Caracterização radiómica e neuroimagiologia comparada dos gliomas no cão e glioblastomas em humanos
A carregar...
Data
Autores
Título da revista
ISSN da revista
Título do volume
Editora
Resumo
Introdução: Os gliomas são tumores cerebrais primários do SNC com um grande impacto na qualidade de vida dos doentes, sejam cães ou humanos. Os glioblastomas humanos são um subtipo mais agressivo, com uma sobrevida média de 15 meses após o diagnóstico. Os gliomas, mais comuns em raças braquicefálicas, apresentam características biológicas semelhantes aos glioblastomas humanos. Este estudo explorou o potencial da radiómica aplicada a imagens de ressonância magnética (RM) ponderadas em T2-FLAIR para caracterizar as lesões tumorais e tecidos sem lesão, com o intuito de desenvolver novos biomarcadores imagiológicos que auxiliem no diagnóstico e prognóstico de gliomas. Métodos: Foram analisados 39 exames de RM cerebral de gliomas caninos (n=23) e glioblastomas humanos (n=16). As imagens foram segmentadas para definir Regiões de Interesse (ROIs), permitindo a extração de características radiómicas, como a textura, forma e intensidade. A análise incluiu testes estatísticos descritivos e não-paramétricos, ANOVA e curvas ROC, para avaliar a discriminação das características radiómicas selecionadas, nomeadamente JointEntropy, SumEntropy, SumAverage, JointAverage, SumSquares e Autocorrelation Resultados: A análise volumétrica revelou diferenças significativas na distribuição de volumes entre os dois grupos de amostras (p < 0,001). As características radiómicas selecionadas apresentam potencial para discriminar os tecidos em estudo. As curvas ROC para as três principais características atingiram valores de AUC superiores a 80% (p < 0,001). O modelo SVM desenvolvido obteve uma exatidão na classificação de 80,33%, Discussão: As características radiómicas extraídas permitiram a identificação de padrões de heterogeneidade que refletem a agressividade e a complexidade biológica dos gliomas. A análise de neuroimagiologia comparada reforça a hipótese de os gliomas caninos serem um bom modelo natural e translacional da doença, para o estudo dos glioblastomas humanos, contribuindo para a medicina comparada. Conclusão: Este estudo demonstrou o potencial da radiómica para caracterizar gliomas caninos e glioblastomas humanos. A identificação de biomarcadores imagiológicos com resultados consistentes fornece uma base para realizar novos avanços na medicina comparada VI e translacional, integrando abordagens comparativas para o desenvolvimento de diagnósticos não invasivos e contribuindo para uma medicina de precisão.
Introduction: Gliomas are primary CNS brain tumors that significantly impact the quality of life of patients, whether in dogs or humans. Human glioblastomas are the most aggressive tumor subtype, with an average survival of 15 months after diagnosis. Brachycephalic dogs are the most predisposed breed to gliomas, exhibiting biological characteristics similar to human glioblastomas. This study explored the potential of radiomics applied to T2-FLAIR-weighted magnetic resonance imaging (MRI) to characterize tumor lesions and non-lesioned tissues, aiming to develop new imaging biomarkers to aid in glioma diagnosis and prognosis. Methods: MRI scans of canine gliomas and human glioblastomas were analyzed. The images were segmented to define Regions of Interest (ROIs), enabling the extraction of radiomic features such as texture, shape, and intensity. The analysis included descriptive and non-parametric statistical tests, ANOVA, and ROC curves to assess the discrimination capacity of radiomic features, particularly those with an area under the curve (AUC) greater than 80%, such as jointaverage and autocorrelation. Results: Volumetric analysis revealed significant differences in volume distribution between the two sample groups (p < 0.001). The selected radiomic features showed potential for discriminating the studied tissues. ROC curves for the three main features achieved AUC values above 80% (p < 0.001). The developed SVM model achieved a classification accuracy of 80.33%. Discussion: The extracted radiomic features allowed the identification of heterogeneity patterns reflecting the aggressiveness and biological complexity of gliomas. Comparative neuroimaging analysis reinforces the hypothesis that canine gliomas are a valuable natural and translational model for studying human glioblastomas, contributing to precision medicine. The study's limitations include the small sample size and the need for validation in future studies with greater population diversity. Conclusion: This study demonstrated the potential of radiomics to characterize canine gliomas and human glioblastomas. The identification of consistent imaging biomarkers provides a foundation for new advances in comparative and translational medicine, integrating comparative approaches to develop non-invasive diagnostics and contributing to precision medicine.
Introduction: Gliomas are primary CNS brain tumors that significantly impact the quality of life of patients, whether in dogs or humans. Human glioblastomas are the most aggressive tumor subtype, with an average survival of 15 months after diagnosis. Brachycephalic dogs are the most predisposed breed to gliomas, exhibiting biological characteristics similar to human glioblastomas. This study explored the potential of radiomics applied to T2-FLAIR-weighted magnetic resonance imaging (MRI) to characterize tumor lesions and non-lesioned tissues, aiming to develop new imaging biomarkers to aid in glioma diagnosis and prognosis. Methods: MRI scans of canine gliomas and human glioblastomas were analyzed. The images were segmented to define Regions of Interest (ROIs), enabling the extraction of radiomic features such as texture, shape, and intensity. The analysis included descriptive and non-parametric statistical tests, ANOVA, and ROC curves to assess the discrimination capacity of radiomic features, particularly those with an area under the curve (AUC) greater than 80%, such as jointaverage and autocorrelation. Results: Volumetric analysis revealed significant differences in volume distribution between the two sample groups (p < 0.001). The selected radiomic features showed potential for discriminating the studied tissues. ROC curves for the three main features achieved AUC values above 80% (p < 0.001). The developed SVM model achieved a classification accuracy of 80.33%. Discussion: The extracted radiomic features allowed the identification of heterogeneity patterns reflecting the aggressiveness and biological complexity of gliomas. Comparative neuroimaging analysis reinforces the hypothesis that canine gliomas are a valuable natural and translational model for studying human glioblastomas, contributing to precision medicine. The study's limitations include the small sample size and the need for validation in future studies with greater population diversity. Conclusion: This study demonstrated the potential of radiomics to characterize canine gliomas and human glioblastomas. The identification of consistent imaging biomarkers provides a foundation for new advances in comparative and translational medicine, integrating comparative approaches to develop non-invasive diagnostics and contributing to precision medicine.
Descrição
Palavras-chave
VETERINARY MEDICINE, GLIOMA, MAGNETIC RESONANCE IMAGING, BIOMARKERS, MEDICAL IMAGING, GLIOBLASTOMA, NERVOUS SYSTEM DISEASES, NEOPLASMS, DOGS, COMPUTER-ASSISTED DIAGNOSIS, MESTRADO INTEGRADO EM MEDICINA VETERINÁRIA, VETERINÁRIA, MEDICINA VETERINÁRIA, GLIOMA, RESSONÂNCIA MAGNÉTICA, BIOMARCADORES, IMAGIOLOGIA, GLIOBLASTOMA, DOENÇAS DO SISTEMA NERVOSO, NEOPLASIAS, CÃES, DIAGNÓSTICO ASSISTIDO POR COMPUTADOR