265 research outputs found
IFI and ISI premitigation for block-code-modulated noncoherent UWB impulse radio: A code optimization approach
No full textCodeword matching and signal aggregation (CMSA) is a recently proposed low-complexity noncoherent receiver for block code modulated UWB Impulse Radio (UWB-IR) systems. As the frame/symbol duration is shortened to boost data rate, inter-frame interference (IFI) or inter-symbol interference (ISI) occurs and degrades detection performance of CMSA. In this paper, an effective IFI/ISI pre-mitigation scheme is proposed for CMSA by means of a code optimization approach. By employing a tailored interference model that highlights the codeword properties, the system performance in the presence of moderate IFI/ISI is evaluated and an average collected channel gain (CCG) is introduced as the metric for code optimization. With the primary focus on binary modulation, two IFI/ISI-robust code properties are generalized as Shifted-Orthogonality and Shifted-Repetition. Based on these properties, the optimal code is constructed. It is observed that, when the optimal code occurs, the leaked signal energy or the interference can be partially used to enhance the detection performance of CMSA in the presence of IFI/ISI. Unlike most of the existing IFI/ISI mitigation schemes for noncoherent UWB-IR that focus mainly on signal processing after the nonlinear detector, the optimized code is exploited to aggregate leaked signal energy along with the linear pre-detection operation already involved in CMSA receiver. Both analysis and simulation show that a distinct performance improvement is achieved
Effect of the Number, Size, and Location of Cannulated Screws on the Incidence of Avascular Necrosis of the Femoral Head in Pediatric Femoral Neck Fractures: A Review of 153 Cases
No full textBackground: The correlation between the number, size, and location of cannulated screws and the incidence of avascular necrosis (AVN) in children with femoral neck fractures treated surgically is uncertain. Methods: We retrospectively reviewed 153 children (mean age: 10.6±3.7 y) with femoral neck fractures treated by internal fixation with 2 (n=112) or 3 (n=41) cannulated screws. The severity of initial displacement was divided into incomplete (type I) and complete (type II, angulation <50 degrees; type III, angulation >50 degrees) fractures. The diameter of the screw was measured and recorded as a percentage of the femoral neck width. The distance (D) between the mid-point of each screw at the base (B) of the femoral neck and at the tip (T) of each screw and the superior and anterior cortices of the femoral neck, respectively, were measured on anteroposterior (AP) and lateral (L) radiographs. Values were expressed as the ratio between the measured distance and the width of the femoral neck (BDAP%, TDAP%, BDL%, and TDL%). The correlation between the number, size, and location of the screws and AVN was analyzed. Results: Patients with type II of initial displacement treated with 2 cannulated screws had a lower AVN rate (21.4%) than those treated with 3 screws (44.8%) (P=0.027). Screw diameter (19%) in patients with AVN was larger than (17%) in patients without AVN (P<0.001); patients with AVN had a lower BDAP% (48.6%) than those without AVN (56.4%) (P<0.001). Screw size and BDAP% were risk factors for AVN (P<0.05). Further, screw diameter >16.5% and BDAP% <51.6% of the femoral neck width were the cutoff values for an increased AVN rate (P<0.05). Conclusions: Patients treated with 2 cannulated screws showed a lower rate of AVN than patients treated with 3 screws. Screws of larger size and screws closer to the piriformis fossa on AP radiographs increased the risk of AVN in children with femoral neck fractures treated surgically. Level of Evidence: Level III
Initial displacement as a risk factor for avascular necrosis of the femoral head in pediatric femoral neck fractures: a review of one hundred eight cases
No full textPurpose: To evaluate the correlation between avascular necrosis (AVN) and the amount (severity) and direction (translation and angulation) of initial displacement of pediatric femoral neck fractures. Methods: We retrospectively reviewed 108 pediatric patients (mean age 10.3 ± 4.1 years) with femoral neck fractures. The amount of initial translation (T) and angulation (A) was measured on anteroposterior (AP; TAP% and AAP) and lateral (TL% and AL) radiographs. The direction of translation was determined on AP (medial or lateral) and lateral radiographs (anterior or posterior). Furthermore, the presence of a comminuted medial cortex on the AP pelvis radiograph was also recorded. Logistic regression analysis, receiver operating characteristic (ROC) curve analysis, student’s t tests, and chi-square tests were used to evaluate the correlation between AVN and the severity and direction of displacement. Results: Twenty-eight out of 108 hips (25.9%) developed AVN of the femoral head. Logistical regression analysis indicated that TAP%, TL%, AAP, and AL were risk factors for AVN (P < 0.05). The analysis of ROC curves found that TAP% over 37.4% and TL% over 29% were the cut-off values for an increased incidence of AVN; similarly, AAP over 8° and AL over 18.6° were the cut-off values for an increased incidence of AVN. The amount of initial translation is a better predictor of AVN than angulation is; fractures with posterior translation (P = 0.002) and/or medial comminution had a significantly higher incidence of AVN (P = 0.005). The mean diagnostic accuracy of translation (74–75%) was significantly higher than that of angulation (65–66%). Conclusions: Displacement severity and direction are important radiological parameters to be assessed in children with femoral neck fractures. Initial translation better predicts AVN than angulation does. Posterior translation and medial comminution are associated with an increased risk of AVN
Risk factors for the development of avascular necrosis after femoral neck fractures in children
No full textAims The aim of this study was to clarify the factors that predict the development of avascular necrosis (AVN) of the femoral head in children with a fracture of the femoral neck. Patients and Methods We retrospectively reviewed 239 children with a mean age of 10.0 years (sd 3.9) who underwent surgical treatment for a femoral neck fracture. Risk factors were recorded, including age, sex, laterality, mechanism of injury, initial displacement, the type of fracture, the time to reduction, and the method and quality of reduction. AVN of the femoral head was assessed on radiographs. Logistic regression analysis was used to evaluate the independent risk factors for AVN. Chi-squared tests and Student’s t-tests were used for subgroup analyses to determine the risk factors for AVN. Results We found that age (p = 0.006) and initial displacement (p = 0.001) were significant independent risk factors. Receiver operating characteristic (ROC) curve analysis indicated that 12 years of age was the cut-off for increasing the rate of AVN. Severe initial displacement (p = 0.021) and poor quality of reduction (p = 0.022) significantly increased the rate of AVN in patients aged 12 years or greater, while in those aged less than 12 years, the rate of AVN significantly increased only with initial displacement (p = 0.048). A poor reduction significantly increased the rate of AVN in patients treated by closed reduction (p = 0.026); screw and plate fixation was preferable to cannulated screw or Kirschner wire (K-wire) fixation for decreasing the rate of AVN in patients treated by open reduction (p = 0.034). Conclusion The rate of AVN increases with age, especially in patients aged 12 years or greater, and with the severity of displacement. In patients treated by closed reduction, anatomical reduction helps to decrease the rate of AVN, while in those treated by open reduction, screw and plate fixation was preferable to fixation using cannulated screws or K-wires
Towards intelligent surgery: dynamic surgical video analysis with deep learning
No full textPh.D.With advancements in medicine and information technologies, the operating room has undergone tremendous transformations evolving into a highly complicated and technologically rich environment. Such transformations further innovate the surgery procedure and greatly enhance the patient safety. To better tackle this new scenario, the computer-assisted and robotic-assisted systems have been gradually developed to provide surgeons with the detailed contextual support as well as the dexterous and precise manipulation capability. Automatic surgical video analysis has become a crucial component when developing these systems, as videos can be routinely collected without using auxiliary devices or disrupting the workflow. It enables systems to have various functionalities, such as decision making support, real-time warning generation and surgical skill assessment. Meanwhile, the deep learning techniques have achieved distinguished successes in analyzing vision-based data by encoding highly discriminative representations.This thesis presents a series of interdisciplinary researches which bring deep learning techniques to dynamic surgical video analysis. The central question in using deep learning to analyze surgical video is how to well complement the visual information and sequential dynamic. This thesis roots in this direction, for encoding powerful spatio-temporal features by developing advanced network architectures and effective learning strategies, which can be applied to a broad variety of essential tasks towards intelligent surgery.We explore a recurrent convolutional network for surgical workflow recognition, which seamlessly integrates a convolutional neural network and a recurrent neural network to take full advantages of the complementary information of visual and temporal features. We effectively train the entire architecture in an end-to-end manner so that both representations can be jointly optimized in the learning process. Following, we develop a more complicated multi-task learning paradigm for joint surgical tool presence detection and phase recognition, which aims to exploit the relatedness to simultaneously boost the performance of both tasks. A novel correlation loss is designed to provide additional regularization inspired from domain knowledge, by minimizing the divergence of probability predictions for the two tasks. The temporal information modeled in the phase branch can benefit the tool detection via multi-task learning.We further target at more fine-grained analysis with pixel-level segmentation tasks. We present a novel framework to incorporate a derived temporal prior to an attention pyramid network for accurate instrument segmentation. Our inferred prior can provide reliable indication of instrument location and shape, which is propagated via inter-frame motion flow. This prior is injected to the middle of an encoder-decoder segmentation network as an initialization of a pyramid of attention modules, to explicitly guide output from coarse to fine. Moreover, we propose to segment surgical scene by deliberately designing networks towards two valuable clues of this task, i.e. sequential dynamics and multi-scale visual features. We develop a sequence to image generator with future frame prediction task to encourage temporal consistent predictions. Based on this, we propose the integration modules to let the generator gradually augment the segmenter in multiple levels. With the joint learning, these two tailored networks well complement to form the discriminative spatio-temporal features for better scene segmentation.隨著醫學以及資訊科技的發展，手術室經歷了巨大的變革，逐漸演變為一個高度複雜且技術豐富的治療環境。這進一步革新了手術流程並且極大地提高了病人診治的安全性。為了更好地應對這種新情況，計算機及機器人輔助系統已逐步發展起來，為外科醫生提供詳細的環境解析以及精確且靈活的操控可能性。由於無需使用輔助設備且不會擾亂手術流程即可收集視頻，因此在開發這些系統時，自動手術視頻分析成為了至關重要的部分。與此同時，深度學習技術依靠著其高度辨識能力在視覺數據分析中取得了顯著的成功。本論文提出一系列跨學科研究，應用深度學習技術進行動態的手術視頻分析。其核心問題是如何更好地使視覺信息和時間信息相互補足與完善。本文正是基於此方向，提出了新型網絡結構以及有效的學習策略來編碼時空特征，並且能夠應用在多種重要任務中，以達到最終的手術智能化的目的。首先本文提出循環卷積網絡用於手術流程識別。通過無縫連接卷積神經網絡與循環神經網絡，來充分利用視覺與時間的互補信息。整個網路同時訓練使得兩種特征共同優化。接下來，本文提出了更複雜的多任務學習框架，借力于任務間的高度關聯性，來同步提升手術工具檢測以及手術流程識別的準確率。我們設計了一種相關性損失函數，通過最小化多任務間的預測分歧來進一步規範網絡。通過多任務學習，用於流程識別的時間信息同樣有益于工具檢測任務。更進一步，本文探索了細粒度的像素級分割任務。我們通過提供網絡一個時間先驗來實現手術工具分割。這個先驗知識通過幀之間的運動流推導出來，能夠準確的指示工具在當前幀的位置與形狀，并通過初始化網絡關注區域來進行顯示化指導。此外，本文研究了整個手術場景的分割。我們提出了序列至圖像生成器，通過預測未來幀的形式來捕捉時間信息。在此基礎上，我們設計了融合模塊，使得生成器可以多層次地調整與增強分割器，從而實現更精準的分割。Jin, Yueming.Thesis Ph.D. Chinese University of Hong Kong 2019.Includes bibliographical references (leaves 133-148).Abstracts also in Chinese.Title from PDF title page (viewed on 25, November, 2020)
Consumers’ Attitudes towards Surcharges on Distributed Renewable Energy Generation and Energy Efficiency Programs
No full textabstract: Increasing penetration of energy efficiency programs and distributed renewable energy generation has imposed significant challenges for utilities to recoup their large upfront costs. There is a heated debate on what surcharges should be implemented to help the utilities recover their fixed costs; however, very few studies focus on consumers’ attitudes regarding this topic. This study surveys about 190 residential consumers throughout the United States in November 2015, investigating their preferences and attitudes towards extra demand charges and volumetric energy price increases. We apply probit models and regress consumers’ attitudes on selected socio-demographic and behavioral variables. The results indicate the homeowners are more likely to prefer demand charges when compared to renters. The demographic and behavioral factors impact consumers’ perception of bill savings from energy efficiency programs or solar panel installation and also influence how consumers perceive the fairness of utilities recovering revenue losses by increasing volumetric energy price. In this paper, we demonstrate there is preference heterogeneity among consumers and that policy makers should be aware of such preference heterogeneity and apply policy targeting based on the identified demographics and behavioral factors impacting consumer preferences.The final version of this article, as published in Sustainability, can be viewed online at: http://www.mdpi.com/2071-1050/9/8/147
Peer Effects and Voluntary Green Building Certification
No full textabstract: Empirical evidence is provided to show that peer effects have statistically significant and positive impacts on the diffusion of green building certificates. Application and approval records of green certificates by commercial buildings in NY and AZ are used. The challenge of self-selection is addressed by the usage of fixed effects and the challenge of reflection is addressed by the time lag delay between a building’s application and its approval. Empirical results show that an additional approved LEED certificate within a zip code will increase the probability of a commercial building in the same zip code to apply for a LEED certificate by 3–4 percentage points; an additional approved Energy Star certificate within a zip code will increase the probability of a commercial building in the same zip code to apply for an Energy Star certificate by 1–2 percentage points
Multi-center Ovarian Tumor Classification Using Hierarchical Transformer-Based Multiple-Instance Learning
No full textMalignant ovarian tumors (OTs) are a leading cause of gynecological cancer deaths, and often remain asymptomatic until advanced stages, making early and accurate diagnosis crucial for effective treatment and good patient outcome. Current diagnostic methods often fall short due to the heterogeneous nature of OTs and the complexities in distinguishing benign from malignant forms. To overcome these limitations, this study proposes a novel framework leveraging transformer-based multiple-instance learning (MIL) and hierarchical self-supervised pre-training. To validate the model, a comprehensive multi-center dataset has been compiled, encompassing diverse patient demographics and imaging protocols. Benchmarking against conventional radiomics methods and other deep learning approaches, the hierarchical MIL model demonstrates superior performance with a median AUROC of 0.84 and high recall of 0.91. These results highlight significant improvements in sensitivity, essential for minimizing false negatives in clinical settings. The performed study emphasizes the importance of multi-center validation and external dataset testing to ensure generalization of the proposed model and obtain a higher robustness. The encountered complexity of multi-center data is found significant, since various clinical factors play an influential role. This makes baseline comparisons virtually impossible and the need for more multi-center research increasingly compelling and encouraging
Optimizing Multi-expert Consensus for Classification and Precise Localization of Barrett's Neoplasia
No full textRecognition of early neoplasia in Barrett’s Esophagus (BE) is challenging, despite advances in endoscopic technology. Even with correct identification, the subtle nature of lesions leads to significant inter-observer variability in placing targeted biopsy markers and delineation of lesions. Computer-Aided Detection (CADe) systems may assist endoscopists, however, compliance of endoscopists with CADe is often suboptimal, reducing joint performance below CADe stand-alone performance. Improved localization performance of CADe could enhance compliance. These systems often use fused consensus ground-truths (GT), which may not capture subtle neoplasia gradations, affecting classification and localization. This study evaluates five consensus GT strategies from multi-expert segmentation labels and four loss functions for their impact on classification and localization performance. The dataset includes 7,995 non-dysplastic BE images (1,256 patients) and 2,947 neoplastic images (823 patients), with each neoplastic image annotated by two experts. Classification, localization for true positives, and combined detection performance are assessed and compared with 14 independent Barrett’s experts. Results show that using multiple consensus GT masks with a compound Binary Cross-Entropy and Dice loss achieves the best classification sensitivity and near-expert level localization, making it the most effective training strategy. The code is made publicly available at: https://github.com/BONS-AI-VCA-AMC/BE-CADe-GT
- …
