Percorrer por autor "Geraldes, Mariana"
A mostrar 1 - 2 de 2
Resultados por página
Opções de ordenação
- Biomarkers in whole slide images stained by: Hemtoxylin-Eosin: A groundbreaking application using artificial intelligencePublication . Borrecho, Gonçalo; Curado, Mónica; Vale, João; Vinagre, Tiago; Geraldes, Mariana; Assunção, Teresa; Coelho, Daniel; Ferreira, Inês; Ferreira, Ana; Fernandes, Ana Isabel; França, Amélia; Mendes, Fernando; Martins, DianaBiomarkers play a fundamental role in the diagnosis, prognosis and prediction of diseases. The study of biomarkers requires the performance of complementary diagnostic tests, which entails high costs and inevitably leads to an increase in response time, which could have a severe impact on the patient’s outcome. The digital transformation in Pathology Laboratories, accompanied by the wide implementation of slide digitalization, has been decisive for the development and application of digital intelligence algorithms in a diagnostic context. The aim of this review is to assess artificial intelligence algorithms for evaluating biomarkers that can be applied to whole slide images stained by hemtoxylin-eosin (WSI-HE) and to understand their advantages and limitations. There are several types of algorithms, some established on the identification and quantification of morphological biomarkers, such as nuclear density, celular heterogeneity, the presence of certain cellular structures, tissue organization and other features. The usage of WSI-HE is enormously promising, as it reveals additional information that is not visually observable but can help or even expand to pathologists capabilities. The identification and validation of morphological biomarkers in WSI-HE still presents challenges, such as the need for large data sets annotated using multimodal data (information from diferente sources, such as histopathological images, clinical data, radiologial information, genomic data, among others), the interpretability of deep learning models, the integration of these biomarkers into clinical practice, among others. The application of algorithms in WSI-HE could represente na importante change in patient management, contributing to timely precision medicine.
- Computacional pathology: What’s newPublication . Coelho, Daniel; Assunção, Teresa; Borrecho, Gonçalo; Geraldes, Mariana; Vinagre, Tiago; Ferreira, Inês; Ferreira, Ana; Fernandes, Ana Isabel; França, Amélia; Vale, João; Curado, Mónica; Mendes, Fernando; Martins, DianaThe term computacional pathology (CPath) has become a buzz-word among the digital pathology community. Adances in scanning systems, imaging technologies and storage devices are generating an ever-increasing volume of whole-slide images (WSI) acquired in clinical settings, which can be computacionally analyzed using artificial intelligence (AI), such as deep learning technologies, in a new área of development called CPath. The purpose of the review is to disseminate the latest news and futures perspectives by CPath. Deep learning in the context of CPath has methodological contributions that can be distinguished into approaches based on the final purpose of the analysis: predicting clinical endpoints such as cancer subtype, patient survival or genetic mutations from WSI and AI-based assistive tools, such as segmentation methods for WSI or virtual staining. The emergence of multipex imaging, spatially resolver genomic assays and 3D pathology, among other methodologies, will accelerate this trend, providing new opportunities for multimodal integration and discovering new biomarkers. Additionally, these developments will help automating labor-intensive manual work and reducing inter-observer variability diagnosis between pathologists, contributing to a better patient care. CPath will underpin the development of the next generation of cancer therapies and diagnostics, changing the clinical research and ultimately leading towards new cures or improved patient outcomes.