The segmentation of airborne laser scanning (ALS) point clouds remains a challenge in remote sensing and photogrammetry. Deep learning methods, such as KPCONV, have proven effective on various datasets. However, the rigid convolutional kernel strategy of KPCONV limits its potential use for 3D object segmentation due to its uniform approach. To address this issue, we propose an Integrated Point Convolution (IPCONV) based on KPCONV, which utilizes two different convolution kernel point generation strategies, one cylindrical and one a spherical cone, for more efficient learning of point cloud data features. We propose a customizable Multi-Shape Neighborhood System (MSNS) to balance the relationship between these convolution kernel point generations. Experiments on the ISPRS benchmark dataset, LASDU dataset, and DFC2019 dataset demonstrate the validity of our method.

We introduce Attention Free Transformer (AFT), an efficient variant of Transformers that eliminates the need for dot product self attention. In an AFT layer, the key and value are first combined with a set of learned position biases, the result of which is multiplied with the query in an element-wise fashion. This new operation has a memory complexity linear w.r.t. both the context size and the dimension of features, making it compatible to both large input and model sizes. We also introduce AFT-local and AFT-conv, two model variants that take advantage of the idea of locality and spatial weight sharing while maintaining global connectivity. We conduct extensive experiments on two autoregressive modeling tasks (CIFAR10 and Enwik8) as well as an image recognition task (ImageNet-1K classification). We show that AFT demonstrates competitive performance on all the benchmarks, while providing excellent efficiency at the same time.

Deep Neural Network (DNN)-based point cloud semantic segmentation has presented significant breakthrough using large-scale labeled aerial laser point cloud datasets. However, annotating such large-scaled point clouds is time-consuming. Self-Supervised Learning (SSL) is a promising approach to this problem by pre-training a DNN model utilizing unlabeled samples followed by a fine-tuned downstream task involving very limited labels. The traditional contrastive learning for point clouds selects the hardest negative samples by solely relying on the distance between the embedded features derived from the learning process, potentially evolving some negative samples from the same classes to reduce the contrastive learning effectiveness. This work proposes a hard-negative sample-aware self-supervised contrastive learning algorithm to pre-train the model for semantic segmentation. We designed a k-means clustering-based Absolute Positive And Negative samples (AbsPAN) strategy to filter the possible false-negative samples. Experiments on two typical ALS benchmark datasets demonstrate that the proposed method is more appealing than supervised training schemes without pre-training. Especially when the labels are severely inadequate (10% of the ISPRS training set), the results obtained by the proposed HAVANA method still exceed 94% of the supervised paradigm performance with full training set.

2021 NCC/CATS/CSTCVS/STM expert consensus on perioperative immunotherapy for esophageal cancer

: There are few prospective studies of brain metastasis from esophageal squamous cell cancer (ESCC). The aim of this study was to investigate the necessity of the brain MRI/CT in the preoperative workup for patients with potentially resectable (cT1-4aN0-3) esophageal squamous cell cancer. Methods: This was a prospective cross-section clinical trial (ChiCTR1800020304). There were a total of 468 patients who were diagnosed ESCC from October 2018 to August 2020. Of these 468 patients, 385 patients with cT1-4aN0-3 who were potentially candidates for surgical resection were consecutively enrolled into the study. Preoperative brain MRI/CT was performed preoperatively. The treatment regimen could be changed if the brain metastasis was confirmed. The primary endpoint was the incidence rate of the treatment regimen being changed because of brain metastasis. Results: In all 385 patients, there are only 4 (1%) patients changed their treatment regimen because of brain metastasis proved by MRI/CT. The rate of positive brain MRI/CT findings is 1%. The MRI/CT diagnostic performance for brain metastasis was as follows: sensitivity, 100%; specificity, 100%; positive predictive value, 100%; negative predictive value, 100% and accuracy, 100%. There were no significant difference of bone metastasis among the Age, Sex, Tumor location and clinical stage. Conclusion: Preoperative brain MRI/CT might help identify brain metastases in patients with esophageal cancer, but we do not recommend the brain MRI/CT in the preoperative workup for patients with potentially resectable esophageal squamous cell cancer.

: 99mTc bone scintigraphy(BS) is still the most widely used method to evaluate bone metastasis in China. The aim of this study was to investigate the necessity of the 99mTc bone scintigraphy in the preoperative workup for patients with potentially resectable esophageal squamous cell cancer (ESCC,cT1-4aN0–3). Methods: This was a prospective cross-section clinical trial (ChiCTR1800020304). There were a total of 471 patients who were diagnosed ESCC in Thoracic Surgery Clinic from October 2018 to September 2020. Of these 471 patients, 385 patients with cT1-4aN0–3 who were potentially candidates for surgical resection were consecutively enrolled into the study. BS was performed preoperatively. The treatment strategy could be changed if the bone metastasis was confirmed. The primary endpoint was the incidence rate of the treatment regimen being changed because of bone metastasis. The secondary endpoint was the rate of positive BS findings. Results: In all 385 patients, there are only 2(0.5%) patients changed their treatment regimen because of bone metastasis proved by BS. The rate of positive BS findings is 1%. The number of patients with false positive and false negative was 2(0.5%) and 2 (0.5%), respectively. The BS diagnostic performance for bone metastasis was as follows: sensitivity, 50%; specificity, 99.5%; positive predictive value, 50%; negative predictive value, 99.5% and accuracy, 99.0%. There were no significant difference of bone metastasis among the Age, Sex, Tumor location and Clinical stage. Conclusion: 99mTc bone scintigraphy is unnecessary in the preoperative workup for patients with potentially resectable esophageal squamous cell cancer.