RESUMO
Advanced object detection methods always face high algorithmic complexity or low accuracy when used in pedestrian target detection for the autonomous driving system. This paper proposes a lightweight pedestrian detection approach called the YOLOv5s-G2 network to address these issues. We apply Ghost and GhostC3 modules in the YOLOv5s-G2 network to minimize computational cost during feature extraction while keeping the network's capability of extracting features intact. The YOLOv5s-G2 network improves feature extraction accuracy by incorporating the Global Attention Mechanism (GAM) module. This application can extract relevant information for pedestrian target identification tasks and suppress irrelevant information, improving the unidentified problem of occluded and small targets by replacing the GIoU loss function used in the bounding box regression with the α-CIoU loss function. The YOLOv5s-G2 network is evaluated on the WiderPerson dataset to ensure its efficacy. Our proposed YOLOv5s-G2 network offers a 1.0% increase in detection accuracy and a 13.2% decrease in Floating Point Operations (FLOPs) compared to the existing YOLOv5s network. As a result, the YOLOv5s-G2 network is preferable for pedestrian identification as it is both more lightweight and more accurate.
RESUMO
BACKGROUND: Radiotherapy remains one of the cornerstones to improve the outcome of colorectal cancer (CRC) patients. Radiotherapy of the CRC not only help to destroy cancer cells but also remodel the tumour microenvironment by enhancing tumour-specific tropism of bone marrow-derived mesenchymal stromal cell (BM-MSC) from the peripheral circulation. However, the role of local MSCs and recruited BM-MSC under radiation were not well defined. Indeed, the functions of BM-MSC without irradiation intervention remained controversial in tumour progression: BM-MSC was previously shown to modulate the immune function of major immune cells, resulting in an impaired immunological sensitivity and to induce an increased risk of tumour recurrence. In contrast, it could also secrete various cytokines and possess anticancer effect. METHODS: Three co-cultivation modules, 3D culture modules, and cancer organoids were established. The induction of cytokines secretion in hBM-MSCs after irradiation was analysed by ELISA array and flow cytometry. AutoMac separator was used to separate hBM-MSC and CRC automatically. Cells from the co-cultured group and the control group were then irradiated by UV-C lamp and X-ray. Proliferation assay and viability assay were performed. RESULTS: In this study, we show that BM-MSCs can induce the EMT progression of CRC cells in vitro. When irradiated with low doses of ultraviolet radiation and X-rays, BM-MSCs show an anti-tumour effect by secreting certain cytokine (TNF-α, IFN-γ) that lead to the inhibition of proliferation and induction of apoptosis of CRC cells. This was further verified in a 3D culture model of a CRC cell in vitro. Furthermore, irradiation on the co-culture system induced the cleavage of caspase3, and attenuated the phosphorylation of phosphatidylinositol 3-kinase (PI3K)/AKT and extracellular signal-regulated kinase in cancer cells. The signal pathways above might contribute to the cancer cell death. CONCLUSIONS: Taken together, we show that BM-MSC can potentially promote the effect of radiotherapy in CRC.