The pre-metastatic niche educated by primary tumor-derived elements contributes to cancer metastasis. However, the role of host stromal cells in metastatic niche formation and organ-specific metastatic tropism is not clearly defined. Here, we demonstrate that lung epithelial cells are critical for initiating neutrophil recruitment and lung metastatic niche formation by sensing tumor exosomal RNAs via Toll-like receptor 3 (TLR3). TLR3-deficient mice show reduced lung metastasis in the spontaneous metastatic models. Mechanistically, primary tumor-derived exosomal RNAs, which are enriched in small nuclear RNAs, activate TLR3 in lung epithelial cells, consequently inducing chemokine secretion in the lung and promoting neutrophil recruitment. Identification of metastatic axis of tumor exosomal RNAs and host lung epithelial cell TLR3 activation provides potential targets to control cancer metastasis to the lung.

No abstract

No abstract

BackgroundOsteosarcoma is the most common primary sarcoma of the bone. Lung osteosarcoma metastases at diagnosis have a significantly poor prognosis, even when surgery plus chemotherapy are performed. Our goal was to analyze clinical and sarcoma characteristics that could help identify factors related to an increased rate of lung metastasis and to identify different modes of treatment and its correlation with survival.Materials and MethodsThe Surveillance, Epidemiology, and End Results (SEER) database was used to identify all osteosarcoma patients diagnosed from 2010 to 2015. Patient characteristics such as age, sex, ethnicity, marital status, tumor location, histologic grade, surgery, chemotherapy, radiation therapy, SEER cause-specific death classification, survival, and lung metastasis were collected. These factors were analyzed using Univariate and multivariate regression models in survival analyses.ResultsA total of 1057 osteosarcoma patients diagnosed from 2010 to 2015 were included, of which 176 were patients with lung metastasis. Substantial disparities in the rate of lung metastasis existed when osteosarcoma patients were stratified according to tumor location (P = 0.0002) and tumor size (P < .001). Using a Multivariate Cox regression model, being older than 30 years (vs. younger than 30, HR = 2.171, 95% CI = 1.623–2.905, P < .0001), having a tumor >5–10 cm (vs. <5 cm, HR = 2.046, 95% CI = 1.153–3.632, P = 0.0014) and >10 cm (vs. <5 cm, HR = 3.610, 95% CI = 2.066–6.310, P < .0001) were related to an increased HR for all-cause death. The HR decreased in patients with surgery (vs. no surgery, HR = 0.189, 95% CI = 0.138–0.260, P < 0.0001) and osteosarcoma. As for osteosarcoma patients with lung metastases, Multivariate Cox regressions revealed that an increased HR was associated with being older than 30 years (vs. younger than 30 years, HR = 2.142, 95% CI = 1.273–3.605, P = .0041) and married (vs. no marriage, HR = 2.418, 95% CI = 1.400–4.176, P= .0015), while a decreased HR was related to having had surgery (vs. no surgery, HR = 0.282, 95% CI = 0.171–0.464, P < .0001) and chemotherapy (vs. no chemotherapy, HR = 0.107, 95% CI = 0.050–0.229, P < .0001).ConclusionsAdvanced age (older than 30 years) and large tumors were related to a higher risk of lung metastases in osteosarcoma patients. Therefore, patients who were diagnosed at advanced age or had large tumors should receive comprehensive chest CT scans. Surgery and chemotherapy can significantly improve the survival of metastatic patients, while radiotherapy did not improve survival in these patients.

Single-image piece-wise planar 3D reconstruction aims to simultaneously segment plane instances and recover 3D plane parameters from an image. Most recent approaches leverage convolutional neural networks (CNNs) and achieve promising results. However, these methods are limited to detecting a fixed number of planes with certain learned order. To tackle this problem, we propose a novel two-stage method based on associative embedding, inspired by its recent success in instance segmentation. In the first stage, we train a CNN to map each pixel to an embedding space where pixels from the same plane instance have similar embeddings. Then, the plane instances are obtained by grouping the embedding vectors in planar regions via an efficient mean shift clustering algorithm. In the second stage, we estimate the parameter for each plane instance by considering both pixel-level and instance-level consistencies. With the proposed method, we are able to detect an arbitrary number of planes. Extensive experiments on public datasets validate the effectiveness and efficiency of our method. Furthermore, our method runs at 30 fps at the testing time, thus could facilitate many real-time applications such as visual SLAM and human-robot interaction. Code is available at https://github.com/ svip-lab/PlanarReconstruction.

No abstract

No abstract

Our work addresses the problem of extracting and classifying tree species from mobile LiDAR data. The work includes tree preprocessing and tree classification. In tree preprocessing, voxel-based upward-growing filtering is proposed to remove ground points from the mobile LiDAR data, followed by a tree segmentation that extracts individual trees via Euclidean distance clustering and voxel-based normalized cut segmentation. In tree classification, first, a waveform representation is developed to model geometric structures of trees. Then, deep learning techniques are used to generate high-level feature abstractions of the trees' waveform representations. Quantitative analysis shows that our algorithm achieves an overall accuracy of 86.1% and a kappa coefficient of 0.8 in classifying urban tree species using mobile LiDAR data. Comparative experiments demonstrate that the uses of waveform representation and deep Boltzmann machines contribute to the improvement of classification accuracies of tree species.