Convolutional neural network for automated mass segmentation in mammography.

BACKGROUND: Automatic segmentation and localization of lesions in mammogram (MG) images are challenging even with employing advanced methods such as deep learning (DL) methods. We developed a new model based on the architecture of the semantic segmentation U-Net model to precisely segment mass lesions in MG images. The proposed end-to-end convolutional neural network (CNN) based model extracts contextual information by combining low-level and high-level features. We trained the proposed model using huge publicly available databases, (CBIS-DDSM, BCDR-01, and INbreast), and a private database from the University of Connecticut Health Center (UCHC). RESULTS: We compared the performance of the proposed model with those of the state-of-the-art DL models including the fully convolutional network (FCN), SegNet, Dilated-Net, original U-Net, and Faster R-CNN models and the conventional region growing (RG) method. The proposed Vanilla U-Net model outperforms the Faster R-CNN model significantly in terms of the runtime and the Intersection over Union metric (IOU). Training with digitized film-based and fully digitized MG images, the proposed Vanilla U-Net model achieves a mean test accuracy of 92.6%. The proposed model achieves a mean Dice coefficient index (DI) of 0.951 and a mean IOU of 0.909 that show how close the output segments are to the corresponding lesions in the ground truth maps. Data augmentation has been very effective in our experiments resulting in an increase in the mean DI and the mean IOU from 0.922 to 0.951 and 0.856 to 0.909, respectively. CONCLUSIONS: The proposed Vanilla U-Net based model can be used for precise segmentation of masses in MG images. This is because the segmentation process incorporates more multi-scale spatial context, and captures more local and global context to predict a precise pixel-wise segmentation map of an input full MG image. These detected maps can help radiologists in differentiating benign and malignant lesions depend on the lesion shapes. We show that using transfer learning, introducing augmentation, and modifying the architecture of the original model results in better performance in terms of the mean accuracy, the mean DI, and the mean IOU in detecting mass lesion compared to the other DL and the conventional models.

Deep convolutional neural networks for mammography: advances, challenges and applications.

BACKGROUND: The limitations of traditional computer-aided detection (CAD) systems for mammography, the extreme importance of early detection of breast cancer and the high impact of the false diagnosis of patients drive researchers to investigate deep learning (DL) methods for mammograms (MGs). Recent breakthroughs in DL, in particular, convolutional neural networks (CNNs) have achieved remarkable advances in the medical fields. Specifically, CNNs are used in mammography for lesion localization and detection, risk assessment, image retrieval, and classification tasks. CNNs also help radiologists providing more accurate diagnosis by delivering precise quantitative analysis of suspicious lesions. RESULTS: In this survey, we conducted a detailed review of the strengths, limitations, and performance of the most recent CNNs applications in analyzing MG images. It summarizes 83 research studies for applying CNNs on various tasks in mammography. It focuses on finding the best practices used in these research studies to improve the diagnosis accuracy. This survey also provides a deep insight into the architecture of CNNs used for various tasks. Furthermore, it describes the most common publicly available MG repositories and highlights their main features and strengths. CONCLUSIONS: The mammography research community can utilize this survey as a basis for their current and future studies. The given comparison among common publicly available MG repositories guides the community to select the most appropriate database for their application(s). Moreover, this survey lists the best practices that improve the performance of CNNs including the pre-processing of images and the use of multi-view images. In addition, other listed techniques like transfer learning (TL), data augmentation, batch normalization, and dropout are appealing solutions to reduce overfitting and increase the generalization of the CNN models. Finally, this survey identifies the research challenges and directions that require further investigations by the community.

A molecular structure matching approach to efficient identification of endogenous mammalian biochemical structures.

Metabolomics is the study of small molecules, called metabolites, of a cell, tissue or organism. It is of particular interest as endogenous metabolites represent the phenotype resulting from gene expression. A major challenge in metabolomics research is the structural identification of unknown biochemical compounds in complex biofluids. In this paper we present an efficient cheminformatics tool, BioSMXpress that uses known endogenous mammalian biochemicals and graph matching methods to identify endogenous mammalian biochemical structures in chemical structure space. The results of a comprehensive set of empirical experiments suggest that BioSMXpress identifies endogenous mammalian biochemical structures with high accuracy. BioSMXpress is 8 times faster than our previous work BioSM without compromising the accuracy of the predictions made. BioSMXpress is freely available at http://engr.uconn.edu/~rajasek/BioSMXpress.zip.

Synthesis, spectral and quantum chemical studies on NO-chelating sulfamonomethoxine-cyclophosph(V)azane and its Er(III) complex.

Computational studies have been carried out at the DFT-B3LYP/6-31G(d) level of theory on the structural and spectroscopic properties of novel ethane-1,2-diol-dichlorocyclophosph(V)azane of sulfamonomethoxine (L), and its binuclear Er(III) complex. Different tautomers of the ligand were optimized at the ab initio DFT level. Keto-form structure is about 15.8 kcal/mol more stable than the enol form (taking zpe correction into account). Simulated IR frequencies were scaled and compared with that experimentally measured. TD-DFT method was used to compute the UV-VIS spectra which show good agreement with measured electronic spectra. The structures of the novel isolated products are proposed based on elemental analyses, IR, UV-VIS, (1)H NMR, (31)P NMR, SEM, XRD spectra, effective magnetic susceptibility measurements and thermogravimetric analysis (TGA).

Integrating Clonal Selection and Deterministic Sampling for Efficient Associative Classification.

Traditional Associative Classification (AC) algorithms typically search for all possible association rules to find a representative subset of those rules. Since the search space of such rules may grow exponentially as the support threshold decreases, the rules discovery process can be computationally expensive. One effective way to tackle this problem is to directly find a set of high-stakes association rules that potentially builds a highly accurate classifier. This paper introduces AC-CS, an AC algorithm that integrates the clonal selection of the immune system along with deterministic data sampling. Upon picking a representative sample of the original data, it proceeds in an evolutionary fashion to populate only rules that are likely to yield good classification accuracy. Empirical results on several real datasets show that the approach generates dramatically less rules than traditional AC algorithms. In addition, the proposed approach is significantly more efficient than traditional AC algorithms while achieving a competitive accuracy.

Disposable screen-printed sensors for determination of duloxetine hydrochloride.

A screen-printed disposable electrode system for the determination of duloxetine hydrochloride (DL) was developed using screen-printing technology. Homemade printing has been characterized and optimized on the basis of effects of the modifier and plasticizers. The fabricated bi-electrode potentiometric strip containing both working and reference electrodes was used as duloxetine hydrochloride sensor. The proposed sensors worked satisfactorily in the concentration range from 1.0 × 10-6-1.0 × 10-2 mol L-1 with detection limit reaching 5.0 × 10-7 mol L-1 and adequate shelf life of 6 months. The method is accurate, precise and economical. The proposed method has been applied successfully for the analysis of the drug in pure and in its dosage forms. In this method, there is no interference from any common pharmaceutical additives and diluents. Results of the analysis were validated statistically by recovery studies.

The use of graph matching algorithms to identify biochemical substructures in synthetic chemical compounds: Application to metabolomics.

Metabolomics is a rapidly growing field studying the small-molecule metabolite profile of a biological organism. Studying metabolism has a potential to contribute to biomedical research as well as drug discovery. One of the current challenges in metabolomics is the identification of unknown metabolites as existing chemical databases are incomplete. We present a novel way of utilizing known mammalian metabolites in an effort to identify unknown ones. The system relies on a mammalian scaffolds database to aid the classification process. The results show that 96% of the mammalian compounds were identified as truly mammalian in a leave-one-out experiment. The system was also tested with a random set of synthetic compounds, downloaded from ChemBridge and ChemSynthesis databases. The system was able to eliminate 54% of the set, leaving 46% of the compounds as potentially unknown mammalian metabolites.

Improved algorithms for parsing ESLTAGs: a grammatical model suitable for RNA pseudoknots.

Formal grammars have been employed in biology to solve various important problems. In particular, grammars have been used to model and predict RNA structures. Two such grammars are Simple Linear Tree Adjoining Grammars (SLTAGs) and Extended SLTAGs (ESLTAGs). Performances of techniques that employ grammatical formalisms critically depend on the efficiency of the underlying parsing algorithms. In this paper, we present efficient algorithms for parsing SLTAGs and ESLTAGs. Our algorithm for SLTAGs parsing takes O(min{m,n4}) time and O(min{m,n4}) space, where m is the number of entries that will ever be made in the matrix M (that is normally used by TAG parsing algorithms). Our algorithm for ESLTAGs parsing takes O(min{m,n4}) time and O(min{m,n4}) space. We show that these algorithms perform better, in practice, than the algorithms of Uemura et al.

New dimeric cyclodiphosph(V)azane complexes of Cr(III), Co(II), Ni(II), Cu(II), and Zn(II): preparation, characterization and biological activity studies.

The complexes of type [CrLCl](2).3.5H(2)O and [ML(H(2)O)](2).nH(2)O in which M = Co(II); n = 1, Ni(II); n = 4, Cu(II); n = 0.5 and Zn(II); n = 2 ions and L is 1,3-dimethyl-2,4-dioxo-2',4'-bis(2iminothiophene)cyclodiphosph(V)azane, were prepared and their structures characterized by elemental analysis, IR, (1)H NMR, (31)P NMR, EPR, XRD, SEM, TGA, mass, molar conductance, magnetic moment and UV-Visible spectra. Antimicrobial activities have been studied using the agar-disc diffusion technique, and the higher antimicrobial activity has been observed for the Chromium(III) complex compared to the other metal complexes.