Improved Target Localization Research Articles

Optimization Study of Coal Gangue Detection in Intelligent Coal Selection Systems Based on the Improved Yolov8n Model

To address the low recognition accuracy of models for coal gangue images in intelligent coal preparation systems—especially in identifying small target coal gangue due to factors such as camera angle changes, low illumination, and motion blur—we propose an improved coal gangue separation model, Yolov8n-improvedGD(GD—Gangue Detection), based on Yolov8n. The optimization strategy includes integrating the GCBlock(Global Context Block) from GCNet(Global Context Network) into the backbone network to enhance the model’s ability to capture long-range dependencies in images and improve recognition performance. The CGFPN (Contextual Guidance Feature Pyramid Network) module is designed to optimize the feature fusion strategy and enhance the model’s feature expression capabilities. The GSConv-SlimNeck architecture is employed to optimize computational efficiency and enhance feature map fusion capabilities, thereby improving the model’s robustness. A 160 × 160 scale detection head is incorporated to enhance the sensitivity and accuracy of small coal and gangue detection, mitigate the effects of low-quality data, and improve target localization accuracy.

Intraoperative navigation in craniofacial surgery

Craniofacial surgery requires comprehensive anatomical knowledge of the head and neck to ensure patient safety and surgical precision. Over recent decades, there have been significant advancements in imaging techniques and the development of real-time surgical navigation systems. Intraoperative navigation technology aligns surgical instruments with imaging-derived information on patient anatomy, enabling surgeons to closely follow preoperative plans. This system functions as a radiologic map, improving the accuracy of instrument placement and minimizing surgical complications. The introduction of first-generation navigation systems in the early 1990s revolutionized surgical procedures by enabling real-time tracking of instruments using preoperative imaging. Initially utilized in neurosurgery, intraoperative navigation has since become standard practice in otolaryngology, cranio-maxillofacial surgery, and orthopedics. Since the 2000s, second-generation navigation systems have been developed to meet the growing demand for precision across various surgical specialties. The adoption of these systems in craniofacial surgery has been slower, but their use is increasing, particularly in procedures such as foreign body removal, facial bone fracture reconstruction, tumor resection, and craniofacial reconstruction and implantation. In Korea, insurance coverage for navigation in craniofacial surgery began in 2021, and new medical technologies for orbital wall fracture treatment were approved in August 2022. These technologies have only recently become clinically available, but are expected to play an increasingly important role in craniofacial surgery. Intraoperative navigation enhances operative insight, improves target localization, and increases surgical safety. Although these systems have a steep learning curve and initially prolong surgery, efficiency improves with experience. Calibration issues, registration errors, and soft tissue deformation can introduce inaccuracies. Nonetheless, navigation technology is evolving, and the integration of intraoperative computed tomography data holds promise for further enhancements of surgical accuracy. This paper discusses the various types and applications of navigation employed in craniofacial surgery, highlighting their benefits and limitations.

Improving the maize crop row navigation line recognition method of YOLOX.

The accurate identification of maize crop row navigation lines is crucial for the navigation of intelligent weeding machinery, yet it faces significant challenges due to lighting variations and complex environments. This study proposes an optimized version of the YOLOX-Tiny single-stage detection network model for accurately identifying maize crop row navigation lines. It incorporates adaptive illumination adjustment and multi-scale prediction to enhance dense target detection. Visual attention mechanisms, including Efficient Channel Attention and Cooperative Attention modules, are introduced to better extract maize features. A Fast Spatial Pyramid Pooling module is incorporated to improve target localization accuracy. The Coordinate Intersection over Union loss function is used to further enhance detection accuracy. Experimental results demonstrate that the improved YOLOX-Tiny model achieves an average precision of 92.2 %, with a detection time of 15.6 milliseconds. This represents a 16.4 % improvement over the original model while maintaining high accuracy. The proposed model has a reduced size of 18.6 MB, representing a 7.1 % reduction. It also incorporates the least squares method for accurately fitting crop rows. The model showcases efficiency in processing large amounts of data, achieving a comprehensive fitting time of 42 milliseconds and an average angular error of 0.59°. The improved YOLOX-Tiny model offers substantial support for the navigation of intelligent weeding machinery in practical applications, contributing to increased agricultural productivity and reduced usage of chemical herbicides.

CCDN-DETR: A Detection Transformer Based on Constrained Contrast Denoising for Multi-Class Synthetic Aperture Radar Object Detection.

The effectiveness of the SAR object detection technique based on Convolutional Neural Networks (CNNs) has been widely proven, and it is increasingly used in the recognition of ship targets. Recently, efforts have been made to integrate transformer structures into SAR detectors to achieve improved target localization. However, existing methods rarely design the transformer itself as a detector, failing to fully leverage the long-range modeling advantages of self-attention. Furthermore, there has been limited research into multi-class SAR target detection. To address these limitations, this study proposes a SAR detector named CCDN-DETR, which builds upon the framework of the detection transformer (DETR). To adapt to the multiscale characteristics of SAR data, cross-scale encoders were introduced to facilitate comprehensive information modeling and fusion across different scales. Simultaneously, we optimized the query selection scheme for the input decoder layers, employing IOU loss to assist in initializing object queries more effectively. Additionally, we introduced constrained contrastive denoising training at the decoder layers to enhance the model's convergence speed and improve the detection of different categories of SAR targets. In the benchmark evaluation on a joint dataset composed of SSDD, HRSID, and SAR-AIRcraft datasets, CCDN-DETR achieves a mean Average Precision (mAP) of 91.9%. Furthermore, it demonstrates significant competitiveness with 83.7% mAP on the multi-class MSAR dataset compared to CNN-based models.

Vehicle Detection Based on Information Fusion of mmWave Radar and Monocular Vision

Single sensors often fail to meet the needs of practical applications due to their lack of robustness and poor detection accuracy in harsh weather and complex environments. A vehicle detection method based on the fusion of millimeter wave (mmWave) radar and monocular vision was proposed to solve this problem in this paper. The method successfully combines the benefits of mmWave radar for measuring distance and speed with the vision for classifying objects. Firstly, the raw point cloud data of mmWave radar can be processed by the proposed data pre-processing algorithm to obtain 3D detection points with higher confidence. Next, the density-based spatial clustering of applications with noise (DBSCAN) clustering fusion algorithm and the nearest neighbor algorithm were also used to correlate the same frame data and adjacent frame data, respectively. Then, the effective targets from mmWave radar and vision were matched under temporal-spatio alignment. In addition, the successfully matched targets were output by using the Kalman weighted fusion algorithm. Targets that were not successfully matched were marked as new targets for tracking and handled in a valid cycle. Finally, experiments demonstrated that the proposed method can improve target localization and detection accuracy, reduce missed detection occurrences, and efficiently fuse the data from the two sensors.

Experimental demonstration of magnitude-only bearing estimation

Traditional bearing estimation systems exploit relative time delays across an array of sensors to localize a target. While highly effective, such systems are costly and cumbersome. Yovel et al. (Science, 2010) found that bats steer their echolocation beams off axis to maximize the Fisher information, implying that spectral magnitude cues improve target localization. This presentation shares experiments demonstrating a man-made system estimating a target's bearing from echo spectrum magnitude-information using a single directional sensor with frequency dependence over the bandwidth of the signal. The output of a 31-element microphone array steered to broadside serves as the frequency dependent directional sensor. Moving a source transmitting a linear FM chirp across a range of angles estimates the beampattern of the receiver. Maximum likelihood estimation using these recorded replicas gives the mean-squared error (MSE) as a function of angle. The MSE is compared with the Cramer-Rao lower bound. Moving away from broadside reduces the received SNR, but paradoxically the MSE exhibits local minima. These MSE local minima are consistent with the optimal angles observed in previous research simulating the impact of magnitude cues (Kloepper et al., JASA-EL, 2018; Tidwell et al., IEEE SSPD, 2019). [Work supported by ONR MURI Program.]

Improved Target Localization in Multiwaveform Multiband Hybrid Multistatic Radar Networks

This study proposes an algorithm to improve the target localization performance. This is implemented in a multiwaveform multiband hybrid (passive and active) multistatic radar network scenario that utilizes broadcasting signals for radar sensing in addition to the radar waveforms. Multiwaveform multiband radar receivers can exploit the broadcast signals transmitted by noncooperative transmitters, such as communication or broadcasting systems, for target sensing in addition to radar waveform. Hence, multiple measurements of the targets can be acquired and fused to improve target detection and parameter estimation. Because of utilizing various waveforms, each transmitter–receiver (Tx–Rx) pair has a different range and velocity estimation accuracy, which is also affected by the bistatic geometry of the bistatic pairs. Taking this into account, this study proposes a target localization algorithm based on bistatic Cramér–Rao lower bounds (CRLBs) for multistatic multiband radar networks. It is shown that modeling the entire network, evaluating the bistatic range CRLB of each bistatic pair in advance, and utilizing this information while estimating the target location significantly improve the localization accuracy. Moreover, the proposed algorithm also includes a target height estimation correction stage to achieve better 3-D localization accuracy.

Put Your Finger on it: The role of Explicit Cognitive Strategies in Laparoscopic Surgery.

This brief commentary highlights the importance of understanding the nature of explicit strategy (or heuristic) use in laparoscopic surgery including why such strategies might be useful and/or necessary in the first place. The use of strategies and other tricks warrants careful experimental investigation to determine their precise impact, and how they can be most effectively leveraged to improve surgical training. Experienced laparoscopic surgeons are familiar with the unsettling phenomenon of making a particular instrument movement (or sequence of movements), only to find that the movement of the instrument tip deviates greatly from their intention. In laparoscopic surgery, at least in complex laparoscopic surgery where the relationship between the camera and instrument ports can vary widely, these surprises can happen frequently. A common example occurs when the laparoscope faces the operator and, although upward hand motion produces the expected downward deflection of the instrument tip (reflective of the normal sensorimotor mapping in laparoscopic surgical contexts), leftward motion produces leftward tip motion and rightward motion produces rightward tip motion. Contending with this mismatch can be challenging for the surgeon. The surgeon is often so absorbed in the laparoscopic image as to feel completely present “within” the laparoscopic domain, oblivious to the screen to the extent that they may reposition their head to look behind a structure. The surgeon must deliberately shift their frame of reference to the outside world by pressing the abdominal wall with their finger at the location they want the instrument tip to go. Applying pressure to the abdominal wall with one’s hand seems to recruit proprioception as another source of information for “where things are.” The technique has the added benefit of bringing degrees of freedom from beyond the instrument, effectively delivering tissue to the instrument rather than having the instrument find the tissue; it is a subtle but powerful technique. This simple action seemingly provides both proprioceptive information and a useful visual reference that can improve target localization and simplify the motor control task. “Stop looking at the screen and look at the patient for a moment” can be a useful admonition. Ask “where do you want the instrument to go?” In surgery, useful tricks like this are called “heuristics.” In the scientific field of motor learning, they are called “explicit cognitive strategies,” which are thought to augment slower, implicit forms of learning in the motor system that tend to occur below the purview of conscious awareness.1 For example, prism goggles that shift the entire visual field to the right by a few degrees perturb the normal relationship between movements and the resulting sensory feedback.2 This change in the sensorimotor mapping immediately causes movement errors. Over successive attempts, the motor system gradually recalibrates this mapping so that movements become accurate once again—a process known as “sensorimotor adaptation.”3 When the shift or perturbation is large enough and human subjects become aware of it, they can volitionally compensate for the perturbation by using an explicit aiming strategy (eg, “If I aim to the left, I’ll be on target”). More generally, there is evidence that explicit strategies can serve as “instructional nudges” during the learning of new motor skills, helping to scaffold lower-level processes of motor skill acquisition and consolidation.4 For this reason, incorporating information about these and other related strategies into training programs could help reduce the learning curve for acquiring the complex hand-eye coordination skills needed for laparoscopic surgery. Although surgical expertise has been the topic of considerable experimental and theoretical investigation,5,6 there are a number of important aspects of skilled surgical performance that remain unexplored.7 This brief commentary highlights the importance of understanding the nature of explicit strategy (or heuristic) use in laparoscopic surgery including why such strategies might be useful and/or necessary in the first place. Second, if strategy use confers performance benefits, there is an important need to document these strategies as well as the specific contexts in which they are useful so that this information can be communicated consistently to trainees. Finally, the use of strategies and other tricks warrants careful experimental investigation to determine their precise impact, and how they can be most effectively leveraged to improve surgical training.

Simulating elastic targets for sonar algorithm development

Targets interrogated during synthetic aperture sonar (SAS) applications may exhibit elastic scattering behavior and re-radiate sound after the initial geometric returns. This behavior can be a discriminating feature in target identification, but may also obscure nearby targets and make the full scene difficult to comprehend. Advancing algorithms to leverage elastic scattering for target localization and identification requires data with accurate ground truth. There is a shortage of field data containing elastic targets with sufficient quality for algorithm development. Stanton’s 1984 model presents the scattering amplitude of an elastic cylinder as a function of frequency and spherical angles. This presentation uses Stanton’s scattering model to simulate the time series of pressure scattered by the cylinder. This approach allows the target scattering model to be readily coupled with time series models for the environment to generate high fidelity simulated data. Simulated time series are generated for SAS collection geometries. The presentation will explore this modeling approach as a substitute for field data during initial stages of algorithm development for improved target localization. It will investigate the impact of elastic effects on imagery and discuss the limitations of methods used to isolate such effects.

Coherence-Induced Bias Reduction in Synthetic Aperture Sonar Along-Track Micronavigation

Subwavelength motion estimation is vital for the production of focused synthetic aperture sonar (SAS) imagery. The required precision is obtainable from the sonar data itself through a process termed micronavigation. Along-track micronavigation is achieved by a similar technique to that used in correlation velocity logs (CVLs), where sparse estimates of the spatial coherence function are interpolated to estimate the location of the peak coherence and hence estimate the interping vehicle motion. However, along-track micronavigation estimates made using this technique are biased, which limits the utility of these measurements for long-term navigation of autonomous underwater vehicles (AUVs). Three sources of along-track motion estimation bias are considered in this article. First, imperfect temporal registration between the signals results in coherence estimates that are negatively biased as a function of the temporal offset. Second, the sparse estimates of the spatial coherence function are obtained by cross-correlation of complex baseband signals, a process which is known to result in positively biased coherence estimates, especially when the true coherence is low. Finally, mismatches between the underlying spatial coherence function and the interpolation kernel used to estimate the peak coherence location also result in along-track micronavigation bias. In this article, we describe and evaluate three methods for reducing along-track micronavigation bias. We introduce a temporal registration of the signals before coherence estimation, which reduces the impact of negative coherence bias due to temporal offsets. The remaining coherence estimation bias is reduced by combining multiple coherence estimates in a Bayesian coherence estimator. Additionally, an improved interpolation kernel is derived with a significantly improved fit compared to the current gold standard Gaussian interpolation kernel. The improvements in along-track micronavigation accuracy are demonstrated using two simulated data sets, which both allow comparison with ground truth. The first involves direct simulation of the spatial coherence from a given interping geometry using the pulse-echo formulation of the van Cittert–Zernike theorem, while the second involves simulation of raw sonar echo data using a point-scatterer model. Using these simulations, a reduction in along-track micronavigation bias of 48.5%–99.5% is demonstrated, with reductions in along-track micronavigation error standard deviation of up to 34%. This improvement expands the potential for SAS-equipped AUVs to reduce their long-term navigation drift, facilitating longer underwater transits, improved target localization, and reduced track misalignment in repeat-pass operations.

P72 - Stereotactic radiosurgery for multiple brain metastases: does MR/CT co-registration improve target localization accuracy compared to the MR-only approach?

P72 - Stereotactic radiosurgery for multiple brain metastases: does MR/CT co-registration improve target localization accuracy compared to the MR-only approach?

Evaluation of patient-specific MR distortion correction schemes for improved target localization accuracy in SRS.

This work aims at promoting target localization accuracy in cranial stereotactic radiosurgery (SRS) applications by focusing on the correction of sequence-dependent (also patient induced) magnetic resonance (MR) distortions at the lesion locations. A phantom-based quality assurance (QA) methodology was developed and implemented for the evaluation of three distortion correction techniques. The same approach was also adapted to cranial MR images used for SRS treatment planning purposes in single or multiple brain metastases cases. A three-dimensional (3D)-printed head phantom was filled with a 3D polymer gel dosimeter. Following treatment planning and dose delivery, volumes of radiation-induced polymerization served as hypothetical lesions, offering adequate MR contrast with respect to the surrounding unirradiated areas. T1-weighted (T1w) MR imaging was performed at 1.5T using the clinical scanning protocol for SRS. Additional images were acquired to implement three distortion correction methods; the field mapping (FM), mean image (MI) and signal integration (SI) techniques. Reference lesion locations were calculated as the averaged centroid positions of each target identified in the forward and reverse read gradient polarity MRI scans. The same techniques and workflows were implemented for the correction of contrast-enhanced T1w MR images of 10 patients with a total of 27 brain metastases. All methods employed in the phantom study diminished spatial distortion. Median and maximum distortion magnitude decreased from 0.7mm (2.10ppm) and 0.8mm (2.36ppm), respectively, to <0.2mm (0.61ppm) at all target locations, using any of the three techniques. Image quality of the corrected images was acceptable, while contrast-to-noise ratio slightly increased. Results of the patient study were in accordance with the findings of the phantom study. Residual distortion in corrected patient images was found to be <0.3mm in the vast majority of targets. Overall, the MI approach appears to be the most efficient correction method from the three investigated. In cranial SRS applications, patient-specific distortion correction at the target location(s) is feasible and effective, despite the expense of longer imaging time since additional MRI scan(s) need to be performed. A phantom-based QA methodology was developed and presented to reassure efficient implementation of correction techniques for sequence-dependent spatial distortion.

Using predicted target state to inform transmit waveform selection in active sonar

Continuous-wave (CW) and frequency-modulated (FM) transmit pulses are commonly used in active sonar systems. CW pulses allow for measurement of target Doppler but have relatively poor range accuracy. In contrast, FM pulses are insensitive to Doppler but allow for improved range accuracy relative to CW. Active sonars often operate in clutter-dominated environments, where choosing a CW pulse for a high Doppler target provides better tracking capability, while an FM pulse is more effective for a low Doppler target. A challenge, however, is that the target Doppler not known before transmission. At present, the complementary strengths of CW and FM pulses can be exploited only by an operator who actively chooses between them, but requiring an operator to select the transmitted waveform on a ping-by-ping basis presents an unreasonable burden. In this talk, we describe the Predicted State-Based Selection (PSBS) algorithm, which uses an estimate of the target Doppler, derived from a state prediction produced by the tracking filter, to select the transmitted pulse. Monte Carlo simulations are conducted to evaluate the performance and behavior of the PSBS algorithm. Results show that PSBS improves target localization estimates by 7.7% on average relative to transmission of the same waveform for all pings.

Daily computed tomography image guidance: Dosimetric advantages outweigh low-dose radiation exposure for treatment of mediastinal lymphoma

CT-guided treatment delivery can improve target localization in mediastinal lymphoma patients treated with “butterfly” intensity-modulated radiotherapy and deep-inspiration breath hold. Although daily CT imaging adds additional radiation exposure, its use may be justified given the greater normal tissue sparing enabled by PTV margin reduction.

Unmanned Aerial vehicle’s runway landing system with efficient target detection by using morphological fusion for military surveillance system

In the surveillance of military system Unmanned Ariel vehicles (UAV’s) offers a remarkable service. Software Defined Network (SDN) is one of approach for UAV for improving target localization with surveillance for many applications . One of the critical issues for an Unmanned Ariel vehicles (UAV’s) landing is the detection of runway in a low visibility conditions with computer vision. Also, the key problem is the accurate and robust detection of runway. A method of runway detection for fixed wing UAVs with a forward-looking camera is uses sensor based communication in network framework between source (UAV’S) and destination (runway sensor). Yet, this depends on the technical experience for landing an UAV’s safely. Nowadays, highly advanced equipments are being developed for improving safety and transportation system to handle the problem of landing under low visibility conditions. However, natural or habitual vision could be reduced during low visibility situations. This may be due to many meteorological factors like haze, darkness, fog etc. From the discussion made in this paper, the authors mainly concentrate on managing an aircraft system under low luminosity condition during the landing. In this research, effective Morphological fusion method is employed for the prediction of virtual runway imagery to avoid accident landing process. For this, fusion of sensor data for DEM (Digital Elevation Map) Data, infrared image (IR) and navigation parameter are used. The performance of this research using Morphological fusion method gives a fused image of the runway prediction for UAV’s under poor visibility condition. After obtaining the fused image, it is involved in ROI (Region of Interest) contour tracing process to get the clear location of landing of an UAV’s without harm. The virtual imaginary model can be produced through contour tracing to predict runway. By this method, a less prediction time or setting time is achieved and maximum accuracy is obtained through simulation using MATLAB tool. Finally, the proposed experimental outcomes are compared with the existing technology. The proposed approach will be use in aircraft sensor based communication between the network (source to destination) to detect the target and to provide a better surveillance for military applications.

Exploiting cooperative sensing for accurate target tracking in industrial Internet of things

Wireless sensor networks are a cornerstone of the Internet of things with many applications. An important aspect of such applications is target tracking using self-positioned known sensor nodes. Over the years, many schemes have been proposed to locate and track the target path. However, accuracy and reliable tracking remain an open area of research. In this article, we propose a dynamic cooperative multilateral sensing scheme for indoor industrial environments to improve target localization and tracking accuracy. The scheme is designed to select reliable nodes based on the distance between nodes within-cluster and to the target for reduced positioning error. Furthermore, a cluster node is dynamically selected based on distance from the base station. We simulate the proposed technique in scenarios with tracking at regular intervals and with the complete path. Furthermore, the performance of the scheme is also tested under different sensor coverage areas. The results show that the proposed scheme provides better target tracking with up to 19% higher accuracy in comparison to the traditional trilateration scheme.

Experimental analysis of the effects of bistatic target scattering on synthetic aperture sonar imagery

Downward looking synthetic aperture sonar (SAS) systems, such as those used to detect surficial and buried unexploded ordinance, may use arrays with large spatial extent compared to the depth of the targets they are imaging. With such systems the targets are in the near field of the physical array formed by the projector and receiver. Beamforming algorithms, data representation schemes, and automated target recognition algorithms can benefit from considering the bistatic scattering patterns of targets in this geometry. In downward looking SAS systems, resonant scattering behavior may be used to discriminate targets from clutter because scattering from the sediment-water interface may obscure the geometric scattering response of targets. An experimental analysis of the effects on SAS imagery due to bistatic collection geometries was conducted using an in-air laboratory setup with resonant targets. This data was used to develop a signal processing algorithm that may help improve target localization, detection, and identification by focusing re-radiated resonant energy from targets. Results from the application of this algorithm are compared for an approximately monostatic and a truly bistatic SAS collection geometry using an aluminum pipe target.

Tactile Cues for Improving Target Localization in Subjects with Tunnel Vision

The loss of peripheral vision is experienced by millions of people with glaucoma or retinitis pigmentosa, and has a major impact in everyday life, specifically to locate visual targets in the environment. In this study, we designed a wearable interface to render the location of specific targets with private and non-intrusive tactile cues. Three experimental studies were completed to design and evaluate the tactile code and the device. In the first study, four different tactile codes (single stimuli or trains of pulses rendered either in a Cartesian or a Polar coordinate system) were evaluated with a head pointing task. In the following studies, the most efficient code, trains of pulses with Cartesian coordinates, was used on a bracelet located on the wrist, and evaluated during a visual search task in a complex virtual environment. The second study included ten subjects with a simulated restrictive field of view (10°). The last study consisted of proof of a concept with one visually impaired subject with restricted peripheral vision due to glaucoma. The results show that the device significantly improved the visual search efficiency with a factor of three. Including object recognition algorithm to smart glass, the device could help to detect targets of interest either on demand or suggested by the device itself (e.g., potential obstacles), facilitating visual search, and more generally spatial awareness of the environment.

GPR target detection using a neural network classifier designed by a multi-objective genetic algorithm

Ground Penetrating Radar (GPR) is an electromagnetic sensing technology employed for localization of underground utilities, pipes, and other types of objects. The radargrams typically obtained have a high dimensionality, containing a number of signatures with hyperbolic pattern shapes, and can be processed to retrieve information about the target’s locations, depths and material type of underground soil. The classical Hough Transform approach used to reconstruct these hyperbola shapes is computationally expensive, given the large dimensionality of the radargrams. In literature, several approaches propose to first approximate the location of hyperbolas to small segments through a classification stage, before applying the Hough transform over these segments. However, the published classifiers designed for this task present a relatively complex architecture.Aiming at an improved target localization, we propose an alternative classification methodology. The goal is to classify windows of GPR radargrams into two classes (with or without target) using a neural network radial basis function (RBF), designed via a multi-objective genetic algorithm (MOGA). To capture samples’ fine details, high order statistic cumulant features (HOS) were used. Feature selection was performed by MOGA, with an optional prior reduction using a mutual information (MIFS) approach. The obtained results demonstrate improvement of the classification performance when compared with other models designed with the same data and are among the best results available in the literature, albeit the large reduction in classifier complexity.

Improving target localization during trans-oral surgery with use of intraoperative imaging.

Trans-oral surgery provides a less invasive means for the surgical management of upper aerodigestive tract malignancies but is limited in its ability to readily assess submucosal tumor extent and location of critical structures intraoperatively. We sought to determine surgeons' baseline target localization accuracy during operative laryngoscopy with preoperative imaging alone and then assess for improvement in localization accuracy when presented with intraoperative CT imaging capturing soft tissue deformation. Fiducial beads were placed submucosally in four cadaver heads. "Preoperative" (PO) and "intraoperative" (IO) neck CTs were acquired before and during suspension laryngoscopy using a CT-compatible laryngoscopy system. Surgeons attempted to localize submucosal fiducials beads using pins based on sequential review of PO and IO images. Mean total error (TE) decreased from 12.8 ± 9.9 to 10 ± 7.5mm from PO to IO (P < 0.001), respectively. TE for base of tongue and vallecula decreased by 1.7 ± 6.7mm (P = 0.015). Right-sided structures were most exposed by scope positioning and experienced a TE reduction of 4.8 ± 9.3mm (P < 0.001). Task completion time decreased from PO to IO by 26% (P < 0.001). Intraoperative imaging significantly improves localization accuracy and task efficiency when targeting submucosal beads in cadaver heads during operative laryngoscopy.