RESUMEN
Following the great success of various deep learning methods in image and object classification, the biomedical image processing society is also overwhelmed with their applications to various automatic diagnosis cases. Unfortunately, most of the deep learning-based classification attempts in the literature solely focus on the aim of extreme accuracy scores, without considering interpretability, or patient-wise separation of training and test data. For example, most lung nodule classification papers using deep learning randomly shuffle data and split it into training, validation, and test sets, causing certain images from the Computed Tomography (CT) scan of a person to be in the training set, while other images of the same person to be in the validation or testing image sets. This can result in reporting misleading accuracy rates and the learning of irrelevant features, ultimately reducing the real-life usability of these models. When the deep neural networks trained on the traditional, unfair data shuffling method are challenged with new patient images, it is observed that the trained models perform poorly. In contrast, deep neural networks trained with strict patient-level separation maintain their accuracy rates even when new patient images are tested. Heat map visualizations of the activations of the deep neural networks trained with strict patient-level separation indicate a higher degree of focus on the relevant nodules. We argue that the research question posed in the title has a positive answer only if the deep neural networks are trained with images of patients that are strictly isolated from the validation and testing patient sets.
RESUMEN
The onset of obstetric antiphospholipid syndrome (APS) occurs when antiphospholipid antibodies act upon the placenta. During pregnancy, APS exhibits traits such as vascular thrombosis, inflammation, and hindered trophoblast implantation. The involvement of microRNA expression has been proposed as a genetic factor contributing to the syndrome's development. MicroRNAs play a role in regulating gene expression in various cellular processes, including the formation of placental tissue. Therefore, additional research is needed to explore the control of placental miRNA in APS. In this study, we aimed to profile miRNA expressions from placenta tissue of patients with APS. Differentially expressed miRNAs were determined for its targeted genes and pathways. Agilent microarray platform was used to measure placental microRNA expressions between normal placental tissue and those obtained from patients with APS. Differentially expressed miRNAs were detected using GeneSpring GX software 14.2 and sequences were mapped using TargetScan software to generate the predicted target genes. Pathway analysis for the genes was then performed on PANTHER and REACTOME software. Selected miRNAs and their associated genes of interest were validated using qPCR. Microarray findings revealed, 9 downregulated and 21 upregulated miRNAs expressed in placenta of patients with APS. Quantitative expressions of 3 selected miRNAs were in agreement with the microarray findings, however only miR-525-5p expression was statistically significant. Pathway analysis revealed that the targeted genes of differentially expressed miRNAs were involved in several hypothesised signalling pathways such as the vascular endothelial (VE) growth factor (VEGF) and inflammatory pathways. VE-cadherin, ras homolog member A (RHOA) and tyrosine kinase receptor (KIT) showed significant downregulation while Retinoblastoma gene (RET), Dual specificity protein phosphatase 10 (DUSP10) and B-lymphocyte kinase (BLK) genes were significantly upregulated. These preliminary findings suggest the involvement of miRNAs and identified novel associated genes involvement in the mechanism of obstetric APS, particularly through the alteration of vascular-associated regulators and the inflammatory signalling cascade.