scholarly journals A Robust Training Method for Pathological Cellular Detector via Spatial Loss Calibration

2021 ◽  
Vol 8 ◽  
Author(s):  
Hansheng Li ◽  
Yuxin Kang ◽  
Wentao Yang ◽  
Zhuoyue Wu ◽  
Xiaoshuang Shi ◽  
...  
Keyword(s):  
Gradient Analysis ◽  
Spatial Information ◽  
Accurate Diagnosis ◽  
Future Research ◽  
Training Method ◽  
Generalization Performance ◽  
Level Model ◽  
Model Training ◽  
Reliable Methods ◽  
Aided Diagnosis

Computer-aided diagnosis of pathological images usually requires detecting and examining all positive cells for accurate diagnosis. However, cellular datasets tend to be sparsely annotated due to the challenge of annotating all the cells. However, training detectors on sparse annotations may be misled by miscalculated losses, limiting the detection performance. Thus, efficient and reliable methods for training cellular detectors on sparse annotations are in higher demand than ever. In this study, we propose a training method that utilizes regression boxes' spatial information to conduct loss calibration to reduce the miscalculated loss. Extensive experimental results show that our method can significantly boost detectors' performance trained on datasets with varying degrees of sparse annotations. Even if 90% of the annotations are missing, the performance of our method is barely affected. Furthermore, we find that the middle layers of the detector are closely related to the generalization performance. More generally, this study could elucidate the link between layers and generalization performance, provide enlightenment for future research, such as designing and applying constraint rules to specific layers according to gradient analysis to achieve “scalpel-level” model training.

Organizational Performance Research in India

2019 ◽  
pp. 1-93
Author(s):  
Shailendra Singh ◽  
Vishal Gupta
Keyword(s):  
Organizational Performance ◽  
Critical Analysis ◽  
Future Research ◽  
Individual Level ◽  
Process Framework ◽  
Level Model ◽  
Multi Level ◽  
Performance Research ◽  
Working Hypotheses ◽  
Level Group

This chapter presents a review of research in the area of organizational performance in India during the last decade, which has become a challenge for organizations and management researchers. The chapter begins with a critical analysis of the nature of performance measurement and associated challenges. Next, it summarizes the research that has linked individual-level, group-level, and organization-level variables to organizational performance. The theoretical and conceptual contributions, limitations, gaps, and the scope of future research in the field are presented by the contributors. Finally, a multi-level model has been presented that provides a process framework, which links antecedent variables to organizational performance. The framework provides a set of working hypotheses for future organizational performance research in the Indian context.

scholarly journals Provider confidence in the telemedicine spine evaluation: results from a global study

2020 ◽  
Author(s):  
Francis Lovecchio ◽  
Grant J. Riew ◽  
Dino Samartzis ◽  
Philip K. Louie ◽  
Niccole Germscheid ◽  
...  
Keyword(s):  
Treatment Plan ◽  
Accurate Diagnosis ◽  
Physical Exam ◽  
Asia Pacific ◽  
Future Research ◽  
Appropriate Use Criteria ◽  
Appropriate Use ◽  
Delphi Approach ◽  
Telemedicine Technology ◽  
Spine Surgeon

Abstract Purpose To utilize data from a global spine surgeon survey to elucidate (1) overall confidence in the telemedicine evaluation and (2) determinants of provider confidence. Methods Members of AO Spine International were sent a survey encompassing participant’s experience with, perception of, and comparison of telemedicine to in-person visits. The survey was designed through a Delphi approach, with four rounds of question review by the multi-disciplinary authors. Data were stratified by provider age, experience, telemedicine platform, trust in telemedicine, and specialty. Results Four hundred and eighty-five surgeons participated in the survey. The global effort included respondents from Africa (19.9%), Asia Pacific (19.7%), Europe (24.3%), North America (9.4%), and South America (26.6%). Providers felt that physical exam-based tasks (e.g., provocative testing, assessing neurologic deficits/myelopathy, etc.) were inferior to in-person exams, while communication-based aspects (e.g., history taking, imaging review, etc.) were equivalent. Participants who performed greater than 50 visits were more likely to believe telemedicine was at least equivalent to in-person visits in the ability to make an accurate diagnosis (OR 2.37, 95% C.I. 1.03–5.43). Compared to in-person encounters, video (versus phone only) visits were associated with increased confidence in the ability of telemedicine to formulate and communicate a treatment plan (OR 3.88, 95% C.I. 1.71–8.84). Conclusion Spine surgeons are confident in the ability of telemedicine to communicate with patients, but are concerned about its capacity to accurately make physical exam-based diagnoses. Future research should concentrate on standardizing the remote examination and the development of appropriate use criteria in order to increase provider confidence in telemedicine technology.

scholarly journals The claustrum: Considerations regarding its anatomy, functions and a programme for research

2017 ◽  
Vol 1 ◽  
pp. 239821281771896 ◽  
Author(s):  
Christopher M. Dillingham ◽  
Maciej M. Jankowski ◽  
Ruchi Chandra ◽  
Bethany E. Frost ◽  
Shane M. O’Mara
Keyword(s):  
Decision Making ◽  
Information Processing ◽  
Brain Region ◽  
Spatial Information ◽  
Mammalian Brain ◽  
Future Research ◽  
Subcortical Structures ◽  
Present Evidence ◽  
Open Questions ◽  
Extensive Range

The claustrum is a highly conserved but enigmatic structure, with connections to the entire cortical mantle, as well as to an extended and extensive range of heterogeneous subcortical structures. Indeed, the human claustrum is thought to have the highest number of connections per millimetre cubed of any other brain region. While there have been relatively few functional investigations of the claustrum, many theoretical suggestions have been put forward, including speculation that it plays a key role in the generation of consciousness in the mammalian brain. Other claims have been more circumspect, suggesting that the claustrum has a particular role in, for example, orchestrating cortical activity, spatial information processing or decision making. Here, we selectively review certain key recent anatomical, electrophysiological and behavioural experimental advances in claustral research and present evidence that calls for a reassessment of its anatomical boundaries in the rodent. We conclude with some open questions for future research.

154 Computer-Aided Diagnosis Based on Endocytoscopy With Narrow-Band Imaging Allows Accurate Diagnosis of Diminutive Colorectal Lesions

2017 ◽  
Vol 85 (5) ◽  
pp. AB57
Author(s):  
Masashi Misawa ◽  
Shinei Kudo ◽  
Yuichi Mori ◽  
Kenichi Takeda ◽  
Shinichi Kataoka ◽  
...  
Keyword(s):  
Narrow Band ◽  
Narrow Band Imaging ◽  
Computer Aided Diagnosis ◽  
Accurate Diagnosis ◽  
Computer Aided ◽  
Aided Diagnosis

scholarly journals Computational approaches to motor learning by imitation

2003 ◽  
Vol 358 (1431) ◽  
pp. 537-547 ◽  
Author(s):  
Stefan Schaal ◽  
Auke Ijspeert ◽  
Aude Billard
Keyword(s):  
Motor Control ◽  
Degrees Of Freedom ◽  
Spatial Information ◽  
Point Of View ◽  
Performance Criteria ◽  
Future Research ◽  
Computational Point ◽  
Pose Tracking ◽  
Active Research ◽  
Learning By Imitation

Movement imitation requires a complex set of mechanisms that map an observed movement of a teacher onto one's own movement apparatus. Relevant problems include movement recognition, pose estimation, pose tracking, body correspondence, coordinate transformation from external to egocentric space, matching of observed against previously learned movement, resolution of redundant degrees–of–freedom that are unconstrained by the observation, suitable movement representations for imitation, modularization of motor control, etc. All of these topics by themselves are active research problems in computational and neurobiological sciences, such that their combination into a complete imitation system remains a daunting undertaking—indeed, one could argue that we need to understand the complete perception–action loop. As a strategy to untangle the complexity of imitation, this paper will examine imitation purely from a computational point of view, i.e. we will review statistical and mathematical approaches that have been suggested for tackling parts of the imitation problem, and discuss their merits, disadvantages and underlying principles. Given the focus on action recognition of other contributions in this special issue, this paper will primarily emphasize the motor side of imitation, assuming that a perceptual system has already identified important features of a demonstrated movement and created their corresponding spatial information. Based on the formalization of motor control in terms of control policies and their associated performance criteria, useful taxonomies of imitation learning can be generated that clarify different approaches and future research directions.

scholarly journals Non-Destructive Monitoring of Maize Nitrogen Concentration Using a Hyperspectral LiDAR: An Evaluation from Leaf-Level to Plant-Level

2021 ◽  
Vol 13 (24) ◽  
pp. 5025
Author(s):  
Kaiyi Bi ◽  
Zheng Niu ◽  
Shunfu Xiao ◽  
Jie Bai ◽  
Gang Sun ◽  
...  
Keyword(s):  
Precision Agriculture ◽  
Spatial Information ◽  
Nitrogen Concentration ◽  
Point Clouds ◽  
Future Research ◽  
Least Squares Regression ◽  
N Concentration ◽  
Leaf Level ◽  
Plant Level ◽  
Remote Sensing Techniques

Advanced remote sensing techniques for estimating crop nitrogen (N) are crucial for optimizing N fertilizer management. Hyperspectral LiDAR (HSL) data, with both spectral and spatial information of the targets, can extract more plant properties than traditional LiDAR and hyperspectral imaging systems. In this study, we tested the ability of HSL in terms of estimating maize N concentration at the leaf-level by using spectral indices and partial least squares regression (PLSR) methods. Subsequently, the N estimation was scaled up to the plant-level based on HSL point clouds. Biomass, extracted with structural proxies, was utilized to exhibit its supplemental effect on N concentration. The results show that HSL has the ability to extract N concentrations at both the leaf-level and the canopy-level, and PLSR showed better performance (R2 > 0.6) than the single spectral index (R2 > 0.4). In comparison to the stem height and maximum canopy width, the plant height had the strongest ability (R2 = 0.88) to estimate biomass. Future research should utilize larger datasets to test the viability of using HSL to monitor the N concentration of crops, which is beneficial for precision agriculture.

scholarly journals Re-Identification in Urban Scenarios: A Review of Tools and Methods

2021 ◽  
Vol 11 (22) ◽  
pp. 10809
Author(s):  
Hugo S. Oliveira ◽  
José J. M. Machado ◽  
João Manuel R. S. Tavares
Keyword(s):  
Neural Networks ◽  
High Performance ◽  
Object Identification ◽  
Future Research ◽  
Great Effort ◽  
Data Limitations ◽  
Video Feed ◽  
Adversarial Training ◽  
Future Research Directions ◽  
Model Training

With the widespread use of surveillance image cameras and enhanced awareness of public security, objects, and persons Re-Identification (ReID), the task of recognizing objects in non-overlapping camera networks has attracted particular attention in computer vision and pattern recognition communities. Given an image or video of an object-of-interest (query), object identification aims to identify the object from images or video feed taken from different cameras. After many years of great effort, object ReID remains a notably challenging task. The main reason is that an object’s appearance may dramatically change across camera views due to significant variations in illumination, poses or viewpoints, or even cluttered backgrounds. With the advent of Deep Neural Networks (DNN), there have been many proposals for different network architectures achieving high-performance levels. With the aim of identifying the most promising methods for ReID for future robust implementations, a review study is presented, mainly focusing on the person and multi-object ReID and auxiliary methods for image enhancement. Such methods are crucial for robust object ReID, while highlighting limitations of the identified methods. This is a very active field, evidenced by the dates of the publications found. However, most works use data from very different datasets and genres, which presents an obstacle to wide generalized DNN model training and usage. Although the model’s performance has achieved satisfactory results on particular datasets, a particular trend was observed in the use of 3D Convolutional Neural Networks (CNN), attention mechanisms to capture object-relevant features, and generative adversarial training to overcome data limitations. However, there is still room for improvement, namely in using images from urban scenarios among anonymized images to comply with public privacy legislation. The main challenges that remain in the ReID field, and prospects for future research directions towards ReID in dense urban scenarios, are also discussed.

scholarly journals A REVIEW OF RECENT RESEARCH IN INDOOR MODELLING & MAPPING

2016 ◽  
Vol XLI-B4 ◽  
pp. 289-294 ◽  
Author(s):  
M. Gunduz ◽  
U. Isikdag ◽  
M. Basaraner
Keyword(s):  
Information Acquisition ◽  
Information Science ◽  
Spatial Information ◽  
Ethical Issues ◽  
Urban Spaces ◽  
Future Research ◽  
Scientific Publications ◽  
Research Directions ◽  
Spatial Information Science ◽  
Future Research Directions

Indoor modeling and mapping has been an active area of research in last 20 years in order to tackle the problems related to positioning and tracking of people and objects indoors, and provides many opportunities for several domains ranging from emergency response to logistics in micro urban spaces. The outputs of recent research in the field have been presented in several scientific publications and events primarily related to spatial information science and technology. This paper summarizes the outputs of last 10 years of research on indoor modeling and mapping within a proper classification which covers 7 areas, i.e. Information Acquisition by Sensors, Model Definition, Model Integration, Indoor Positioning and LBS, Routing & Navigation Methods, Augmented and Virtual Reality Applications, and Ethical Issues. Finally, the paper outlines the current and future research directions and concluding remarks.

scholarly journals Deepfake-Image Anti-Forensics with Adversarial Examples Attacks

2021 ◽  
Vol 13 (11) ◽  
pp. 288
Author(s):  
Li Fan ◽  
Wei Li ◽  
Xiaohui Cui
Keyword(s):  
State Of The Art ◽  
Poisson Noise ◽  
Future Research ◽  
Detection Accuracy ◽  
Generalization Performance ◽  
Image Forensic ◽  
Adversarial Examples ◽  
Adversarial Example ◽  
The Impact

Many deepfake-image forensic detectors have been proposed and improved due to the development of synthetic techniques. However, recent studies show that most of these detectors are not immune to adversarial example attacks. Therefore, understanding the impact of adversarial examples on their performance is an important step towards improving deepfake-image detectors. This study developed an anti-forensics case study of two popular general deepfake detectors based on their accuracy and generalization. Herein, we propose the Poisson noise DeepFool (PNDF), an improved iterative adversarial examples generation method. This method can simply and effectively attack forensics detectors by adding perturbations to images in different directions. Our attacks can reduce its AUC from 0.9999 to 0.0331, and the detection accuracy of deepfake images from 0.9997 to 0.0731. Compared with state-of-the-art studies, our work provides an important defense direction for future research on deepfake-image detectors, by focusing on the generalization performance of detectors and their resistance to adversarial example attacks.

Sign in / Sign up

Export Citation Format

Share Document