol monophosphate by IMPase is a critical and rate-limiting step for the regeneration of PI.From a clinical point of view, IMPase has attracted much interest (1-4). Bipolar disorder (also known as manic depressive illness) is characterized by chronically recurring episodes of fluctuating moods between mania and depression. Lithium has been used...
Special delivery: Transporters constructed on a sorbitol scaffold with eight guanidine residues show significant translocation across the cell membrane and the blood–brain barrier and high affinities toward mitochondria in HeLa and KG1a leukemia cells. The picture shows colocalisation of one such transporter (T) with MitoTracker red in KG1a cells (scale bars: 10 μm).
A facile synthesis of a combinatorial ceramide library and their activities in the NF-kappaB pathway and in apoptosis induction/prevention were demonstrated. A novel NF-kappaB activating molecule was discovered among ceramide containing beta-galactose, and the structural requirements of ceramides for apoptosis induction was elucidated.
Intraoperative evaluation of skin flap viability has primarily been dependent on clinical judgment. The purpose of this study was to determine whether an orthogonal polarization spectral imaging device could be used to accurately predict viability of random-pattern skin flaps. Orthogonal polarization spectral imaging is a newly developed technique that visualizes the microcirculation using reflected light without the use of fluorescent dyes and allows for noninvasive real-time observation of functional microvascular networks. In Sprague-Dawley rats (n = 24), three types of random skin flaps were designed with unknown zones of viability (n = 8 per group). After flap elevation, the skin flaps were evaluated by both clinical examination and orthogonal polarization spectral imaging. Areas of the flap determined to be nonviable by clinical examination were measured and marked. Orthogonal polarization spectral imaging was subsequently performed, and areas of the skin flap with stasis (i.e., cessation of red blood cell movement) in the dermal microcirculation on orthogonal polarization spectral imaging were measured and marked. The skin flaps were then secured in place. Flaps were evaluated on a daily basis for clinical signs of ischemia and necrosis. On postoperative day 7, the total amount of random skin flap necrosis was measured and recorded. Clinical examination of the random skin flaps significantly underestimated the actual amount of eventual flap necrosis, and as result was a very poor predictor of flap necrosis. By contrast, assessment of microcirculatory stasis using the orthogonal polarization spectral imaging device correlated well with the subsequent development of necrosis in all groups. In the three groups, the average amount of flap necrosis predicted by clinical examination deviated from actual necrosis by approximately 2 to 4 cm. However, the amount that orthogonal polarization spectral imaging differed from actual necrosis was 0.1 to 0.3 cm. Therefore, orthogonal polarization spectral imaging was an excellent predictor of eventual flap necrosis and much more accurate than clinical observation (p < 0.001). Intraoperative evaluation of axial and random pattern flap viability has traditionally been based on clinical examination as no other reliable, convenient test currently exists. The authors demonstrated that an orthogonal polarization spectral imaging device accurately predicts zones of necrosis in random pattern flaps by directly visualizing cessation of microcirculatory flow. Intraoperative stasis in the dermal microcirculation correlated precisely with subsequent flap necrosis. Orthogonal polarization spectral imaging was significantly more accurate than clinical examination, which consistently underestimated flap necrosis. The orthogonal polarization spectral imaging technique may have value in the intraoperative assessment of skin flap perfusion such as that required after skin-sparing mastectomy.
Spezialzustellung: Transporter, die auf einem Sorbitolgerüst mit acht Guanidinresten aufgebaut wurden, zeigen eine signifikante Translokation über die Zellmembran und die Blut‐Hirn‐Schranke sowie hohe Affinitäten gegenüber Mitochondrien in HeLa‐Zellen und KG1a‐Leukämiezellen. In den Bildern ist die Colokalisierung eines solchen Transporters (T) mit MitoTracker (rot) in KG1a‐Zellen gezeigt (Maßstab: 10 μm).
e Misfolding of proteins containing abnormal expansions of polyglutamine (polyQ) repeats is associated with cytotoxicity in several neurodegenerative disorders, including Huntington's disease. Recently, the eukaryotic chaperonin TRiC hetero-oligomeric complex has been shown to play an important role in protecting cells against the accumulation of misfolded polyQ protein aggregates. It is essential to elucidate how TRiC function is regulated to better understand the pathological mechanism of polyQ aggregation. Here, we propose that vaccinia-related kinase 2 (VRK2) is a critical enzyme that negatively regulates TRiC. In mammalian cells, overexpression of wild-type VRK2 decreased endogenous TRiC protein levels by promoting TRiC ubiquitination, but a VRK2 kinase-dead mutant did not. Interestingly, VRK2-mediated downregulation of TRiC increased aggregate formation of a polyQ-expanded huntingtin fragment. This effect was ameliorated by rescue of TRiC protein levels. Notably, small interference RNA-mediated knockdown of VRK2 enhanced TRiC protein stability and decreased polyQ aggregation. The VRK2-mediated reduction of TRiC protein levels was subsequent to the recruitment of COP1 E3 ligase. Among the members of the COP1 E3 ligase complex, VRK2 interacted with RBX1 and increased E3 ligase activity on TRiC in vitro. Taken together, these results demonstrate that VRK2 is crucial to regulate the ubiquitination-proteosomal degradation of TRiC, which controls folding of polyglutamine proteins involved in Huntington's disease.
We have established a new system for studying phytic acid, myo-inositol hexakisphosphate (InsP 6 ) synthesis in suspensioncultured cells of Catharanthus. InsP 6 and other intermediates of myo-inositol (Ins) phosphate metabolism were measured using an ion chromatography method. The detection limit for InsP 6 was less than 50 nM, which was sufficient to analyze Ins phosphates in living cells. Synthesis of Ins phosphates was induced by incubation in high inorganic phosphate medium. InsP 6 was mainly accumulated in vacuoles and was enhanced when cells were grown in high concentration of inorganic phosphates with the cations K 1
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.