243 research outputs found
Image Embeddings Extracted from CNNs Outperform Other Transfer Learning Approaches in Classification of Chest Radiographs
Full text linkTo identify the best transfer learning approach for the identification of the most frequent abnormalities on chest radiographs (CXRs), we used embeddings extracted from pretrained convolutional neural networks (CNNs). An explainable AI (XAI) model was applied to interpret black-box model predictions and assess its performance. Seven CNNs were trained on CheXpert. Three transfer learning approaches were thereafter applied to a local dataset. The classification results were ensembled using simple and entropy-weighted averaging. We applied Grad-CAM (an XAI model) to produce a saliency map. Grad-CAM maps were compared to manually extracted regions of interest, and the training time was recorded. The best transfer learning model was that which used image embeddings and random forest with simple averaging, with an average AUC of 0.856. Grad-CAM maps showed that the models focused on specific features of each CXR. CNNs pretrained on a large public dataset of medical images can be exploited as feature extractors for tasks of interest. The extracted image embeddings contain relevant information that can be used to train an additional classifier with satisfactory performance on an independent dataset, demonstrating it to be the optimal transfer learning strategy and overcoming the need for large private datasets, extensive computational resources, and long training times
Explainable domain transfer of distant supervised cancer subtyping model via imaging-based rules extraction
Full text linkImage texture analysis has for decades represented a promising opportunity for cancer assessment and disease progression evaluation, evolving in a discipline, i.e., radiomics. However, the road to a complete translation into clinical practice is still hampered by intrinsic limitations. As purely supervised classification models fail in devising robust imaging-based biomarkers for prognosis, cancer subtyping approaches would benefit from the employment of distant supervision, for instance exploiting survival/recurrence information. In this work, we assessed, tested, and validated the domain-generality of our previously proposed Distant Supervised Cancer Subtyping model on Hodgkin Lymphoma. We evaluate the model performance on two independent datasets coming from two hospitals, comparing and analyzing the results. Although successful and consistent, the comparison confirmed the instability of radiomics due to an across-center lack of reproducibility, leading to explainable results in one center and poor interpretability in the other. We thus propose a Random Forest-based Explainable Transfer Model for testing the domain-invariance of imaging biomarkers extracted from retrospective cancer subtyping. In doing so, we tested the predictive ability of cancer subtyping in a validation and perspective setting, which led to successful results and supported the domain-generality of the proposed approach. On the other hand, the extraction of decision rules enables to draw of risk factors and robust biomarkers to inform clinical decisions. This work shows the potentialities of the Distant Supervised Cancer Subtyping model to be further evaluated in larger multi-center datasets, to reliably translate radiomics into medical practice
EMoCy: Towards Physiological Signals-Based Stress Detection
No full textThe prompt detection of stress can prevent long-term consequences on people's health, economy and society. In this work, we present EMoCy, a reproducible methodology and analysis pipeline for automatic and continuous stress detection based on physiological signals. By providing reproducible experiments and exploring more constrained settings with state-of-the-art accuracy, we set new benchmarks for future works on stress detection. Our study includes signal selection and preprocessing, feature engineering, and classification using machine learning algorithms. We have tested our approach on WESAD dataset using blood volume pulse, electrodermal activity, and respiration signals. We identify a set of features that allow us to reach effective detection using low sampling frequency and short time windows. This reduces the computational power and the detection delay, going towards real-time applications on wearable devices. Using 60s windows, we reach an accuracy of 0.972 and an F1 score of 0.979 on the stress/baseline discrimination task. Moreover, our machine learning models achieve an accuracy greater than 0.93 on windows of 25s sampled at 64Hz. We have also observed that it is possible to obtain a good estimate of the models' performance by training them with a small window overlap
The Development of an Intelligent Agent to Detect and Non-Invasively Characterize Lung Lesions on CT Scans: Ready for the “Real World”?
Full text link(1) Background: Once lung lesions are identified on CT scans, they must be characterized by assessing the risk of malignancy. Despite the promising performance of computer-aided systems, some limitations related to the study design and technical issues undermine these tools’ efficiency; an “intelligent agent” to detect and non-invasively characterize lung lesions on CT scans is proposed. (2) Methods: Two main modules tackled the detection of lung nodules on CT scans and the diagnosis of each nodule into benign and malignant categories. Computer-aided detection (CADe) and computer aided-diagnosis (CADx) modules relied on deep learning techniques such as Retina U-Net and the convolutional neural network; (3) Results: Tests were conducted on one publicly available dataset and two local datasets featuring CT scans acquired with different devices to reveal deep learning performances in “real-world” clinical scenarios. The CADe module reached an accuracy rate of 78%, while the CADx’s accuracy, specificity, and sensitivity stand at 80%, 73%, and 85.7%, respectively; (4) Conclusions: Two different deep learning techniques have been adapted for CADe and CADx purposes in both publicly available and private CT scan datasets. Experiments have shown adequate performance in both detection and diagnosis tasks. Nevertheless, some drawbacks still characterize the supervised learning paradigm employed in networks such as CNN and Retina U-Net in real-world clinical scenarios, with CT scans from different devices with different sensors’ fingerprints and spatial resolution. Continuous reassessment of CADe and CADx’s performance is needed during their implementation in clinical practice
Homozygous loss of autism-risk gene CNTNAP2 results in reduced local and long-range prefrontal functional connectivity
Full text linkFunctional connectivity aberrancies, as measured with resting-state fMRI (rsfMRI), have been consistently observed in the brain of autism spectrum disorders (ASD) patients. However, the genetic and neurobiological underpinnings of these findings remain unclear. Homozygous mutations in Contactin Associated Protein-like 2 (CNTNAP2), a neurexin-related cell-adhesion protein, are strongly linked to autism and epilepsy. Here we used rsfMRI to show that homozygous mice lacking Cntnap2 exhibit reduced long-range and local functional connectivity in prefrontal and midline brain “connectivity hubs”. Long-range rsfMRI connectivity impairments affected heteromodal cortical regions and were prominent between frontoposterior components of the mouse default-mode network (DMN), an effect that was associated with reduced social investigation, a core “autism trait” in mice. Notably, viral tracing revealed reduced frequency of prefrontal-projecting neural clusters in the cingulate cortex of Cntnap2-/- mutants, suggesting a possible contribution of defective mesoscale axonal wiring to the observed functional impairments. Macroscale cortico-cortical white matter organization appeared to be otherwise preserved in these animals. These findings reveal a key contribution of ASD-associated gene CNTNAP2 in modulating macroscale functional connectivity, and suggest that homozygous loss-of-function mutations in this gene may predispose to neurodevelopmental disorders and autism through a selective dysregulation of connectivity in integrative prefrontal areas
Domestication process of two Solanum (section Lasiocarpa) species among Amerindians in the Upper Orinoco, Venezuela with special focus on Piaroa indians
No full textVolpato, Gabriele, Rossella Marcucci, Noemi Tornadore, and Maurizio G. Paoletti (Department
of Biology, University of Padua, 325100 Padova, Italy; email corresponding author:
[email protected]). DOMESTICATION PROCESS OF TWO SOLANUM SECTION LASIOCARPA SPECIES
AMONG AMERINDIANS IN THE UPPER ORINOCO, VENEZUELA, WITH SPECIAL FOCUS ON PIAROA
INDIANS. Economic Botany 58(2):000–000, 2004. Two semi-cultivated Solanum species (S. sessiliflorum
Dunal and S. stramonifolium Jacq.) are utilized by the Amazonian Indians of the
Upper Orinoco Basin in Venezuela. The manner by which they have become partially domesticated
by the Piaroas and other native tribes of this rain forest region is elucidated in the
following text. Both species have two varieties, with and without prickles, the latter the result
of human selection. Patterns of indigenous utilization of these species by the selection of morphologic
forms and to the differentiation of karyotypes of varieties, and exploitation of the
species reflects also in the perception of them among users. S. sessiliflorum is cultivated in
swiddens and has an economic role, whereas S. stramonifolium is grown in dooryards. This
difference is detectable to the Piaroas, as they recognize in their folk taxonomy three different
varieties of the S. sessiliflorum and one of S. stramonifolium, this according to the stage of
domestication of the species and the way in which they are utilized
The Umma Messenger Texts from the Harvard Semitic Museum and the Yale Babylonian Collection - Part 1. Nisaba 27. Eisenbrauns
No full textIn this volume, the author publishes 240 Messenger Texts from the city of Umma, texts that are currently housed in the collections of the Yale Babylonian Collection and the Harvard Semitic Museum. Earlier volumes of Nisaba published nearly 900 similar Messenger Texts that are in the collections of the British Museum. The texts published here range in date from the fifth month of Amar-Suen 3 to the twelfth month of Ibbi-Sîn. These administrative records provide data on the allotment of rations and disbursement of goods and thus form a basis for further study of the sociology and economics of Neo-Sumerian times in and around the city of Umma
La devozione al Sacro Chiodo fra Lombardia e Piemonte nell'età moderna
No full textDrawing on local sources and on old and new historical literature, the author examines the social and geographical spread of the cult of the Holy Nail, a relic from Christ’s Cross preserved in the Milan Cathedral. The related iconography centers round the procession organized by Carlo Borromeo during the plague of 1576, when the Sacro Chiodo was carried through the streets of Milan, and the Saint’s personal veneration for the relic. From Milan, where most of the surviving paintings were produced, the cult spread to the near-by province of Novara, including the town itself, lake Orta and the Sesia and Ossola valleys. The Sacro Chiodo devotion can thus be regarded as typical of the religious climate of the area spanning present-day western Lombardy and eastern Piedmont
First isolation of Leptospira interrogans from the arboreal squirrel Callosciurus erythraeus introduced in Argentina
Full text linkWe isolated Leptospira strains from renal tissue samples of Asiatic red-bellied squirrels Callosciurus erythraeus, captured in Argentina. In December 2008, we captured 34 squirrels in the province of Santa Fe during a short control campaign, which were screened for Leptospira spp. by direct immunofluorescence assays and isolation attempts in renal tissue samples. A prevalence of 38% of Leptospira spp. was obtained by direct immunofluorescence. The isolates, genotyped by the MLVA, were identified as L. interrogans Canicola (N = 2) and L. interrogans Icterohaemorraghiae (N = 1), showing an identical genetic profile to the reference strains Hond Utrech IV (serovar Canicola, serogroup Canicola) and M20 (serovar Copenhageni, serogroup Icterohaemorrhagiae), respectively. Our study shows for the first time that the red-bellied squirrel is a renal carrier of Leptospira interrogans and might act as a reservoir of leptospirosis both in native range and areas of introduction within several European countries, Japan and Argentina. The close association of this introduced squirrel species with humans and domestic animals may increase the probability of disease transmission, particularly in favourable environments.Fil: Gozzi, Ana Cecilia. Universidad Nacional de Luján. Departamento de Ciencias Básicas; ArgentinaFil: Guichon, Maria Laura. Universidad Nacional de Luján. Departamento de Ciencias Básicas; Argentina. Consejo Nacional de Investigaciones Científicas y Técnicas. Centro Científico Tecnológico Patagonia Norte. Instituto de Investigación en Biodiversidad y Medioambiente; ArgentinaFil: Benitez, Verónica Victoria. Universidad Nacional de Luján. Departamento de Ciencias Básicas; ArgentinaFil: Romero, Graciela Noemi. Instituto Nacional de Tecnologia Agropecuaria. Centro Nacional de Investigaciones Agropecuarias. Centro de Investigación de Ciencias Veterinarias y Agronomicas; ArgentinaFil: Auteri, Carmelo. Instituto Nacional de Tecnologia Agropecuaria. Centro Nacional de Investigaciones Agropecuarias. Centro de Investigación de Ciencias Veterinarias y Agronomicas; ArgentinaFil: Brihuega, Bibiana. Instituto Nacional de Tecnologia Agropecuaria. Centro Nacional de Investigaciones Agropecuarias. Centro de Investigación de Ciencias Veterinarias y Agronomicas; Argentin
Luigi Meneghello traduttore di Hopkins, Cummings, Yeats e Donne: uno studio genetico dei trapianti editi e inediti
Full text linkLa ricerca si propone di indagare l’attività traduttoria di Luigi Meneghello condotta sui testi di Gerard Manley Hopkins, John Donne, William Butler Yeats e Edward Estlin Cummings.
Risale al 2002 la pubblicazione, per Rizzoli, dei Trapianti, volume che raccoglie 41 traduzioni vicentine realizzate da Meneghello a partire da testi di Shakespeare, Wordsworth, Hopkins, Cummings, Empson, Yeats e Campbell.
Le carte autografe che testimoniano le diverse stesure dei trapianti – di proprietà della Fondazione Maria Corti – sono conservate presso il Fondo Meneghello del Centro per gli studi sulla tradizione manoscritta di autori moderni e contemporanei dell’Università di Pavia, all’interno del quale sono presenti però anche le stesure di numerose traduzioni inedite, realizzate a partire da testi sia di autori presenti nei Trapianti (cummings, Hopkins, Yeats), sia di autori che ne sono invece stati esclusi (Arnold, Don Marquis, Larkin, Donne).
Nel corso del lavoro – dopo un primo esame del materiale conservato al Fondo – sono stati selezionati i quattro autori sopracitati, seguendo principalmente un criterio di rappresentatività dell’opera traduttoria meneghelliana.
Prendendo le mosse dallo studio delle carte d’archivio autografe e instaurando un costante raffronto con altre opere di Meneghello, nel corso del lavoro è stato realizzato per ciascun trapianto – integrale o meno – un apparato genetico che rende conto delle varie fasi di elaborazione del testo. Esso permette di orientarsi entro i processi variantistici, spesso fitti e dispiegati lungo numerose stesure.
Per ciascun testo, all’apparato genetico segue un’analisi discorsiva incentrata sui passaggi ritenuti maggiormente esemplificativi della tensione espressiva meneghelliana.
Dal lavoro emerge come i Trapianti siano un’opera tesa a colmare il dislivello che separa una lingua regionale da una lingua nazionale e letteraria, vanificando ogni tendenza alla gerarchizzazione e instaurando un confronto tra culture su un piano di pari dignità.The research aims to investigate Luigi Meneghello’s translations of texts by Gerard Manley Hopkins, John Donne, William Butler Yeats and Edward Estlin Cummings.
Most part of Meneghello’s translation work is published in Trapianti (Rizzoli, 2002). The volume contains forty-three vicentine translations from Shakespeare, Wordsworth, Hopkins, Cummings, Empson, Yeats and Campbell.
The autograph documents, which testify the various drafts of Trapianti, are owned by the Maria Corti Foundation. They are preserved in the Meneghello Fund of the Centre for Studies on the Manuscript Tradition of Modern and Contemporary Authors at the University of Pavia. The Fund includes the drafts of numerous unpublished translations, based on texts by both authors present in the volume (Cummings, Hopkins, Yeats) and authors who were instead excluded (Arnold, Don Marquis, Larkin, Donne).
After an initial examination of the material preserved at the Fund, I selected the four aforementioned authors. Such choice was made following the criterion of representativeness of Meneghello’s translation work.
In this study, every translation made by the Italian author was analyzed – even those unfinished or constituted by a few verses. Every analysis was completed by a genetic apparatus. The research was in fact conducted both from a critical and philological perspective, in order to produce an in-depth study on every phase of the translation work. In addition, a comparison with other literary works by Meneghello was made, to orient the critical readings both synchronically and diachronically.
In the research, the genetic apparatus of each text is followed by a critical analysis. These insights are dedicated to identifying points where translation work deviates most from a literal translation, showing a deeper authorial intervention.
The broader aim is to deepen the programmatic approach of Meneghello’s translation practice. His work of linguistic adaptation appears in fact free and not referential in most cases, as the study allows us to observe.
In conclusion, the study highlighted how Trapianti is a work aimed at bridging the gap that separates a regional language from a national and literary language. The author has thus established a comparison between cultures on equal basis, counteracting the risk of a linguistic hierarchy
- …
