Search

The application of machine learning techniques to histopathology images enables advances in the field, providing valuable tools that can speed up and facilitate the diagnosis process. The classification of these images is a relevant aid for physicians who have to process a large number of images in long and repetitive tasks. This work proposes the adoption of metric learning that, beyond the task of classifying images, can provide additional information able to support the decision of the classification system. In particular, triplet networks have been employed to create a representation in the embedding space that gathers together images of the same class while tending to separate images with different labels. The obtained representation shows an evident separation of the classes with the possibility of evaluating the similarity and the dissimilarity among input images according to distance criteria. The model has been tested on the BreakHis dataset, a reference and largely used dataset that collects breast cancer images with eight pathology labels and four magnification levels. Our proposed classification model achieves relevant performance on the patient level, with the advantage of providing interpretable information for the obtained results, which represent a specific feature missed by the all the recent methodologies proposed for the same purpose.

Recognition of diseases associated with mutations of the chaperone system genes, e.g., chaperonopathies, is on the rise. Hereditary and clinical aspects are established, but the impact of the mutation on the chaperone molecule and the mechanisms underpinning the tissue abnormalities are not. Here, histological features of skeletal muscle from a patient with a severe, early onset, distal motor neuropathy, carrying a mutation on the CCT5 subunit (MUT) were examined in comparison with normal muscle (CTR). The MUT muscle was considerably modified; atrophy of fibers and disruption of the tissue architecture were prominent, with many fibers in apoptosis. CCT5 was diversely present in the sarcolemma, cytoplasm, and nuclei in MUT and in CTR and was also in the extracellular space; it colocalized with CCT1. In MUT, the signal of myosin appeared slightly increased, and actin slightly decreased as compared with CTR. Desmin was considerably delocalized in MUT, appearing with abnormal patterns and in precipitates. Alpha-B-crystallin and Hsp90 occurred at lower signals in MUT than in CTR muscle, appearing also in precipitates with desmin. The abnormal features in MUT may be the consequence of inactivity, malnutrition, denervation, and failure of protein homeostasis. The latter could be at least in part caused by malfunction of the CCT complex with the mutant CCT5 subunit. This is suggested by the results of the in silico analyses of the mutant CCT5 molecule, which revealed various abnormalities when compared with the wild-type counterpart, mostly affecting the apical domain and potentially impairing chaperoning functions. Thus, analysis of mutated CCT5 in vitro and in vivo is anticipated to provide additional insights on subunit involvement in neuromuscular disorders.

Abstract Background Nucleosomes wrap the DNA into the nucleus of the Eukaryote cell and regulate its transcription phase. Several studies indicate that nucleosomes are determined by the combined effects of several factors, including DNA sequence organization. Interestingly, the identification of nucleosomes on a genomic scale has been successfully performed by computational methods using DNA sequence as input data. Results In this work, we propose CORENup, a deep learning model for nucleosome identification. CORENup processes a DNA sequence as input using one-hot representation and combines in a parallel fashion a fully convolutional neural network and a recurrent layer. These two parallel levels are devoted to catching both non periodic and periodic DNA string features. A dense layer is devoted to their combination to give a final classification. Conclusions Results computed on public data sets of different organisms show that CORENup is a state of the art methodology for nucleosome positioning identification based on a Deep Neural Network architecture. The comparisons have been carried out using two groups of datasets, currently adopted by the best performing methods, and CORENup has shown top performance both in terms of classification metrics and elapsed computation time.

Machine-learning techniques, properly combined with data structures, have resulted in Learned Static Indexes, innovative and powerful tools that speed up Binary Searches with the use of additional space with respect to the table being searched into. Such space is devoted to the machine-learning models. Although in their infancy, these are methodologically and practically important, due to the pervasiveness of Sorted Table Search procedures. In modern applications, model space is a key factor, and a major open question concerning this area is to assess to what extent one can enjoy the speeding up of Binary Searches achieved by Learned Indexes while using constant or nearly constant-space models. In this paper, we investigate the mentioned question by (a) introducing two new models, i.e., the Learned k-ary Search Model and the Synoptic Recursive Model Index; and (b) systematically exploring the time–space trade-offs of a hierarchy of existing models, i.e., the ones in the reference software platform Searching on Sorted Data, together with the new ones proposed here. We document a novel and rather complex time–space trade-off picture, which is informative for users as well as designers of Learned Indexing data structures. By adhering to and extending the current benchmarking methodology, we experimentally show that the Learned k-ary Search Model is competitive in time with respect to Binary Search in constant additional space. Our second model, together with the bi-criteria Piece-wise Geometric Model Index, can achieve speeding up of Binary Search with a model space of 0.05% more than the one taken by the table, thereby, being competitive in terms of the time–space trade-off with existing proposals. The Synoptic Recursive Model Index and the bi-criteria Piece-wise Geometric Model complement each other quite well across the various levels of the internal memory hierarchy. Finally, our findings stimulate research in this area since they highlight the need for further studies regarding the time–space relation in Learned Indexes.

BACKGROUND/AIMS: To preliminarily evaluate the repeatability of central corneal thickness (CCT) measurements performed with Anterior Segment Optical Coherence Tomography (AS-OCT) on eye bank posterior corneal lenticules. METHODS: Six donor lenticules were created with a 350 µm head microkeratome (Moria, Antony, France). All donor tissues were stored at 4°C in Eusol-C solution (Alchimia S.r.l, Ponte S. Nicolò, Italy), without the anterior cornea lamella. The CCT of each lenticule, maintained in the glass phial, was measured using a commercial AS-OCT instrument (Visante, Carl Zeiss Meditec, Dublin, California, USA) and a specially designed adaptor immediately and 4, 24 and 48 hours after dissection. Immediately after AS-OCT, CCT values were measured with the ultrasound pachymetry method used at the Eye Bank. RESULTS: The mean donor cornea central thickness was 647±36 µm and 660±38 µm (p=0.001) as measured by AS-OCT and ultrasound, respectively; immediately after dissection, CCT values of posterior lenticules were 235±43 µm and 248±44 µm, respectively (p=0.001). No statistically significant changes in CCT values of donor lenticules were assessed over the 48 h period with both methods. There was a high level of agreement, evidenced by Bland–Altman analysis, between the two methods of pachymetry. CONCLUSION: AS-OCT, with the corneal tissue in the vial, was revealed to be a repeatable and reliable method for measuring posterior donor lenticule central thickness. Lenticule CCT values measured with the investigational AS-OCT method were on average 10 µm thinner than those measured with the established ultrasound method. [ABSTRACT FROM AUTHOR]