Acquiring rule sets as a product of learning in a logical neuralarchitecture

Envisioning neural networks as systems that learn rules calls forth the verification issues already being studied in knowledge-based systems engineering, and complicates these with neural-network concepts such as nonlinear dynamics and distributed memories. We show that the issues can be clarified and the learned rules visualized symbolically by formalizing the semantics of rule-learning in the mathematical language of two-valued predicate logic. We further show that this can, at least in some cases, be done with a fairly simple logical model. We illustrate this with a combination of two example neural-network architectures, LAPART, designed to learn rules as logical inferences from binary data patterns, and the stack interval network, which converts real-valued data into binary patterns that preserve the semantics of the ordering of real values. We discuss the significance of the formal model in facilitating the analysis of the underlying logic of rule-learning and numerical data representation. We provide examples to illustrate the formal model, with the combined stack interval/LAPART networks extracting rules from numerical data.     

Large cell neuroendocrine carcinoma of the thymus

AIM: We highlight the occurrence of an unusual neuroendocrine tumour, a large cell neuroendocrine carcinoma, arising from the thymus. CASE DETAILS: A 68-year-old man with a history of cigarette smoking had a large mediastinal tumour arising from the thymus removed. Two years later the tumour recurred; it was debulked surgically but the patient died 2 months later: Histological examination of both tumour specimens revealed a tumour with an endocrine pattern, composed of large pleomorphic cells with large nuclei and prominent nucleoli. The mitotic count ranged from 19 to 26 per 10 high-power fields and large tracks of coagulative tumour necrosis were present. The tumour cells were strongly positive for neuron-specific enolase (NSE), chromogranin, CAM5.2 and AE1/3, with cytoplasmic dot-like accentuation for the latter three markers. The tumour fulfilled the criteria for a diagnosis of large cell neuroendocrine carcinoma. CONCLUSIONS: Large cell neuroendocrine carcinoma should be distinguished from atypical carcinoid and small cell carcinoma. It is a distinctive neuroendocrine malignancy with a prognosis between that of atypical carcinoid and small cell carcinoma, and needs to be treated aggressively.     

Protruded lumbar intervertebral nucleus pulposus in a 12-year-old girl who recovered after nonsurgical treatment: a follow-up case report

Restless legs syndrome affects millions, but most healthcare providers don't know it exists. Its sleep-deprived victims are plagued by indescribable crawling sensations in their legs that occur at night and that are only relieved by moving the legs. Many home healthcare clients, already challenged by their primary illness, are also affected by the persistent symptoms of Restless Legs Syndrome. It is important that home care nurses learn about Restless Legs Syndrome-what it is, how to assess it, and what can be done about it.     

Simplified 0.35-μm flash EEPROM process using high-temperatureoxide (HTO) deposited by LPCVD as interpoly dielectrics and peripheraltransistors gate oxide

A simplified flash EEPROM process was developed using high-temperature LPCVD oxide both as flash cells interpoly dielectrics and as peripheral transistors gate oxide (decoding logic). An O₂ anneal at 850°C lowers charge trapping and interface trap density induced by Fowler-Nordheim injection. However, electron trapping remains slightly higher than with dry thermal oxide. Similar memory charge loss and write-erase endurance are obtained as for ONO-insulated cells. HTO thus proves to have the required quality and reliability to be used in flash EEPROMs     

Computer simulation of the growth of heterostructure systems

We have investigated the growth of heterostructures with high lattice mismatch and the growth of quantum wires on the top of ridges and through shadowing masks. Simulations are performed within a Monte Carlo scheme using tetrahedral lattice structure of semiconductor materials. It is shown that results of different simulations present similarities that we attribute to the primary role of kinetic effects as the driving force during epitaxial growth. The formation of 3D islands showing (111) facets, or (111) side walls when depositing through a shadowing mask, has been observed. The facets are of better quality than the top (001) surface because of the higher mobility of atoms on these facets.     

Expression of functional MHC class II molecules by a mouse pro-B cell clone

OBJECTIVE: Pooled bronchoalveolar lavage fluid (BALF), the return of lavage, contains both bronchial and alveolar material which differ from each other. Artifacts may be created by filtering, centrifuging and washing cells before cytopreparation. This study presents reference values of healthy volunteers for the alveolar sample, ALF, cytopreparation being performed without filtration or centrifugation. METHODS: Eighteen healthy, non-smoking volunteers underwent a standard bronchoalveolar lavage using 10 aliquots of 20 ml of saline. Excluding the return of the first and second aliquots, the rest were pooled and examined cytologically, immunocytochemically and biochemically. The mean, standard deviation, and 95% confidence limits were calculated for the following variables: amount of return, estimated content of epithelial lining fluid (ELF), total and differential cell counts on filter and cytocentrifuge (CCF) preparations, computed cell counts per unit volume of ALF, distribution of lymphocyte subgroups CD3+CD2, CD4, CD8, CD19, CD25 and CD57, and the ratio of CD4 to CD8, the amounts of lymphocytes in the same subgroups per volume of ALF, and the concentrations of total protein, albumin, immunoglobulins A, G and M, hyaluronic acid, eosinophilic cationic protein (ECP), procollagen III aminoterminal propeptide (PCP) and beta 2-microglobulin in ALF and in ELF, as well as the ratios of the concentrations of the solutes in ALF to the same in serum. RESULTS: The 95% confidence limits of means for the most important variables were as follows: estimated ELF content 0.42-0.74%; total cells in ALF 76.6-143.0 x 10(6) l-1; distribution of inflammatory cells on filter and CCF slides: macrophages 74.9-83.6 and 81.4-90.1%, lymphocytes 13.1-22.5 and 8.1-16.4%, and neutrophils 1.0-4.1 and 0.7-2.7%, respectively; distribution of lymphocyte subsets: CD3+CD2 85.6-90.6%, CD4 44.3-53.1%, CD8 26.9-35.8%; concentration of solutes in ALF: total protein 44.8-61.3 mg l-1, albumin 15.4-22.2 mg l-1, IgA 1.8-3.4 mg l-1, IgG 3.1-6.1 mg l-1, IgM 0.05-0.26 mg l-1, hyaluronic acid 8.8-11.1 micrograms l-1, ECP 0.19-0.77 micrograms l-1, PCP 0.005-0.58 micrograms l-1, beta 2-microglobulin 62.2-81.5 micrograms l-1. CONCLUSIONS: Our results show that excluding the bronchial sample from ALF of volunteer subjects and omitting filtering and washing before cytopreparation produces cytologic, immunocytochemical and biochemical reference values with reasonable 95% confidence limits to be used in clinical settings.     

Selective catalytic reduction of NOx by hydrocarbons on Pt/Al2O3 catalysts at low temperatures without the formation of N2O

The selective reduction of nitrogen oxides by hydrocarbons on a Pt catalyst has been investigated. Some variations are observed between different reductants. However, the most significant result is the observation that no N₂O is observed as a by-product when toluene is used as a reductant.     

An auricular prosthesis attached by intra-mastoid implants. Apropos of 11 cases

The use of osseointegrated implants to provide support for auricle prosthesis as been carried out by Tjellstrom. This method has provided the maxillofacial surgeon with another approach to the treatment of defect of the auricle. The principles and concepts of extra-oral osseointegrated implants are the same as those of the intra-oral application developed by Branemark 26 years ago. This simple and reliable method allow the attachment to splint bar fixated to three or two mastoid osseointegrated implants. This method represent an interesting possibility for the management of auricle defects. We report on our experience of 11 cases.     

DNA analysis in hepatoblastoma by flow and image cytometry

BACKGROUND: In several types of tumors, including hepatocellular carcinoma, prognosis could be correlated with DNA ploidy. Few studies have been performed on hepatoblastoma with contradictory results. METHODS: Twenty-nine cases of nonpretreated hepatoblastoma were studied with flow cytometry and image cytometry for DNA index and proliferation index using paraffin-embedded tissue. RESULTS: Twenty-three (79.9%) tumors were diploid, and 6 (20.7%) were aneuploid (hyperdiploid). Patients with diploid tumors were younger than those with aneuploid tumors. With regard to stage, diploid tumors were almost equally distributed among stages (tumor, lymph node metastases, distant metastases), whereas aneuploid tumors tended to occur in higher stages (tumor, lymph node metastases, distant metastases). Diploid tumors had clearly a better prognosis than aneuploid tumors, although the difference was not statistically significant (flow cytometry, P = 0.06; image cytometry, P = 0.16). A more favorable prognosis was also noted for hepatoblastomas with low-proliferation index (< or = 7%), but the difference from tumors with high-proliferation index (> 7%) again was not statistically significant (P = 0.16). CONCLUSIONS: Although no statistically significant differences in prognosis between hepatoblastomas with diploid and aneuploid DNA content, respectively, were found, there is a clear tendency that diploid hepatoblastomas behave more favorably. The same is true for hepatoblastomas with low-proliferation index.