Targeting Task Performance Research Articles

A Remote Sensing Image Quality Interpretation Scale Characterization Method Based on the TTP Criterion

Accurate grading of remote sensing image interpretation is crucial for improving image classification and screening efficiency. Through extensive research, the General Image Quality Equation (GIQE) based on the National Imagery Interpretability Rating Scale (NIIRS) has been developed. However, poor reliability and low accuracy issues remain due to the failure to consider human visual characteristics. This paper introduces the Target Task Performance (TTP) criterion as a key parameter to reflect the cascading degradation factors of human visual perception characteristics and system imaging links, which improves the reliability of the model. A New Optimized Remote Sensing Image Quality Equation (NORSIQE), which effectively predicted the interpretability of image information, is constructed. Using 200 sets of test data, the quantitative relationship between key parameters (GSD, TTP, and SNR) of the NORSIQE and the subjective NIIRS level is obtained by a least squares regression fit, and the determination coefficient of the model is as high as 0.916. The model is evaluated for accuracy using 120 sets of validation data, showing an 87% improvement compared to the GIQE4. This method provides theoretical support for the development of new methods for remote sensing image quality evaluation and the design of payloads for remote sensing imaging systems.

Classification of tumor from computed tomography images: A brain-inspired multisource transfer learning under probability distribution adaptation.

Preoperative diagnosis of gastric cancer and primary gastric lymphoma is challenging and has important clinical significance. Inspired by the inductive reasoning learning of the human brain, transfer learning can improve diagnosis performance of target task by utilizing the knowledge learned from the other domains (source domain). However, most studies focus on single-source transfer learning and may lead to model performance degradation when a large domain shift exists between the single-source domain and target domain. By simulating the multi-modal information learning and transfer mechanism of human brain, this study designed a multisource transfer learning feature extraction and classification framework, which can enhance the prediction performance of the target model by using multisource medical data (domain). First, this manuscript designs a feature extraction network that takes the maximum mean difference based on the Wasserstein distance as an adaptive measure of probability distribution and extracts the domain-specific invariant representations between source and target domain data. Then, aiming at the random generation of parameters bringing uncertainties to prediction accuracy and generalization ability of extreme learning machine network, the 1-norm regularization is used to implement sparse constraints of the output weight matrix and improve the robustness of the model. Finally, some experiments are carried out on the data of two medical centers. The experimental results show that the area under curves (AUCs) of the method are 0.958 and 0.929 in the two validation cohorts, respectively. The method in this manuscript can provide doctors with a better diagnostic reference, which has certain practical significance.

Generalization in transfer learning: robust control of robot locomotion

AbstractIn this paper, we propose a set of robust training methods for deep reinforcement learning to transfer learning acquired in one control task to a set of previously unseen control tasks. We improve generalization in commonly used transfer learning benchmarks by a novel sample elimination technique, early stopping, and maximum entropy adversarial reinforcement learning. To generate robust policies, we use sample elimination during training via a method we call strict clipping. We apply early stopping, a method previously used in supervised learning, to deep reinforcement learning. Subsequently, we introduce maximum entropy adversarial reinforcement learning to increase the domain randomization during training for a better target task performance. Finally, we evaluate the robustness of these methods compared to previous work on simulated robots in target environments where the gravity, the morphology of the robot, and the tangential friction coefficient of the environment are altered.

Hands-on Simulation Workshop for Obstetric Ultrasound-Guided Invasive Procedures.

IntroductionDue to the introduction of cell-free DNA genetic testing options, the number of clinical ultrasound procedures has greatly diminished in recent years. With fewer real-life ultrasound-guided procedures being performed, it is difficult for OB/GYN and maternal-fetal medicine (MFM) trainees to achieve competency in doing them. Simulation can be utilized to address this issue and supplement a learner's real-life training.MethodsWe developed a simulation workshop incorporating previously described ultrasound guidance task trainers and simulators of amniocentesis and chorionic villus sampling. The workshop had three parts: needle guidance basics, targeting task performance, and procedure-specific simulation. A form of this workshop has been held at the annual meeting of the Society for Maternal-Fetal Medicine since 2015 and as a regional course for MFM fellows since 2017. During the 2019 and 2020 courses, participants completed Likert-scale surveys evaluating the course.ResultsSince the workshops began in 2015, approximately 300 people have participated. In 2019–2020, 41 MFM attending physicians, 136 MFM fellows, and three OB/GYN residents took our course and completed a postcourse survey. Participants rated the course highly and thought it was highly effective.DiscussionWe created an introductory simulation workshop for obstetric ultrasound-guided invasive procedures that participants rated highly and thought was very effective. Objective clinical assessment of skill improvement after completion of this course is needed to verify its true impact. Repeated exposure to this introductory simulation and creation of more challenging workshops are needed to achieve a sustained high level of procedural skill.

Image quality assessment by overlapping task-specific and task-agnostic measures: application to prostate multiparametric MR images for cancer segmentation

Image quality assessment (IQA) in medical imaging can be used to ensure that downstream clinical tasks can be reliably performed. Quantifying the impact of an image on the specific target tasks, also named as task amenability, is needed. A <em>task-specific</em> IQA has recently been proposed to learn an image-amenability-predicting controller simultaneously with a target task predictor. This allows for the trained IQA controller to measure the impact an image has on the target task performance, when this task is performed using the predictor, e.g. segmentation and classification neural networks in modern clinical applications. In this work, we propose an extension to this task-specific IQA approach, by adding a <em>task-agnostic</em> IQA based on auto-encoding as the target task. Analysing the intersection between low-quality images, deemed by both the task-specific and task-agnostic IQA, may help to differentiate the underpinning factors that caused the poor target task performance. For example, common imaging artefacts may not adversely affect the target task, which would lead to a low task-agnostic quality and a high task-specific quality, whilst individual cases considered clinically challenging, which can not be improved by better imaging equipment or protocols, is likely to result in a high task-agnostic quality but a low task-specific quality. We first describe a flexible reward shaping strategy which allows for the adjustment of weighting between task-agnostic and task-specific quality scoring. Furthermore, we evaluate the proposed reinforcement learning algorithm, using a clinically challenging target task of prostate tumour segmentation on multiparametric magnetic resonance (mpMR) images. Based on experimental results using mpMR images from 850 patients, it was found that <em>a</em>) The task-agnostic IQA may identify artefacts, but with limited impact on the accuracy of cancer segmentation networks. A Dice score of 0.367±0.017 was obtained after rejecting 10% of low quality images, compared to 0.354±0.016 from a non-selective baseline; <em>b</em>} The task-specific IQA alone improved the performance to 0.415±0.020, at the same rejection ratio. However, this system indeed rejected both images that impact task performance due to imaging defects and due to being clinically challenging; and <em>c</em>) The proposed reward shaping strategy, when the task-agnostic and task-specific IQA are weighted appropriately, successfully identified samples that need re-acquisition due to defected imaging process, as opposed to clinically challenging cases due to low contrast in pathological tissues or other equivocacy in radiological presentation.<br>Our code is available at https://github.com/s-sd/task-amenability/tree/v1

Infrared sensor performance with boost and restoration filtering.

The targeting task performance (TTP) model for prediction of target identification range suggests that boost filtering with a well-sampled, low-noise long-wave infrared (LWIR) sensor can substantially increase target ID range (by enhancing contrast at high spatial frequencies). We model a notional high-performance LWIR imaging system with a high F-number, deep electron wells, and a small-pitch focal plane array. System analysis performed with the Night Vision Integrated Performance Model (NVIPM) predicts that a range enhancement upwards of 50% is achievable with Wiener restoration applied to imagery from the modeled sensor. Human perception experiments were performed on simulated target imagery, with range through different boost filters (including a Wiener restoration filter) compared to the no-post-filter case. The TTP model was found to significantly overestimate the performance improvement due to boost and restoration filtering. Alternate predictions based on the Johnson criteria were also performed, and these underestimated the impact of boost. We speculate on reasons for the discrepancy and on promising avenues for future research. Sensor parameters, NVIPM predictions, filter parameters, and experimental data are provided.

UMS-Rep: Unified modality-specific representation for efficient medical image analysis.

Medical image analysis typically includes several tasks such as enhancement, segmentation, and classification. Traditionally, these tasks are implemented using separate deep learning models for separate tasks, which is not efficient because it involves unnecessary training repetitions, demands greater computational resources, and requires a relatively large amount of labeled data. In this paper, we propose a multi-task training approach for medical image analysis, where individual tasks are fine-tuned simultaneously through relevant knowledge transfer using a unified modality-specific feature representation (UMS-Rep). We explore different fine-tuning strategies to demonstrate the impact of the strategy on the performance of target medical image tasks. We experiment with different visual tasks (e.g., image denoising, segmentation, and classification) to highlight the advantages offered with our approach for two imaging modalities, chest X-ray and Doppler echocardiography. Our results demonstrate that the proposed approach reduces the overall demand for computational resources and improves target task generalization and performance. Specifically, the proposed approach improves accuracy (up to ∼ 9% ↑) and decreases computational time (up to ∼ 86% ↓) as compared to the baseline approach. Further, our results prove that the performance of target tasks in medical images is highly influenced by the utilized fine-tuning strategy.

Analysis of Factors Affecting the Efficiency of Electro-Optical Systems

According to present knowledge thermal imaging electro-optical surveillance systems with object detection are widely used in various fields of human activity. Thermal imagers use infrared radiation emitted by objects and people, so it makes possible to take a sight in complete darkness and miscellaneous weather conditions. Typical applications of this technology are security and surveillance systems, along with autonomous navigation industries in aircraft and ground vehicle. The aim of the work is to determine the necessary condition for an accurate parameter estimation of the electro-optical systems allowing for adverse externalities.First of all, the various infrared spectrum ranges will be considered and corresponding wavelengths will be pointed for each of the following. Thermal imaging cameras are capable of receiving emission from an object in the infrared spectral ranges, which have an atmospheric transparency window, i.e. MWIR and LWIR.<p >Then the structural diagram of the phased generation process of infrared images in thermal imaging systems will be discussed. It appears that, the image formation chain includes such components as irradiation source, optics, imaging sensor, read-out electronics, processing electronics, compression and transmission of data, decoding, search & track engine, display and finally human observer.The next chapter will describe the probability method of detecting objects based on the effective resolution of the depicted object. There will be resulted the table of visual perception levels, namely orientation, recognition, identification. And relative levels of perception will be provided the corresponding resolution of the image.The next section will analyze the factors affecting the objects identification. Before the radiation from the object reaches the thermal imaging sensor optics, it must go through many obstacles, which include atmospheric factors such as precipitation and air condition (humidity, temperature), solar radiation, etc. Various equations and models will be carried out to take into account the above factors. For example, Johnson's criterion, and a more advanced Target Task Performance recognition model.This paper considers the main criteria for detecting objects using electro-optical systems in the infrared range. The robust analysis of literature sources will be conduct and describe the basics of visualization and the infrared imaging in thermal systems. Then Johnson criteria, which is considering for calculating the efficiency of electro-optical systems, will be evaluated for validity using in real-world environment. Also, it will be pointed out, that the greatest negative contribution on visualization is made by weather and atmospheric conditions, clutter effects and density, motion-blur and defocusing, along with target properties – aspect ratio and viewing angle. Finally, it will be suggested prediction models and software packages that will help determine the real characteristics of the elaborate device and identify the weak link of the system.

Retrospective auditory cues can improve detection of near-threshold visual targets

Recent studies have demonstrated that visually cueing attention towards a stimulus location after its disappearance can facilitate visual processing of the target and increase task performance. Here, we tested whether such retro-cueing effects can also occur across different sensory modalities, as cross-modal facilitation has been shown in pre-cueing studies using auditory stimuli prior to the onset of a visual target. In the present study, participants detected low-contrast Gabor patches in a speeded response task. These patches were presented in the left or right visual periphery, preceded or followed by a lateralized and task-irrelevant sound at 4 stimulus-onset asynchronies (SOA; −600 ms, −150 ms, +150 ms, +450 ms). We found that pre-cueing at the −150 ms SOA led to a general increase in detection performance irrespective of the sound’s location relative to the target. On top of this temporal effect, sound-cues also had a spatially specific effect, with further improvement when cue and target originated from the same location. Critically, the temporal effect was absent, but the spatial effect was present in the short-SOA retro-cueing condition (+150 ms). Drift-diffusion analysis of the response time distributions allowed us to better characterize the evidenced effects. Overall, our results show that sounds can facilitate visual processing, both pre- and retro-actively, indicative of a flexible and multisensory attentional system that underlies our conscious visual experience.

PB16. Prefrontal bipolar versus multichannel tDCS: Impact on working memory performance

Introduction Recent studies show that tDCS applied over the left dorsolateral prefrontal cortex (lDLPFC) can successfully affect working memory (WM) performance ( Coffman et al., 2014 ). The present study investigates the influences of a classical bipolar and more focal multichannel tDCS over the lDLPFC on WM performance (2-back task) and continuous performance task (CPT) as non target task as well as on resting-state EEG oscillations and ERPs. Methods The study was approved by the Ethic Committee of the Faculty of Medicine Christian-Albrechts University Kiel. Eighteen healthy young adults (18–30 years) participated in the study. Participants were stimulated 3 times (bipolar, multichannel, sham stimulation in randomized order) over the lDLPFC for 20 min, minimum period between sessions was 7 days. For bipolar stimulation 1 mA was applied by two circular saline-soaked surface sponge electrodes (25 cm2), with anode positioned over F3 and reference over FP2. For multichannel tDCS we used five 3.14 cm2 circular PiStim electrodes (Neuroelectrics, Barcelona, Spain), positioned at AF3 (897 uA), AF7 (284 uA), F3 (819 uA), FP2 (−1000 uA) and T7 (−1000 uA). After stimulation EEG at rest (2 min eyes closed, 2 min eyes open) and during 2-back and CPT performance was recorded. Results Behavioral results show significant improvement of accuracy in the 2-back task after multichannel and bipolar tDCS compared to sham stimulation but not for CPT performance. Incidence and intensity of aversive side effects did not differ between multichannel, bipolar or sham tDCS. Conclusion Our results show that multichannel and bipolar tDCS over the lDLPFC have the potential of modulating WM performance, with multifocal stimulation not eliciting greater effects, which is in contrary to results of motor cortex area stimulation ( Fischer et al., 2017 ). Since WM performance but not non target-task performance was affected, our results suggest that the area of interest has been successfully stimulated. Our results on aversive side effect underline the tolerability of multichannel tDCS and bipolar tDCS. Additionally, participants were not able to discriminate real and sham stimulation effectively. However, future analyses, especially of obtained EEG data, will give further insight on the efficacy and tolerability of multichannel tDCS versus bipolar tDCS. This study was conducted as part of the STIPED program. STIPED has received funding from European Union’s Horizon2020 research and innovation programme under grant agreement No. 731827 .

Night vision integrated performance model: impact of a recent change on the model's predictive accuracy.

The U. S. Army has adopted the targeting task performance (TTP) image quality metric for formal analyses of electro-optical imager performance. The most recent release of the model, NVIPM, changes the way TTP is calculated in order to improve the accuracy of facial identification. Predictions of the original and modified models are compared to both laboratory and field data. The change degrades NVIPM predictive accuracy for the tactical vehicle target set which is often used in hardware procurement specifications.

Cultural differences in visual attention: Implications for distraction processing.

We investigated differences between participants of East Asian and Western descent in attention to and implicit memory for irrelevant words which participants were instructed to ignore while completing a target task (a Stroop Task in Experiment 1 and a 1-back task on pictures in Experiment 2). Implicit memory was measured using two conceptual priming tasks (category generation in Experiment 1 and general knowledge in Experiment 2). Participants of East Asian descent showed reliable implicit memory for previous distractors relative to those of Western descent with no evidence of differences on target task performance. We also found differences in a Corsi Block spatial memory task in both studies, with superior performance by the East Asian group. Our findings suggest that cultural differences in attention extend to task-irrelevant background information, and demonstrate for the first time that such information can boost performance when it becomes relevant on a subsequent task.

When and where to transfer for Bayesian network parameter learning

Learning Bayesian networks from scarce data is a major challenge in real-world applications where data are hard to acquire. Transfer learning techniques attempt to address this by leveraging data from different but related problems. For example, it may be possible to exploit medical diagnosis data from a different country. A challenge with this approach is heterogeneous relatedness to the target, both within and across source networks. In this paper we introduce the Bayesian network parameter transfer learning (BNPTL) algorithm to reason about both network and fragment (sub-graph) relatedness. BNPTL addresses (i) how to find the most relevant source network and network fragments to transfer, and (ii) how to fuse source and target parameters in a robust way. In addition to improving target task performance, explicit reasoning allows us to diagnose network and fragment relatedness across Bayesian networks, even if latent variables are present, or if their state space is heterogeneous. This is important in some applications where relatedness itself is an output of interest. Experimental results demonstrate the superiority of BNPTL at various scarcities and source relevance levels compared to single task learning and other state-of-the-art parameter transfer methods. Moreover, we demonstrate successful application to real-world medical case studies.

Tasking on Natural Statistics of Infrared Images.

Natural scene statistics (NSSs) provide powerful, perceptually relevant tools that have been successfully used for image quality analysis of visible light images. Since NSS capture statistical regularities that arise from the physical world, they are relevant to long wave infrared (LWIR) images, which differ from visible light images mainly by the wavelengths captured at the imaging sensors. We show that NSS models of bandpass LWIR images are similar to those of visible light images, but with different parameterizations. Using this difference, we exploit the power of NSS to successfully distinguish between LWIR images and visible light images. In addition, we study distortions unique to LWIR and find directional models useful for detecting the halo effect, simple bandpass models useful for detecting hotspots, and combinations of these models useful for measuring the degree of non-uniformity present in many LWIR images. For local distortion identification and measurement, we also describe a method for generating distortion maps using NSS features. To facilitate our evaluation, we analyze the NSS of LWIR images under pristine and distorted conditions, using four databases, each captured with a different IR camera. Predicting human performance for assessing distortion and quality in LWIR images is critical for task efficacy. We find that NSS features improve human targeting task performance prediction. Furthermore, we conducted a human study on the perceptual quality of noise-and blur-distorted LWIR images and create a new blind image quality predictor for IR images.

Human vision noise model validation for the U.S. Army sensor performance metric

Image noise originating from a sensor system is often the limiting factor in target acquisition perfor- mance, especially when limited by atmospheric transmission or low-light conditions. To accurately predict target acquisition range performance for a wide variety of imaging systems, image degradation introduced by the sen- sor must be properly combined with the limitations of the human visual system (HVS). This crucial step of incor- porating the HVS has been improved and updated within NVESD's latest imaging system performance model. The new noise model discussed here shows how an imaging system's noise and blur are combined with the contrast threshold function (CTF) to form the system CTF. Model calibration constants were found by presenting low-contrast sine gratings with additive noise in a two alternative forced choice experiment. One of the principal improvements comes from adding an eye photon noise term allowing the noise CTF to be accurate over a wide range of luminance. The latest HVS noise model is then applied to the targeting task performance metric respon- sible for predicting system performance from the system CTF. To validate this model, human target acquisition performance was measured from a series of infrared and visible-band noise-limited imaging systems. © 2014

Motivation and performance in a game-based intelligent tutoring system.

One strength of educational games stems from their potential to increase students’ motivation and engagement during educational tasks. However, game features may also detract from principle learning goals and interfere with students’ ability to master the target material. To assess the potential impact of game-based learning environments, in this study we examined motivation and learning for 84 high-school students across eight 1-hr sessions comparing 2 versions of a reading strategy tutoring system, an intelligent tutoring system (iSTART) and its game-based version (iSTART–ME). The results demonstrate equivalent target task performance (i.e., learning) across environments at pretest, posttest, and retention, but significantly higher levels of enjoyment and motivation for the game-based system. Analyses of performance across sessions reveal an initial decrease in performance followed by improvement within the game-based training condition. These results suggest possible constraints and benefits of game-based training, including time-scale effects. The findings from this study offer a potential explanation for some of the mixed findings within the literature and support the integration of game-based features within intelligent tutoring environments that require long-term interactions for students to develop skill mastery.

Target acquisition performance in a cluttered environment

Most existing target acquisition (TA) models neglect the influence of background clutter, which results in inaccurate prediction of TA performance in a complicated environment. In this paper, all the background clutter is first quantitatively characterized by the distribution of edge clutter metric, and its effects on the target detection probability are analyzed. Further, a novel TA model is developed by combining this proposed clutter metric and the target task performance metric based on probability statistics theory. Moreover, this proposed model is validated by the search_2 dataset, and experiment results show that it is more consistent with the subjective detection probability than other models.

Small detectors in infrared system design

Recent progress, in small infrared detector fabrication, has raised interest in determining the minimum useful detector size. We approach detector size analysis, from an imaging system point of view, with reasonable assumptions for future sensor design. The analysis is a simplified version of the target task performance model using the parameter Fλ/d for generalization. Our figure-of-merit is a system characteristic. The results are easy to use and yield minimum useful detector size of 2 μm for the mid-wave infrared region (MWIR) and 5 μm for the long-wave infrared region (LWIR) when coupled with an F/1 optical system under high signal-to-noise ratio conditions. Final size depends upon optical design difficulty, manufacturing constraints, noise equivalent differential temperature, and the operational scenario. For challenging signal-to-noise ratio conditions and more reasonable F/1.2 optics, a 3 μm MWIR detector and a 6 μm LWIR detector are recommended. There are many benefits to approaching these detector sizes with low F -number optics. They include lower cost detectors, no need for dual FOV or continuous zoom optics, and no need for dual F -number optics. Our approach provides the smallest volume and lowest weight sensor with maximum range performance. While this paper focuses on infrared design, our approach applies to all imaging sensors.

Infrared detector size: how low should you go?

In the past five years, significant progress has been accomplished in the reduction of infrared detector pitch and detector size. Recently, longwave infrared (LWIR) detectors in limited quantities have been fabricated with a detector pitch of 5 μm. Detectors with 12-μm pitch are now becoming standard in both midwave infrared (MWIR) and LWIR sensors. Persistent surveillance systems are pursuing 10-μm detector pitch in large format arrays. The fundamental question that most system designers and detector developers desire an answer to is: "How small can you produce an infrared detector and still provide value in performance?" If a system is mostly diffraction-limited, then developing a smaller detector is of limited benefit. If a detector is so small that it does not collect enough photons to produce a good image, then a smaller detector is not much benefit. Resolution and signal-to-noise are the primary characteristics of an imaging system that contribute to targeting, pilotage, search, and other human warfighting task performance. We investigate the task of target discrimination range performance as a function of detector size/pitch. Results for LWIR and MWIR detectors are provided and depend on a large number of assumptions that are reasonable.

After-training emotional interference may modulate sequence awareness in a serial reaction time task

The purpose of the present experiment was to investigate the effects of emotional interference on consolidation of sequential learning. In different sessions, 6 groups of subjects were initially trained on a serial reaction time task (SRTT). To modulate consolidation of the newly learned skill, subjects were exposed, after the training, to 1 of 3 (positive, negative or neutral) different classes of emotional stimuli which consisted of a set of emotional pictures combined with congruent emotional musical pieces or neutral sound. Emotional intervention for each subject group was done in 2 different time intervals (either directly after the training session or 6 h later). After a 72 h post-training interval, each group was retested on the SRTT. Re-test performance was evaluated in terms of response times and accuracy during execution of a target sequence. Emotional intervention did not influence either response times or accuracy of re-testing SRTT target task performance, both variables sensitive to implicit knowledge acquired during SRTT training. However, explicit awareness of sequence knowledge after 72 h was enhanced when negative stimuli had been applied at 0 h after training. These findings suggest that consolidation of explicit aspects of procedural learning may be more responsive toward emotional interference than implicit aspects.