Predictable Target Motion Research Articles

Predicting cervical cancer target motion using a multivariate regression model to enable patient selection for adaptive external beam radiotherapy

Background and purposeInterfraction motion during cervical cancer radiotherapy is substantial in some patients, minimal in others. Non-adaptive plans may miss the target and/or unnecessarily irradiate normal tissue. Adaptive radiotherapy leads to superior dose-volume metrics but is resource-intensive. The aim of this study was to predict target motion, enabling patient selection and efficient resource allocation. Materials and methodsForty cervical cancer patients had CT with full-bladder (CT-FB) and empty-bladder (CT-EB) at planning, and daily cone-beam CTs (CBCTs). The low-risk clinical target volume (CTVLR) was contoured. Mean coverage of the daily CTVLR by the CT-FB CTVLR was calculated for each patient. Eighty-three investigated variables included measures of organ geometry, patient, tumour and treatment characteristics. Models were trained on 29 patients (171 fractions). The Two-CT multivariate model could use all available data. The Single-CT multivariate model excluded data from the CT-EB. A univariate model was trained using the distance moved by the uterine fundus tip between CTs, the only method of patient selection found in published cervix plan-of-the-day studies. Models were tested on 11 patients (68 fractions). Accuracy in predicting mean coverage was reported as mean absolute error (MAE), mean squared error (MSE) and R2. ResultsThe Two-CT model was based upon rectal volume, dice similarity coefficient between CT-FB and CT-EB CTVLR, and uterine thickness. The Single-CT model was based upon rectal volume, uterine thickness and tumour size. Both performed better than the univariate model in predicting mean coverage (MAE 7 %, 7 % and 8 %; MSE 82 %2, 65 %2, 110 %2; R2 0.2, 0.4, −0.1). ConclusionUterocervix motion is complex and multifactorial. We present two multivariate models which predicted motion with reasonable accuracy using pre-treatment information, and outperformed the only published method.

Study on Multi-Heterogeneous Sensor Data Fusion Method Based on Millimeter-Wave Radar and Camera.

This study presents a novel multimodal heterogeneous perception cross-fusion framework for intelligent vehicles that combines data from millimeter-wave radar and camera to enhance target tracking accuracy and handle system uncertainties. The framework employs a multimodal interaction strategy to predict target motion more accurately and an improved joint probability data association method to match measurement data with targets. An adaptive root-mean-square cubature Kalman filter is used to estimate the statistical characteristics of noise under complex traffic scenarios with varying process and measurement noise. Experiments conducted on a real vehicle platform demonstrate that the proposed framework improves reliability and robustness in challenging environments. It overcomes the challenges of insufficient data fusion utilization, frequent leakage, and misjudgment of dangerous obstructions around vehicles, and inaccurate prediction of collision risks. The proposed framework has the potential to advance the state of the art in target tracking and perception for intelligent vehicles.

An improved dynamic programming tracking-before-detection algorithm based on LSTM network

The detection and tracking of small and weak maneuvering radar targets in complex electromagnetic environments is still a difficult problem to effectively solve. To address this problem, this paper proposes a dynamic programming tracking-before-detection method based on a long short-term memory (LSTM) network (LSTM-DP-TBD). With the predicted target motion state provided by the LSTM network, the state transition range of the traditional DP-TBD algorithm can be updated in real time, and the detection and tracking effect achieved for maneuvering small and weak targets is also improved. Utilizing the LSTM network to model the moving state of the target, the moving features of the maneuvering target can be learned from the noisy input data. By incorporating these features into the traditional DP-TBD algorithm, the state transition set can be adjusted in time with the changes in the moving state of the target so that the new algorithm is capable of effectively recursively accumulating the movement trend of the maneuvering small and weak target. Simulation results show that the new algorithm is able to effectively accomplish the task of detecting and tracking maneuvering small and weak targets, and it achieves improved detection and tracking probabilities.

Predictability of sinusoidally moving stimuli does not improve the accuracy of the accommodative response

Previous research work suggests that predictable target motion such as sinusoidal movement can be anticipated by the visual system, thereby improving the accommodative response. The validity of predictable motion for studying human dynamic accommodation is sometimes put into question. The aim of this work was to assess the effect of anticipation along with learning (and motivation, etc.) and fatigue (and boredom, loss of attention, etc.) on dynamic accommodation experiments from a practical perspective. Specifically, changes in amplitude and temporal phase lag were estimated within and between trials as 9 adult observers were instructed to focus on a stimulus that oscillated sinusoidally towards and away from the eye at specific temporal frequencies. On average, amplitude decreased whereas phase increased within trials. No evidence of anticipation or learning was observed either within or between trials. Fatigue consistently dominated anticipation and learning within the course of each trial. Even if the eye is equipped by a prediction operator as it is often assumed, fatigue confounds the results from dynamic accommodation experiments more than anticipation or learning.

Center of Pressure Feedback Modulates the Entrainment of Voluntary Sway to the Motion of a Visual Target

Visually guided weight shifting is widely employed in balance rehabilitation, but the underlying visuo-motor integration process leading to balance improvement is still unclear. In this study, we investigated the role of center of pressure (CoP) feedback on the entrainment of active voluntary sway to a moving visual target and on sway’s dynamic stability as a function of target predictability. Fifteen young and healthy adult volunteers (height 175 ± 7 cm, body mass 69 ± 12 kg, age 32 ± 5 years) tracked a vertically moving visual target by shifting their body weight antero-posteriorly under two target motion and feedback conditions, namely, predictable and less predictable target motion, with or without visual CoP feedback. Results revealed lower coherence, less gain, and longer phase lag when tracking the less predictable compared to the predictable target motion. Feedback did not affect CoP-target coherence, but feedback removal resulted in greater target overshooting and a shorter phase lag when tracking the less predictable target. These adaptations did not affect the dynamic stability of voluntary sway. It was concluded that CoP feedback improves spatial perception at the cost of time delays, particularly when tracking a less predictable moving target.

Both four-dimensional computed tomography and four-dimensional cone beam computed tomography under-predict lung target motion during radiotherapy

Background and purposeTo test the hypothesis that 4DCT and 4DCBCT-measured target motion ranges predict target motion ranges during lung cancer SABR. Materials and methodsTen lung SABR patients were implanted with Calypso beacons. 4DCBCT was reconstructed for 29 fractions (1–4fx/patient) from a 1 min CBCT scan. The beacon centroid motion segmented for all 4DCT and 4DCBCT bins was compared with the real-time imaging and treatment beacon centroid (“target”) motion range (4SDs) for each fraction. We tested the hypotheses that (1) 4DCT and 4CBCT predict treatment motion range and (2) there is no difference between 4DCT and 4DCBCT for predicting treatment motion range. Phase-wise root-mean-square errors (RMSEs) between imaging and treatment motion and reconstructed motion (4DCT, 4DCBCT) were calculated. Relationships between motion ranges in 4DCT and 4DCBCT and imaging and treatment motion ranges were investigated for the superior–inferior (SI), left–right (LR) and anterior–posterior (AP) directions. Baseline drifts and amplitude variability were investigated as potential factors leading to motion misrepresentation. ResultsSI 4DCT, 4DCBCT, imaging and treatment motion ranges were 6.3 ± 3.6 mm, 7.1 ± 4.5 mm, 11.1 ± 7.5 mm and 10.9 ± 6.9 mm, respectively. Similar 4DCT and 4DCBCT under-predictions were observed in the LR and AP directions. Hypothesis (1) was rejected (p < 0.0001). Treatment target motion range was under-predicted in 4DCT by factors of 1.7, 1.9 and 1.7 and in 4DCBCT by factors of 1.5, 1.6 and 1.6 in the SI, LR, and AP directions, respectively. RMSEs were generally lower for end-exhale than inhale. 4DCBCT showed higher correlations with the imaging and treatment target motion than 4DCT and testing hypothesis (2) a statistically significant difference between 4DCT and 4DCBCT was shown in the SI direction (p = 0.03). ConclusionFor lung SABR patients both 4DCT and 4DCBCT significantly under-predict treatment target motion ranges.

Predictive Filtering in Motion Compensation with Steerable Cardiac Catheters.

Robotic cardiac catheterization using ultrasound (US) imaging catheters provides real time imaging from within the heart while reducing the difficulty in manually steering a four degree-of-freedom (4-DOF) catheter. Accurate robotic catheter navigation in the heart is challenging due to a variety of disturbances including cyclical physiological motions, such as respiration. In this work we compensate for respiratory motion by using an Extended Kalman Filter (EKF) to predict target motion and by applying the predictions to steer the US imaging catheter. The system performance was measured in bench top experiments with phantom vasculature. The robotic system with predictive filtering tracked cyclically moving targets with 1.59 mm and 0.72° mean error. Accurately tracking moving structures can improve intra-procedural treatments and visualization.

The Role of Motion Extrapolation in Amphibian Prey Capture

Sensorimotor delays decouple behaviors from the events that drive them. The brain compensates for these delays with predictive mechanisms, but the efficacy and timescale over which these mechanisms operate remain poorly understood. Here, we assess how prediction is used to compensate for prey movement that occurs during visuomotor processing. We obtained high-speed video records of freely moving, tongue-projecting salamanders catching walking prey, emulating natural foraging conditions. We found that tongue projections were preceded by a rapid head turn lasting ∼ 130 ms. This motor lag, combined with the ∼ 100 ms phototransduction delay at photopic light levels, gave a ∼ 230 ms visuomotor response delay during which prey typically moved approximately one body length. Tongue projections, however, did not significantly lag prey position but were highly accurate instead. Angular errors in tongue projection accuracy were consistent with a linear extrapolation model that predicted prey position at the time of tongue contact using the average prey motion during a ∼ 175 ms period one visual latency before the head movement. The model explained successful strikes where the tongue hit the fly, and unsuccessful strikes where the fly turned and the tongue hit a phantom location consistent with the fly's earlier trajectory. The model parameters, obtained from the data, agree with the temporal integration and latency of retinal responses proposed to contribute to motion extrapolation. These results show that the salamander predicts future prey position and that prediction significantly improves prey capture success over a broad range of prey speeds and light levels. Neural processing delays cause actions to lag behind the events that elicit them. To cope with these delays, the brain predicts what will happen in the future. While neural circuits in the retina and beyond have been suggested to participate in such predictions, few behaviors have been explored sufficiently to constrain circuit function. Here we show that salamanders aim their tongues by using extrapolation to estimate future prey position, thereby compensating for internal delays from both visual and motor processing. Predictions made just before a prey turn resulted in the tongue being projected to a position consistent with the prey's pre-turn trajectory. These results define the computations and operating regimen for neural circuits that predict target motion.

Simultaneous Tracking of Multiple Ground Targets from a Multirotor Unmanned Aerial Vehicle

An algorithm for simultaneous tracking of multiple ground targets by an unmanned aerial vehicle is presented. The algorithm is specifically tailored toward multirotor vehicles, and it consists of a particle filter to predict target motion, a reference trajectory generator, and a finite-horizon model predictive controller for trajectory tracking. Two versions of the algorithm are proposed: for a vehicle equipped with a gimbaled camera, and for a vehicle equipped with a fixed camera. Furthermore, a target rejection algorithm is included to prune targets that inhibit accurate tracking of the majority of the target set. The tracking algorithm and multirotor vehicle dynamic model are first described, followed by example simulations for both the gimbaled and fixed-camera cases. Trade studies are presented, analyzing the effects of controller tuning parameters and camera field of view, as well as performance of the target rejection algorithm. In simulation experiments, real-world target data are used to improve simulation fidelity. Overall, results show that the algorithm is effective in capturing a set of ground targets within the field of view simultaneously when using either gimbaled or fixed-camera configurations, but performance is somewhat degraded in the fixed-camera case when target dynamics occur on timescales similar to tracking vehicle dynamics.

SU-GG-T-442: Dose Verification of SRS Monte Carlo Plan with a Moving Anthropomorphic Phantom

Purpose: Dose verification of Stereotactic Radiosurgery plan using Monte Carlo for tissue heterogeneity correction with a moving anthropomorphic phantom Method and Materials: An anthropomorphic lung phantom has gold fiducials, TLD capsules and GAFChromic films imbedded in a target in the left lung, TLD capsules are also inserted in heart and cord structures. The phantom was scanned, CT images were sent to Cyberknife Multiplan for planning. Target and critical structures were contoured. A SRS treatment plan with no tissue heterogeneity correction was generated to give 6Gy to 95% of the target, 5.4Gy to 99% of the target, no point > 2cm from target > 3.5Gy, conformai index < 1.2, cord < 1.8Gy, heart < 3Gy, 10% of whole lung < 2Gy. The plan was recalculated with Monte Carlo with 2% uncertainty for tissue heterogeneity correction for dose delivery. During delivery the phantom moved in superior/inferior and anterior/posterior direction, Cyberknife Synchrony tracking system took orthogonal x-ray images of the phantom while infra-red camera tracked LED diodes placed on the phantom to track phantom motion in real time. After the fiducials were identified by the tracking program and correlated with the motion of LED diodes, a motion model of the target was built; radiation started with the Cyberknife robot following the predicted target motion and adjusted its position during irradiation. Throughout delivery, the motion model was updated and adjusted with new x-ray images of the fiducials. Results: Target TLD was 98.3% of Monte Carlo dose; heart and cord TLD were within 7% acceptability. Profiles from 3 orthogonal films were displaced L/R 0mm/3mm, P/A 1mm/5mm and I/S 1mm/2mm from treatment plan profiles, within 5mm acceptability. Conclusion: Cyberknife Multiplan with Monte Carlo accurately predicts dose in heterogeneous tissue; Cyberknife Synchrony tracking system delivers dose accurately to a target in a moving phantom.

TU-FF-A3-05: Real-Time Lung Tumor Motion Prediction Using Neural Network Based Models Constructed From Unsorted Cine CT Images

Purpose: To develop neural network based models for predicting respiratory induced tumor motion from gating signals and unsorted cine CT images. The ability of the models to predict target motion given real-time surrogate marker motion was verified using a dynamic phantom with a target moving according to motion profiles derived from patient respiratory data. Methods & Materials: Tumor motion profiles were derived using patient data and imported into CIRS Dynamic Thorax Phantom. The phantom was imaged with a 4-slice CT scanner in cine mode. Surrogate marker motion from the phantom exterior was recorded simultaneously using Varian's RPM system. Neural networks were constructed to model relationships between target presence or absence at each slice location and external marker motion. The combined networks for each of the slice locations were used to predict target motion. To verify the ability of the models to infer target position from surrogate signal, additional scans were performed. Models were used to predict target motion from external marker signals and compared to the motion patterns imported into the phantom. Results: Five preliminary experiments were performed with motion patterns from different patients. The models performed well for two cases, predicting motion with average errors on the order of image resolution, 1.26 mm and 1.39 mm root mean squared errors (RMSEs). For motion patterns with large peak-to-peak displacements or highly irregular frequencies / amplitudes, the predictions were less accurate. Conclusion: A neural network based method has been developed for modeling internal target motion using external surrogate signals and unsorted cine CT images. Preliminary results indicate that the approach may be useful for some cases. Adaptive networks utilizing real-time image information during the prediction process may be used to improve prediction accuracy for high frequency / amplitude motion patterns. Conflict of Interest: Partially supported by Varian research grant.

The Contribution of the Human FEF and SEF to Smooth Pursuit Initiation

Smooth pursuit eye movements function to keep moving targets foveated. Behavioral studies have shown that pursuit is particularly effective for predictable target motion. There is evidence that both the frontal eye field (FEF) and supplementary eye field (SEF) (also known as the dorsomedial frontal cortex) contribute to pursuit control. The goal of the current experiment was to determine whether these 2 areas made different contributions to the initiation of pursuit in response to predictable compared with unpredictable target motion. Transcranial magnetic stimulation (TMS) was used in 5 healthy human participants to temporarily disrupt each area around the time of target motion onset. TMS over the FEF delayed contraversive pursuit markedly more than ipsiversive pursuit and this direction-dependent difference was more deeply modulated during pursuit of unpredictable than predictable target motion. By contrast, TMS over the SEF resulted in a much more muted modulation of pursuit latency that was similar across both predictable and unpredictable conditions. Taken together, we conclude that the human FEF, but not the SEF, makes a significant contribution to the processing required during the preparation of contraversive pursuit responses to unpredictable target motion and this contribution is less vital during pursuit to predictable target motion.

Controlling reaching movements with predictable and unpredictable target motion in 10-year-old children and adults

We investigated interception behavior in adults and 10-year-old children. Participants had to intercept virtual targets moving on either a predictable (linear) or unpredictable (non-linear) path (with random direction changes). Targets moved at two different velocities, which varied randomly from trial to trial. Participants reached for the targets via a force-feedback device. Reaching movements for linearly moving targets in a block of linearly moving targets were compared with reaching movements for linearly moving targets in the context of non-linearly moving targets. Movement direction and maximum speed of the first sub-movement were analyzed as well as frequency of target hits and number of sub-movements. Unpredictable target motion caused faster movement speeds than predictable target motion for both children and adults. Additionally, unpredictable target motion caused children and adults to gear their initial movement direction further towards the current position of the target, while with predictable target motion, they geared their initial movement direction further ahead of the target towards an anticipated interception position. Together, these results suggest differential processing of predictable and unpredictable object behavior in an interception task, and that this differential processing is already in place in 10-year-olds.

Supplementary eye fields stimulation facilitates anticipatory pursuit.

Anticipatory movements are motor responses occurring before likely sensory events in contrast to reflexive actions. Anticipatory movements are necessary to compensate for delays present in sensory and motor systems. Smooth pursuit eye movements are often used as a paradigmatic example for the study of anticipation. However, the neural control of anticipatory pursuit is unknown. A previous study suggested that the supplementary eye fields (SEFs) could play a role in the guidance of smooth pursuit to predictable target motion. In this study, we favored anticipatory responses in monkeys by making the parameters of target motion highly predictable and electrically stimulated the SEF before and during this behavior. Stimulation sites were restricted to regions of the SEF where saccades could not be evoked at the same low currents. We found that electrical microstimulation in the SEF increased the velocity of anticipatory pursuit movements and decreased their latency. These effects will be referred to as anticipatory pursuit facilitation. The degree of facilitation was the largest if the stimulation train was delivered near the end of the fixation period, before the moment when anticipatory pursuit usually begins. No anticipatory smooth eye movements could be evoked during fixation without an expectation of target motion. These results suggest that the SEF pursuit area might be involved in the process of guiding anticipatory pursuit.

A Blueprint for Target Motion: fMRI Reveals Perceived Sequential Complexity to Modulate Premotor Cortex

The execution of movements that are guided by an increasingly complex target motion is known to draw on premotor cortices. Whole-brain functional magnetic resonance imaging was used to investigate whether, in the absence of any movement, attending to and predicting increasingly complex target motion also rely on premotor cortices. Complexity was varied as a function of number of sequential elements and amount of dynamic sequential trend in a pulsing target motion. As a result, serial prediction caused activations in premotor and parietal cortices, particularly within the right hemisphere. Parametric analyses revealed that the right ventrolateral premotor cortex and the right anterior intraparietal sulcus were the only areas that, in addition, covaried positively with both behavioral and physical measures of sequential complexity. Further areas that covaried positively with increasing task difficulty reflected influences of both number and trend manipulation. In particular, increasing element number drew on dorsal premotor and corresponding posterior intraparietal regions, whereas increasing trend drew on the visual motion area and area V4. The present findings demonstrate that premotor involvement directly reflects perceptual complexity in attended and predicted target motion. It is suggested that when we try to predict how a target will move, the motor system generates a “blueprint” of the observed motion that allows potential sensorimotor integration. In the absence of any motor requirement, this blueprint appears to be not a by-product of motor planning, but rather the basis for target motion prediction.

A matter of design: No representational momentum without predictability

When observers are asked to localize the final position of a moving target, a forward shift of the judged final position is observed. So far, the forward shift has been attributed to the influence of mental continuation of the final target position (representational momentum). However, studies investigating forward displacement have used highly predictable target motion. The direction of target motion and the final target position were often varied between subjects. Thus, observers may have expected the target to travel in a particular direction or vanish at a particular location before a given trial started. In this study, direction of motion and final position were treated as fixed or random factors. The forward shift and the reversal of the shift with time (memory averaging) were absent when both factors were randomized. Thus, the forward shift with implied motion is restricted to repeatedly observed motion sequences that allow for pre-trial motion prediction.

Ocular pursuit of predicted motion trajectories.

The initiation and maintenance of slow eye movements (SEMs) usually depend on the perception of a moving stimulus. However, the endogenous representation of predictable target motion can be sufficient to initiate and maintain brief episodes of SEM even when the stimulus is not present. In this note, we show that expectancies generated by predictable stimulus motion trajectories can also produce smooth deceleration, reversal of direction, and subsequent acceleration in these movements, and explore the limits of the predictive component of the SEM control system quantitatively.

Single-neuron activity in the dorsomedial frontal cortex during smooth-pursuit eye movements to predictable target motion.

A region of dorsomedial frontal cortex (DMFC) has been implicated in planning and executing saccadic eye movements; hence it has been referred to as a supplementary eye field (SEF). Recently, activity related to executing smooth-pursuit eye movements has been recorded from the DMFC, and microstimulation here has been shown to evoke smooth eye movements. This report documents neuronal activity present in smooth-pursuit tasks where the predictability of target motion was manipulated. The activity of many neurons in the DMFC reached a peak when a predictable change in target motion occurred. Furthermore, the peak activity of some cells was systematically shifted by manipulating the duration of the target event, indicating that the network these neurons were in could learn the temporal characteristics of new target motion. Finally, the activity of most neurons tested was greater when target motion was predictable than when it was unpredictable. The results suggest that the DMFC participates in planning smooth-pursuit eye movements based on past stimulus history.

Ocular vergence under natural conditions. I. Continuous changes of target distance along the median plane

Horizontal binocular eye movements of four subjects were recorded with the scleral sensor coil--revolving magnetic field technique while they fixated a natural target, whose distance was varied in a normally illuminated room. The distance of the target relative to the head of the subject was changed in three ways: (a) the target was moved manually by the experimenter; (b) the target was moved manually by the subject; (c) the target remained stationary while the subject moved his upper torso towards and away from the target. The rate of change of target distance was varied systematically in four levels, ranging from 'slow' to 'very fast', corresponding to changes in target vergence from about 10 degrees s-1 to about 100 degrees s-1. The dynamics of ocular vergence with regard to delay and speed were, under all three conditions, considerably better than could be expected from the literature on ocular vergence induced by disparity and/or blur. When 'very fast' changes in the distance of the target were made, subjects achieved maximum vergence speeds of up to about 100 degrees s-1. Delays of these fast vergence responses were generally smaller than 125 ms. Negative delays, i.e. ocular vergence leading the change in target distance, were observed. The eyes led the target (i.e. predicted target motion) by about 90 ms on average, when the subject used his hand to move the target. Vergence tracking was almost perfect when changes in distance were produced by moving the upper torso. In this condition, the eye led the target by about 5 ms. In the 'slow' and 'medium' conditions (stimulus speeds about 10-40 degrees s-1) tracking was accurate to within 1-2 degrees, irrespective of the way in which the target was moved. In the 'fast' and 'very fast' conditions (stimulus speeds about 40-100 degrees s-1), the accuracy of vergence tracking was better for self-induced than for experimenter-induced target displacements, and accuracy was best during voluntary movements of the upper torso. In the last case, ocular vergence speed was within about 10% of the rate of change of the vergence angle formed by the eyes and the stationary target. The dynamics of convergent and divergent vergence responses varied considerably. These variations were idiosyncratic. They were consistent within, but not between, subjects. Ocular vergence associated with attempted fixation of an imagined target, changing distance in darkness, could only be made by two of the four subjects.(ABSTRACT TRUNCATED AT 400 WORDS)

Predictive eye saccades are different from visually triggered saccades

The metrics of eye saccades in response to random and predictable target motion were determined in 8 normal young subjects. The peak velocity of predictive saccades (PS) was significantly reduced when compared with randomly elicited saccades (RS) of equal amplitude. Saccadic gain (initial saccadic amplitude/target amplitude) was also decreased in PS. Saccadic inaccuracy was extremely frequent in PS and, in contrast to RS, it was independent of target amplitude displacement. These stimulus-dependent changes in saccadic characteristics may reflect activation of the saccadic pontine generators by different supranuclear pathways.