Repositório DSpace

Comunidades no DSpace

Selecione uma comunidade para navegar por suas coleções

Agora exibindo 1 - 4 de 4

Gestão 0
Publicações Científicas e Culturais 23308
Recursos Educacionais 450
Trabalhos Acadêmicos 53791

Submissões Recentes

Previsão de resposta patológica completa do câncer de mama triplo negativo utilizando virtual staining
(Universidade Federal de Minas Gerais, 2026-01-20) Henrique Colonese Echternacht
Background and Objectives: Predicting pathological complete response (pCR) to neoadjuvant chemotherapy is particularly relevant in triple-negative breast cancer (TNBC), a highly aggressive subtype with limited targeted treatment options. Recent deep learning systems for pCR prediction have shown strong performance, but often rely on multiple whole-slide images (WSIs) per case: a routine hematoxylin and eosin (H&E) slide plus additional immunohistochemical (IHC) stains such as Ki-67 and PHH3. These IHC stains are costly, time-consuming, and not routinely available in many clinical settings, which limits the scalability of such multi-stain AI models. This work aims to investigate whether virtual IHC staining derived from H&E can reduce the dependence on physical IHC slides in TNBC pCR prediction. Methods: We developed and evaluated a computational pathology pipeline that replaces physical Ki-67 and PHH3 slides with virtually stained counterparts generated from H&E WSIs using generative deep learning. Among several generative models, a CycleGAN-based architecture was selected for virtual staining of Ki-67 and PHH3. The resulting real and virtual IHC images were then used to derive spatial attention maps, which guided a deep learning model for pCR prediction. The system was trained and validated on a cohort of 73 TNBC patients with serial H&E, Ki-67 and PHH3 slides. We quantitatively assessed (i) the quality of the virtual stains, (ii) the similarity between real and virtual attention maps, and (iii) the impact of using virtual versus real IHC on pCR prediction performance. Results: The virtual staining module produced synthetic Ki-67 and PHH3 images with high structural and intensity similarity to their real counterparts, with mean Intersection over Union (IoU) values of approximately 0.81 (Ki-67) and 0.80 (PHH3), and consistently high perceptual similarity measures. Attention maps derived from virtual IHC showed low divergence from those obtained with real IHC (Jensen–Shannon divergence below 0.04), indicating that spatially relevant regions were largely preserved. In the pCR prediction task, the best configuration using real H&E plus real IHC slides achieved a patient-level Area Under the Curve (AUC) of about 0.94, whereas the corresponding configuration using virtual Ki-67 and PHH3 reached an AUC close to 0.90, substantially outperforming a baseline model using H&E only (AUC 0.81). Conclusions: The proposed virtual staining pipeline retains most of the discriminative power of a multi-stain, IHC-based pCR prediction system, while removing the need for costly and time-consuming wet-lab IHC procedures. By enabling pCR prediction from routinely available H&E slides plus virtually generated Ki-67 and PHH3, this approach offers a cost-effective and scalable alternative for large-scale retrospective studies and, potentially, for future clinical deployment in settings where physical IHC is limited or unavailable.
Dual RNA-seq identifies genes and pathways modulated during Clostridioides difficile colonization
(Universidade Federal de Minas Gerais, 2023) Lucy Frost; Ricardo Stark; Bênção Anonye; Thomas MacCreath; Ludmila Rodrigues Pinto Ferreira Camargo; Meera Unnikrishnan
The gastrointestinal pathogen Clostridioides difficile is the most common cause of hospital-acquired diarrhea. Bacterial interactions with the gut mucosa are crucial for the establishment of C. difficile infection; however, key infection events like bacterial attachment and gut penetration are still poorly defined. To better understand the initial events that occur when this anaerobe interacts with the human gut epithelium, we employed a dual RNA-sequencing approach to study the bacterial and host transcriptomic profiles during C. difficile infection in a dual environment in vitro human gut model. Temporal changes in gene expression during infection were studied in bacterial and epithelial cells over 3–24 hours. While there were several common differentially expressed bacterial genes across different timepoints after infection, mammalian transcriptional profiles were quite distinct, with little overlap. Interestingly, an induction of colonic receptors for C. difficile toxins was observed, along with the downregulation of genes encoding immune response markers. Several cell wall-associated proteins were downregulated in C. difficile when associated with host cells, including slpA, which encodes the main S-layer protein. Gene function and pathway enrichment analyses revealed a potential modulation of the purine/pyrimidine synthesis pathways both in the mammalian and bacterial cells. We observed that proline-proline endopeptidase, a secreted metalloprotease responsible for cell surface protein cleavage, is downregulated during infection, and a mutant lacking this enzyme demonstrated enhanced adhesion to epithelial cells during infection. This study provides new insight into the host and bacterial pathways based on gene expression modulation during the initial contact of C. difficile with gut cells.
A Major Downregulation of Circulating microRNAs in Zika Acutely Infected Patients: potential implications in innate and adaptive immune response signaling pathways
(Universidade Federal de Minas Gerais, 2022) Ana Carolina Carvalho Silva; Almir Ribeiro da Silva Junior; Vagner Oliveira Carvalho Rigaud; Waleska Kerllen Martins; Verônica Coelho; Irmtraut Araci Hoffmann Pfrimer; Jorge Kalil; Simone Gonçalves Fonseca; Edecio Cunha Neto; Ludmila Rodrigues Pinto Ferreira
Zika virus (ZIKV) is an arbovirus mainly transmitted by mosquitos of the genus Aedes. The first cases of ZIKV infection in South America occurred in Brazil in 2015. The infection in humans causes diverse symptoms from asymptomatic to a syndrome-like dengue infection with fever, arthralgia, and myalgia. Furthermore, ZIKV infection during pregnancy is associated with fetal microcephaly and neurological disorders. The identification of host molecular mechanisms responsible for the modulation of different signaling pathways in response to ZIKV is the first step to finding potential biomarkers and therapeutic targets and understanding disease outcomes. In the last decade, it has been shown that microRNAs (miRNAs) are important post-transcriptional regulators involved in virtually all cellular processes. miRNAs present in body fluids can not only serve as key biomarkers for diagnostics and prognosis of human disorders but also contribute to cellular signaling offering new insights into pathological mechanisms. Here, we describe for the first time ZIKV-induced changes in miRNA plasma levels in patients during the acute and recovery phases of infection. We observed that during ZIKV acute infection, among the dysregulated miRNAs (DMs), the majority is with decreased levels when compared to convalescent and control patients. We used systems biology tools to build and highlight biological interactions between miRNAs and their multiple direct and indirect target molecules. Among the 24 DMs identified in ZIKV + patients, miR-146, miR-125a-5p, miR-30-5p, and miR-142-3p were related to signaling pathways modulated during infection and immune response. The results presented here are an effort to open new vistas for the key roles of miRNAs during ZIKV infection.
Um estudo sobre a equação de avanço de Kolmogorov em escalas temporais
(Universidade Federal de Minas Gerais, 2025-02-27) Patrick de Souza Oliveira
This work investigates the evolution of distributions associated with the solutions of stochastic dynamic equations on time scales over time given an initial condition. Similar to dynamic differential equations on time scales, explicit solutions of stochastic dynamic equa-tions are rarely accessible but can be indirectly analyzed through their distributions, when these exist. Initially, we determine the Kolmogorov forward equation (or Fokker-Planck equation) for stochastic problems on time scales, identifying the partial dynamic equation governing these distributions and proving their existence and uniqueness. With these results, we address the explicit representation of these distributions by employing differen-tiable extensions and Neumann series, providing a functional formula that systematically describes them.
Blood DNA methylation marks discriminate Chagas cardiomyopathy disease clinical forms
(Universidade Federal de Minas Gerais, 2022) Pauline Brochet; Barbara Ianni; João Paulo Silva Nunes; Amanda Frade; Priscila Camillo Teixeira; Charles Mady; Ludmila Rodrigues Pinto Ferreira Camargo; Andreia Kuramoto; Cristina Wide Pissetti; Bruno Saba; Darlan da Silva Cândido; Fabrício Dias; Marcelo Sampaio; José Marin Neto; Abílio Fragata; Ricardo Costa Fernandes Zaniratto; Sergio Siqueira; Giselle de Lima Peixoto; Vagner Oliveira Carvalho Rigaud; Paula Buck; Rafael Ribeiro Almeida; Hui Tzu Lin-Wang; André Schmidt; Martino Martinelli; Mario Hiroyuki Hirata; Eduardo Donadi; Virmondes Rodrigues Junior; Alexandre Costa Pereira; Jorge Kalil; Lionel Spinelli; Edecio Cunha Neto; Christophe Chevillard
Chagas disease is a parasitic disease from South America, affecting around 7 million people worldwide. Decades after the infection, 30% of people develop chronic forms, including Chronic Chagas Cardiomyopathy (CCC), for which no treatment exists. Two stages characterized this form: the moderate form, characterized by a heart ejection fraction (EF) ≥ 0.4, and the severe form, associated to an EF < 0.4. We propose two sets of DNA methylation biomarkers which can predict in blood CCC occurrence, and CCC stage. This analysis, based on machine learning algorithms, makes predictions with more than 95% accuracy in a test cohort. Beyond their predictive capacity, these CpGs are located near genes involved in the immune response, the nervous system, ion transport or ATP synthesis, pathways known to be deregulated in CCCs. Among these genes, some are also differentially expressed in heart tissues. Interestingly, the CpGs of interest are tagged to genes mainly involved in nervous and ionic processes. Given the close link between methylation and gene expression, these lists of CpGs promise to be not only good biomarkers, but also good indicators of key elements in the development of this pathology.

Busca

Comunidades no DSpace

Submissões Recentes