Indoor Scene Change Captioning Based on Multimodality Data.

This study proposes a framework for describing a scene change using natural language text based on indoor scene observations conducted before and after a scene change. The recognition of scene changes plays an essential role in a variety of real-world applications, such as scene anomaly detection. Most scene understanding research has focused on static scenes. Most existing scene change captioning methods detect scene changes from single-view RGB images, neglecting the underlying three-dimensional structures. Previous three-dimensional scene change captioning methods use simulated scenes consisting of geometry primitives, making it unsuitable for real-world applications. To solve these problems, we automatically generated large-scale indoor scene change caption datasets. We propose an end-to-end framework for describing scene changes from various input modalities, namely, RGB images, depth images, and point cloud data, which are available in most robot applications. We conducted experiments with various input modalities and models and evaluated model performance using datasets with various levels of complexity. Experimental results show that the models that combine RGB images and point cloud data as input achieve high performance in sentence generation and caption correctness and are robust for change type understanding for datasets with high complexity. The developed datasets and models contribute to the study of indoor scene change understanding.

Multi-View Visual Question Answering with Active Viewpoint Selection.

This paper proposes a framework that allows the observation of a scene iteratively to answer a given question about the scene. Conventional visual question answering (VQA) methods are designed to answer given questions based on single-view images. However, in real-world applications, such as human-robot interaction (HRI), in which camera angles and occluded scenes must be considered, answering questions based on single-view images might be difficult. Since HRI applications make it possible to observe a scene from multiple viewpoints, it is reasonable to discuss the VQA task in multi-view settings. In addition, because it is usually challenging to observe a scene from arbitrary viewpoints, we designed a framework that allows the observation of a scene actively until the necessary scene information to answer a given question is obtained. The proposed framework achieves comparable performance to a state-of-the-art method in question answering and simultaneously decreases the number of required observation viewpoints by a significant margin. Additionally, we found our framework plausibly learned to choose better viewpoints for answering questions, lowering the required number of camera movements. Moreover, we built a multi-view VQA dataset based on real images. The proposed framework shows high accuracy (94.01%) for the unseen real image dataset.

Temporal and Fine-Grained Pedestrian Action Recognition on Driving Recorder Database.

The paper presents an emerging issue of fine-grained pedestrian action recognition that induces an advanced pre-crush safety to estimate a pedestrian intention in advance. The fine-grained pedestrian actions include visually slight differences (e.g., walking straight and crossing), which are difficult to distinguish from each other. It is believed that the fine-grained action recognition induces a pedestrian intention estimation for a helpful advanced driver-assistance systems (ADAS). The following difficulties have been studied to achieve a fine-grained and accurate pedestrian action recognition: (i) In order to analyze the fine-grained motion of a pedestrian appearance in the vehicle-mounted drive recorder, a method to describe subtle change of motion characteristics occurring in a short time is necessary; (ii) even when the background moves greatly due to the driving of the vehicle, it is necessary to detect changes in subtle motion of the pedestrian; (iii) the collection of large-scale fine-grained actions is very difficult, and therefore a relatively small database should be focused. We find out how to learn an effective recognition model with only a small-scale database. Here, we have thoroughly evaluated several types of configurations to explore an effective approach in fine-grained pedestrian action recognition without a large-scale database. Moreover, two different datasets have been collected in order to raise the issue. Finally, our proposal attained 91.01% on National Traffic Science and Environment Laboratory database (NTSEL) and 53.23% on the near-miss driving recorder database (NDRDB). The paper has improved +8.28% and +6.53% from baseline two-stream fusion convnets.

[Update on the Use of Lower Extremity Peripheral Nerve Blocks and its Future].

Regional anesthesia for lower limb surgery not only provides satisfactory analgesia, but also improves the overall postoperative outcome with less postoperative nausea and vomiting by decreasing the opioid con- sumption, encouraging early postoperative mobility. Therefore, high-quality anesthesia and postoperative analgesia accelerate the rehabilitation process and shorten the hospital stay. In the past decade, ultra- sound-guided lower extremity peripheral nerve blocks have become popular in Japanese hospitals. This tech- nique enables the visualization of thee target nerve structures, controlles needle movement and the spread of injected local anesthetic solution in a real real-time manner, and this has been adapted not only for tradi- tional blocks, but also in "new approach" blocks such as adductor canal blocks, depending solely on ultra- sound images. In the decades to come, we hope to obtain more established evidence supporting the utility of ultra- sound-guided techniques for lower extremity nerve blocks based on high-quality clinical studies. These findings may support the development of sustained- release formulation local anesthetics and new devices or techniques in the future.

[Effect of tolvaptan on ascites due to malignancy].

Ascites accompanying a malignancy is often refractory to conventional treatment with saline diuretics, making it difficult to control. We administered a new diuretic, Tolvaptan, to 10 individuals with malignancy and heart failure accompanied by ascites, which was refractory to saline diuretics, and assessed its efficacy and adverse events. We observed a significant reduction in abdominal distension following 2 weeks of Tolvaptan administration. However, we also observed significant increases in serum potassium, urea nitrogen, and creatinine levels, but no serious adverse events. This suggests that Tolvaptan may also be effective as treatment for ascites.

Effects of oral rehydration therapy on gastric volume and pH in patients with preanesthetic H2 antagonist.

Recent preanesthetic fasting practice allows patients to take clear fluids up to 2 h before surgery without taking any antacid for the prophylaxis of aspiration pneumonia; this practice is defined as oral rehydration therapy (ORT). It has been reported that with ORT the gastric volume may be significantly lower than that with a standard fasting regimen, although in a standard fasting regimen without preanesthetic antacid, gastric pH and volume values could be critical for causing aspiration pneumonia. In this study we compared gastric fluid status in patients with ORT and those with a standard fasting regimen; patients in both groups received a preanesthetic H(2) antagonist. One hundred and four patients were randomly assigned to either the ORT or standard fasting group, and all were given roxatidine 75 mg 2 h before surgery. After the induction of anesthesia, the gastric contents were collected via a gastric tube to measure gastric volume and pH. Neither gastric volume nor pH differed between the groups (ORT 9.6 ± 8.2 ml and 5.6 ± 1.7, respectively, vs. standard fasting 8.5 ± 6.8 ml and 5.5 ± 1.7, respectively). These data suggest that ORT may not reduce gastric volume in patients receiving a preanesthetic H(2) antagonist.

[Complexities of the stress experienced by employees of the Fukushima nuclear plants].

Fukushima Daiichi Nuclear Power Plants suffered serious damage by the Great East Japan Earthquake and Tsunami. The employees of the plant worked very hard to stabilize the nuclear reactor and to prevent any secondary accidents. They were in one of the most severe situations in this disaster, but they were the people who hesitated most to request help for themselves. We started visiting the Fukushima Daini Plant office that was used as the frontline base for Daiichi Plant workers since July, 2011. These visits were held once or twice a month and we offered mental health support to the employees. We have completed interview with the total number of 339 plant workers by April, 2012. We offered several ways of mental support including clinical treatment, continuous counseling, or one time advice, depending on mental condition of each interviewee. Complexity of huge disaster and individuality of suffering from it were discussed in this article. Like local residents, many plant workers also experienced death/missing of family, loss of housing, refuge life, and dispersion of family. Furthermore, they have been suffering from various kinds of criticism and slander against Tokyo Electric Power Company. Many workers, even though they were not in management positions, seemed to have guilty conscience and sense of responsibility that forced them to stay in the risky working site. We could find some struggling coexistence of sense of guilt (as a causer of disaster) and sense of victim in their mind. It was suggested that continuous effort to listen and pay attention to their talk is important in order to support their mission to stabilize the power plant and to prevent them from over-stress and burnout.

CT findings of leukemic pulmonary infiltration with pathologic correlation.

The aim of this study was to demonstrate the characteristic CT findings of leukemic pulmonary infiltration based on the pathologic findings. The CT findings of 11 leukemic patients with leukemic pulmonary infiltration were compared with those of 22 leukemic patients with other diseases as a control group. Evaluated pulmonary parenchymal CT findings included thickening of bronchovascular bundles and interlobular septa, prominence of peripheral pulmonary arteries, ground-glass opacities, air-space consolidation, and nodules. The CT-pathologic correlations for leukemic infiltration were evaluated in 7 patients. Frequent parenchymal CT findings were thickening of bronchovascular bundles (81.8%), prominence of peripheral pulmonary arteries (81.8%), and non-lobular and non-segmental ground-glass opacities (90.9%). The first two findings were significantly more frequently observed in leukemic infiltration than in the control group, had good interobserver agreement, and corresponded pathologically to leukemic cell infiltration around the pulmonary arteries, bronchi, or bronchioles. Non-lobular and non-segmental ground-glass opacity corresponded to leukemic cell infiltration within alveolar spaces and septa adjacent to the pulmonary arteries or bronchi and also corresponded to hemorrhage, edema, or diffuse alveolar damage. Thickening of bronchovascular bundles and prominence of peripheral pulmonary arteries are CT findings suggestive for leukemic infiltration and correspond to peribronchovascular tumor extension.