Protecting Prostate Cancer Classification From Rectal Artifacts via Targeted Adversarial Training.

Magnetic resonance imaging (MRI)-based deep neural networks (DNN) have been widely developed to perform prostate cancer (PCa) classification. However, in real-world clinical situations, prostate MRIs can be easily impacted by rectal artifacts, which have been found to lead to incorrect PCa classification. Existing DNN-based methods typically do not consider the interference of rectal artifacts on PCa classification, and do not design specific strategy to address this problem. In this study, we proposed a novel Targeted adversarial training with Proprietary Adversarial Samples (TPAS) strategy to defend the PCa classification model against the influence of rectal artifacts. Specifically, based on clinical prior knowledge, we generated proprietary adversarial samples with rectal artifact-pattern adversarial noise, which can severely mislead PCa classification models optimized by the ordinary training strategy. We then jointly exploited the generated proprietary adversarial samples and original samples to train the models. To demonstrate the effectiveness of our strategy, we conducted analytical experiments on multiple PCa classification models. Compared with ordinary training strategy, TPAS can effectively improve the single- and multi-parametric PCa classification at patient, slice and lesion level, and bring substantial gains to recent advanced models. In conclusion, TPAS strategy can be identified as a valuable way to mitigate the influence of rectal artifacts on deep learning models for PCa classification.

The effect of elimination of gibbs ringing, noise and systematic errors on the DTI metrics and tractography in a rat brain.

Diffusion tensor imaging (DTI) metrics and tractography can be biased due to low signal-to-noise ratio (SNR) and systematic errors resulting from image artifacts and imperfections in magnetic field gradients. The imperfections include non-uniformity and nonlinearity, effects caused by eddy currents, and the influence of background and imaging gradients. We investigated the impact of systematic errors on DTI metrics of an isotropic phantom and DTI metrics and tractography of a rat brain measured at high resolution. We tested denoising and Gibbs ringing removal methods combined with the B matrix spatial distribution (BSD) method for magnetic field gradient calibration. The results showed that the performance of the BSD method depends on whether Gibbs ringing is removed and the effectiveness of stochastic error removal. Region of interest (ROI)-based analysis of the DTI metrics showed that, depending on the size of the ROI and its location in space, correction methods can remove systematic bias to varying degrees. The preprocessing pipeline proposed and dedicated to this type of data together with the BSD method resulted in an even - 90% decrease in fractional anisotropy (FA) (globally and locally) in the isotropic phantom and - 45% in the rat brain. The largest global changes in the rat brain tractogram compared to the standard method without preprocessing (sDTI) were noticed after denoising. The direction of the first eigenvector obtained from DTI after denoising, Gibbs ringing removal and BSD differed by an average of 56 and 10 degrees in the ROI from sDTI and from sDTI after denoising and Gibbs ringing removal, respectively. The latter can be identified with the amount of improvement in tractography due to the elimination of systematic errors related to imperfect magnetic field gradients. Based on the results, the systematic bias for high resolution data mainly depended on SNR, but the influence of non-uniform gradients could also be seen. After denoising, the BSD method was able to further correct both the metrics and tractography of the diffusion tensor in the rat brain by taking into account the actual distribution of magnetic field gradients independent of the examined object and uniquely dependent on the scanner and sequence. This means that in vivo studies are also subject to this type of errors, which should be taken into account when processing such data.

Enhancing image quality in computed tomography angiography follow-ups after endovascular aneurysm repair: a comparative study of reconstruction techniques.

BACKGROUND: The image quality of computed tomography angiography (CTA) images following endovascular aneurysm repair (EVAR) is not satisfactory, since artifacts resulting from metallic implants obstruct the clear depiction of stent and isolation lumens, and also adjacent soft tissues. However, current techniques to reduce these artifacts still need further advancements due to higher radiation doses, longer processing times and so on. Thus, the aim of this study is to assess the impact of utilizing Single-Energy Metal Artifact Reduction (SEMAR) alongside a novel deep learning image reconstruction technique, known as the Advanced Intelligent Clear-IQ Engine (AiCE), on image quality of CTA follow-ups conducted after EVAR. MATERIALS: This retrospective study included 47 patients (mean age ± standard deviation: 68.6 ± 7.8 years; 37 males) who underwent CTA examinations following EVAR. Images were reconstructed using four different methods: hybrid iterative reconstruction (HIR), AiCE, the combination of HIR and SEMAR (HIR + SEMAR), and the combination of AiCE and SEMAR (AiCE + SEMAR). Two radiologists, blinded to the reconstruction techniques, independently evaluated the images. Quantitative assessments included measurements of image noise, signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR), the longest length of artifacts (AL), and artifact index (AI). These parameters were subsequently compared across different reconstruction methods. RESULTS: The subjective results indicated that AiCE + SEMAR performed the best in terms of image quality. The mean image noise intensity was significantly lower in the AiCE + SEMAR group (25.35 ± 6.51 HU) than in the HIR (47.77 ± 8.76 HU), AiCE (42.93 ± 10.61 HU), and HIR + SEMAR (30.34 ± 4.87 HU) groups (p < 0.001). Additionally, AiCE + SEMAR exhibited the highest SNRs and CNRs, as well as the lowest AIs and ALs. Importantly, endoleaks and thrombi were most clearly visualized using AiCE + SEMAR. CONCLUSIONS: In comparison to other reconstruction methods, the combination of AiCE + SEMAR demonstrates superior image quality, thereby enhancing the detection capabilities and diagnostic confidence of potential complications such as early minor endleaks and thrombi following EVAR. This improvement in image quality could lead to more accurate diagnoses and better patient outcomes.

Verification of the effect of data-driven brain motion correction on PET imaging.

INTRODUCTION: Brain positron emission tomography/computed tomography (PET/CT) scans are useful for identifying the cause of dementia by evaluating glucose metabolism in the brain with F-18-fluorodeoxyglucose or Aß deposition with F-18-florbetaben. However, since imaging time ranges from 10 to 30 minutes, movements during the examination might result in image artifacts, which interfere with diagnosis. To solve this problem, data-driven brain motion correction (DDBMC) techniques are capable of performing motion corrected reconstruction using highly accurate motion estimates with high temporal resolution. In this study, we investigated the effectiveness of DDBMC techniques on PET/CT images using a Hoffman phantom, involving continuous rotational and tilting motion, each expanded up to approximately 20 degrees. MATERIALS AND METHODS: Listmode imaging was performed using a Hoffman phantom that reproduced rotational and tilting motions of the head. Brain motion correction processing was performed on the obtained data. Reconstructed images with and without brain motion correction processing were compared. Visual evaluations by a nuclear medicine specialist and quantitative parameters of images with correction and reference still images were compared. RESULTS: Normalized Mean Squared Error (NMSE) results demonstrated the effectiveness of DDBMC in compensating for rotational and tilting motions during PET imaging. In Cases 1 and 2 involving rotational motion, NMSE decreased from 0.15-0.2 to approximately 0.01 with DDBMC, indicating a substantial reduction in differences from the reference image across various brain regions. In the Structural Similarity Index (SSIM), DDBMC improved it to above 0.96 Contrast assessment revealed notable improvements with DDBMC. In continuous rotational motion, % contrast increased from 42.4% to 73.5%, In tilting motion, % contrast increased from 52.3% to 64.5%, eliminating significant differences from the static reference image. These findings underscore the efficacy of DDBMC in enhancing image contrast and minimizing motion induced variations across different motion scenarios. CONCLUSIONS: DDBMC processing can effectively compensate for continuous rotational and tilting motion of the head during PET, with motion angles of approximately 20 degrees. However, a significant limitation of this study is the exclusive validation of the proposed method using a Hoffman phantom; its applicability to the human brain has not been investigated. Further research involving human subjects is necessary to assess the generalizability and reliability of the presented motion correction technique in real clinical scenarios.

T1-weighted Motion Mitigation in Abdominal MRI: Technical Principles, Clinical Applications, Current Limitations, and Future Prospects.

T1-weighted (T1W) pulse sequences are an indispensable component of clinical protocols in abdominal MRI but usually require multiple breath holds (BHs) during the examination, which not all patients can sustain. Patient motion can affect the quality of T1W imaging so that key diagnostic information, such as intrinsic signal intensity and contrast enhancement image patterns, cannot be determined. Patient motion also has a negative impact on examination efficiency, as multiple acquisition attempts prolong the duration of the examination and often remain noncontributory. Techniques for mitigation of motion-related artifacts at T1W imaging include multiple arterial acquisitions within one BH; free breathing with respiratory gating or respiratory triggering; and radial imaging acquisition techniques, such as golden-angle radial k-space acquisition (stack-of-stars). While each of these techniques has inherent strengths and limitations, the selection of a specific motion-mitigation technique is based on several factors, including the clinical task under investigation, downstream technical ramifications, patient condition, and user preference. The authors review the technical principles of free-breathing motion mitigation techniques in abdominal MRI with T1W sequences, offer an overview of the established clinical applications, and outline the existing limitations of these techniques. In addition, practical guidance for abdominal MRI protocol strategies commonly encountered in clinical scenarios involving patients with limited BH abilities is rendered. Future prospects of free-breathing T1W imaging in abdominal MRI are also discussed. ©RSNA, 2024 See the invited commentary by Fraum and An in this issue.

The GoldX Fiducial Eraser.

Gold nanoparticles with sizes in the range of 5-15 nm are a standard method of providing fiducial markers to assist with alignment during reconstruction in cryogenic electron tomography. However, due to their high electron density and resulting contrast when compared to standard cellular or biological samples, they introduce artifacts such as streaking in the reconstructed tomograms. Here, we demonstrate a tool that automatically detects these nanoparticles and suppresses them by replacing them with a local background as a post-processing step, providing a cleaner tomogram without removing any sample relevant information or introducing new artifacts or edge effects from uniform density replacements.

[A dual-domain cone beam computed tomography reconstruction framework with improved differentiable domain transform for cone-angle artifact correction].

OBJECTIVE: We propose a dual-domain cone beam computed tomography (CBCT) reconstruction framework DualCBR-Net based on improved differentiable domain transform for cone-angle artifact correction. METHODS: The proposed CBCT dual-domain reconstruction framework DualCBR-Net consists of 3 individual modules: projection preprocessing, differentiable domain transform, and image post-processing. The projection preprocessing module first extends the original projection data in the row direction to ensure full coverage of the scanned object by X-ray. The differentiable domain transform introduces the FDK reconstruction and forward projection operators to complete the forward and gradient backpropagation processes, where the geometric parameters correspond to the extended data dimension to provide crucial prior information in the forward pass of the network and ensure the accuracy in the gradient backpropagation, thus enabling precise learning of cone-beam region data. The image post-processing module further fine-tunes the domain-transformed image to remove residual artifacts and noises. RESULTS: The results of validation experiments conducted on Mayo's public chest dataset showed that the proposed DualCBR-Net framework was superior to other comparison methods in terms of artifact removal and structural detail preservation. Compared with the latest methods, the DualCBR-Net framework improved the PSNR and SSIM by 0.6479 and 0.0074, respectively. CONCLUSION: The proposed DualCBR-Net framework for cone-angle artifact correction allows effective joint training of the CBCT dual-domain network and is especially effective for large cone-angle region.

[A deep blur learning-based motion artifact reduction algorithm for dental cone-beam computed tomography images].

OBJECTIVE: We propose a motion artifact correction algorithm (DMBL) for reducing motion artifacts in reconstructed dental cone-beam computed tomography (CBCT) images based on deep blur learning. METHODS: A blur encoder was used to extract motion-related degradation features to model the degradation process caused by motion, and the obtained motion degradation features were imported in the artifact correction module for artifact removal. The artifact correction module adopts a joint learning framework for image blur removal and image blur simulation for treatment of spatially varying and random motion patterns. Comparative experiments were conducted to verify the effectiveness of the proposed method using both simulated motion data sets and clinical data sets. RESULTS: The experimental results with the simulated dataset showed that compared with the existing methods, the PSNR of the proposed method increased by 2.88%, the SSIM increased by 0.89%, and the RMSE decreased by 10.58%. The results with the clinical dataset showed that the proposed method achieved the highest expert level with a subjective image quality score of 4.417 (in a 5-point scale), significantly higher than those of the comparison methods. CONCLUSION: The proposed DMBL algorithm with a deep blur joint learning network structure can effectively reduce motion artifacts in dental CBCT images and achieve high-quality image restoration.

ReMAR: a preoperative CT angiography guided metal artifact reduction framework designed for follow-up CTA of endovascular coiling.

Objective. Follow-up computed tomography angiography (CTA) is necessary for ensuring occlusion effect of endovascular coiling. However, the implanted metal coil will introduce artifacts that have a negative spillover into radiologic assessment.Method. A framework named ReMAR is proposed in this paper for metal artifacts reduction (MARs) from follow-up CTA of patients with coiled aneurysms. It employs preoperative CTA to provide the prior knowledge of the aneurysm and the expected position of the coil as a guidance thus balances the metal artifacts removal performance and clinical feasibility. The ReMAR is composed of three modules: segmentation, registration and MAR module. The segmentation and registration modules obtain the metal coil knowledge via implementing aneurysms delineation on preoperative CTA and alignment of follow-up CTA. The MAR module consisting of hybrid convolutional neural network- and transformer- architectures is utilized to restore sinogram and remove the artifact from reconstructed image. Both image quality and vessel rendering effect after metal artifacts removal are assessed in order to responding clinical concerns.Main results. A total of 137 patients undergone endovascular coiling have been enrolled in the study: 13 of them have complete diagnosis/follow-up records for end-to-end validation, while the rest lacked of follow-up records are used for model training. Quantitative metrics show ReMAR significantly reduced the metal-artifact burden in follow-up CTA. Qualitative ranks show ReMAR could preserve the morphology of blood vessels during artifact removal as desired by doctors.Significance. The ReMAR could significantly remove the artifacts caused by implanted metal coil in the follow-up CTA. It can be used to enhance the overall image quality and convince CTA an alternative to invasive follow-up in treated intracranial aneurysm.

Single-beam digital holographic reconstruction: a phase-support enhanced complex wavefront on phase-only function for twin-image elimination.

Significance: In in-line digital holographic microscopy (DHM), twin-image artifacts pose a significant challenge, and reduction or complete elimination is essential for object reconstruction. Aim: To facilitate object reconstruction using a single hologram, significantly reduce inaccuracies, and avoid iterative processing, a digital holographic reconstruction algorithm called phase-support constraint on phase-only function (PCOF) is presented. Approach: In-line DHM simulations and tabletop experiments employing the sliding-window approach are used to compute the arithmetic mean and variance of the phase values in the reconstructed image. A support constraint mask, through variance thresholding, effectively enabled twin-image artifacts. Results: Quantitative evaluations using metrics such as mean squared error, peak signal-to-noise ratio, and mean structural similarity index show PCOF's superior capability in eliminating twin-image artifacts and achieving high-fidelity reconstructions compared with conventional methods such as angular spectrum and iterative phase retrieval methods. Conclusions: PCOF stands as a promising approach to in-line digital holographic reconstruction, offering a robust solution to mitigate twin-image artifacts and enhance the fidelity of reconstructed objects.

Evaluation of a prototype metal artifact reduction algorithm for cone beam CT in patients undergoing radioembolization.

Metal artifacts notoriously pose significant challenge in computed tomography (CT), leading to inaccuracies in image formation and interpretation. Artifact reduction tools have been designed to improve cone beam computed tomography (CBCT) image quality by reducing artifacts caused by certain high-density materials. Metal artifact reduction (MAR) tools are specific algorithms that are applied during image reconstruction to minimize or eliminate artifacts degrading CBCT images. The purpose of the study is to evaluate the effect of a MAR algorithm on image quality in CBCT performed for evaluating patients before transarterial radioembolization (TARE). We retrospectively included 40 consecutive patients (aged 65 ± 13 years; 23 males) who underwent 45 CBCT examinations (Allura FD 20, XperCT Roll protocol, Philips Healthcare, Best, The Netherlands) in the setting of evaluation for TARE between January 2017 and December 2018. Artifacts caused by coils, catheters, and surgical clips were scored subjectively by four readers on a 5-point scale (1 = artifacts affecting diagnostic information to 5 = no artifacts) using a side-by-side display of uncorrected and MAR-corrected images. In addition, readers scored tumor visibility and vessel discrimination. MAR-corrected images were assigned higher scores, indicating better image quality. The differences between the measurements with and without MAR were most impressive for coils with a mean improvement of 1.6 points (95%CI [1.5 1.8]) on the 5-point likert scale, followed by catheters 1.4 points (95%CI [1.3 1.5]) and clips 0.7 points (95%CI [0.3 1.1]). Improvements for other artifact sources were consistent but relatively small (below 0.25 points on average). Interrater agreement was good to perfect (Kendall's W coefficient = 0.68-0.95) and was higher for MAR-corrected images, indicating that MAR improves diagnostic accuracy. A metal artifact reduction algorithm can improve diagnostic and interventional accuracy of cone beam CT in patients undergoing radioembolization by reducing artifacts caused by diagnostic catheters and coils, lowering interference of metal artifacts with adjacent major structures, and improving tumor visibility.

Conservative method for vertical electrooculogram attenuation based on local suppression of ongoing EEG artifact templates.

Eye movement during blinking can be a significant artifact in Event-Related Potentials (ERP) analysis. Blinks produce a positive potential in the vertical electrooculogram (VEOG), spreading towards the posterior direction. Two methods are frequently used to suppress VEOGs: linear regression to subtract the VEOG signal from the electroencephalogram (EEG) and Independent Component Analysis (ICA). However, some information is lost in both. The present algorithm (1) statistically identifies the position of VEOGs in the frontopolar channels; (2) performs EEG averaging for each channel, which results in 'blink templates'; (3) subtracts each template from the respective EEG at each VEOG position, only when the linear correlation index between the template and the segment is greater than a chosen threshold L. The signals from twenty subjects were acquired using a behavioral test and were treated using FilterBlink for subsequent ERP analysis. A model was designed to test the method for each subject using twenty copies of the EEG signal from the subject's mid-central channel (with nearly no VEOG) representing the EEG channels and their respective blink templates. At the same 200 equidistant time points (marks), a signal (2.5 sinusoidal cycles at 1050 ms emulating an ERP) was mixed with each model channel and the respective blink template of that channel, between 500 to 1200 ms after each mark. According to the model, VEOGs interfered with both ERPs and the ongoing EEG, mainly on the anterior medial leads, and no significant effect was observed on the mid-central channel (Cz). FilterBlink recovered approximately 90% (Fp1) to 98% (Fz) of the original ERP and EEG signals for L = 0.1. The method reduced the VEOG effect on the EEG after ERP and blink-artifact averaging in analyzing real signals. The method is straightforward and effective for VEOG attenuation without significant distortion in the EEG signal and embedded ERPs.

Combining traceological analysis and ZooMS on Early Neolithic bone artefacts from the cave of Coro Trasito, NE Iberian Peninsula: Cervidae used equally to Caprinae.

Few studies have combined the analysis of use-wear traces, traceology, and the proteomic taxonomic identification method Zooarchaeology by Mass Spectrometry (ZooMS). Traceology provides information on the usage, in this case, of bone artefacts, while ZooMS allows for taxonomic identifications where diagnostic features are otherwise gone. The approaches therefore offer complementary information on bone artefacts, allowing for insights into species selection strategies in bone tool manufacture and their subsequent use. Here we present a case study of 20 bone artefacts, mainly bone points, from the Early Neolithic cave site of Coro Trasito located on the southern slope of the Central Pyrenees. Hitherto, studies on Early Neolithic bone artefacts from the Iberian Peninsula have suggested based on morphological assessments that Ovis aries/Capra hircus constituted the majority of the bone material selected for bone tool production. However, the taxonomic identification in this study suggests that, at this site, Cervidae was selected equally to that of O. aries/C. hircus. Furthermore, bone artefacts made from Cervidae specimens seem to be utilised in a wider range of artefact types compared to O. aries/C. hircus. Coro Trasito's bone artefact species composition is probably site-specific to some degree, however, morphological assessments of bone artefacts might not be representative and could be biased towards certain species. Therefore, research on bone artefacts' usage could possibly gain new insights by implementing ZooMS in combination with traceology.

A method for the synchronization of inertial sensor signals and local field potentials from deep brain stimulation systems.

Objective. Recent innovative neurostimulators allow recording local field potentials (LFPs) while performing motor tasks monitored by wearable sensors. Inertial sensors can provide quantitative measures of motor impairment in people with subthalamic nucleus deep brain stimulation. To the best of our knowledge, there is no validated method to synchronize inertial sensors and neurostimulators without an additional device. This study aims to define a new synchronization method to analyze disease-related brain activity patterns during specific motor tasks and evaluate how LFPs are affected by stimulation and medication.Approach. Fourteen male subjects treated with subthalamic nucleus deep brain stimulation were recruited to perform motor tasks in four different medication and stimulation conditions. In each condition, a synchronization protocol was performed consisting of taps on the implanted neurostimulator, which produces artifacts in the LFPs that a nearby inertial sensor can simultaneously record.Main results. In 64% of the recruited subjects, induced artifacts were detected at least in one condition. Among those subjects, 83% of the recordings could be synchronized offline analyzing LFPs and wearables data. The remaining recordings were synchronized by video analysis.Significance. The proposed synchronization method does not require an external system (e.g., TENS electrodes) and can be easily integrated into clinical practice. The procedure is simple and can be carried out in a short time. A proper and simple synchronization will also be useful to analyze subthalamic neural activity in the presence of specific events (e.g., freezing of gait events) to identify predictive biomarkers.

Knotted artifacts in predicted 3D RNA structures.

Unlike proteins, RNAs deposited in the Protein Data Bank do not contain topological knots. Recently, admittedly, the first trefoil knot and some lasso-type conformations have been found in experimental RNA structures, but these are still exceptional cases. Meanwhile, algorithms predicting 3D RNA models have happened to form knotted structures not so rarely. Interestingly, machine learning-based predictors seem to be more prone to generate knotted RNA folds than traditional methods. A similar situation is observed for the entanglements of structural elements. In this paper, we analyze all models submitted to the CASP15 competition in the 3D RNA structure prediction category. We show what types of topological knots and structure element entanglements appear in the submitted models and highlight what methods are behind the generation of such conformations. We also study the structural aspect of susceptibility to entanglement. We suggest that predictors take care of an evaluation of RNA models to avoid publishing structures with artifacts, such as unusual entanglements, that result from hallucinations of predictive algorithms.

Influence of tube current and metal artifact reduction on the diagnosis of external cervical resorption in teeth adjacent to a dental implant in CBCT: an ex-vivo study.

OBJECTIVES: This ex-vivo study aimed to assess the influence of tube current (mA) and metal artifact reduction (MAR) on the diagnosis of early external cervical resorption (EECR) in cone-beam computed tomography (CBCT) in the presence of an adjacent dental implant. MATERIALS AND METHODS: Twenty-three single-rooted teeth were sectioned longitudinally and EECR was induced using a spherical drill and 5% nitric acid in 10 teeth. Each tooth was positioned in the socket of the lower right canine of a dry human mandible and CBCT scans were acquired using 90 kVp, voxel of 0.085 mm, field of view of 5 x 5 cm, and varying tube current (4, 8 or 12 mA), MAR (enabled or disabled) and implant conditions (with a zirconia implant in the socket of the lower right first premolar or without). Five oral radiologists evaluated the presence of EECR in a 5-point scale and the diagnostic values (area under the receiver operating characteristic curve - AUC, sensitivity, and specificity) were compared using multi-way Analysis of Variance (α = 0.05). Kappa test assessed intra-/inter-evaluator agreement. RESULTS: The tube current only influenced the AUC values in the presence of the implant and when MAR disabled; in this case, 8 mA showed lower values (p<0.007). MAR did not influence the diagnostic values (p>0.05). In general, the presence of an implant reduced the AUC values (p<0.0001); sensitivity values with 8 mA and MAR disabled, and specificity values with 4 mA and MAR enabled and 8 mA regardless MAR were also decreased (p<0.0001). CONCLUSIONS: Variations in tube current and MAR were unable to improve EECR detection, which was impaired by the presence of an adjacent implant. CLINICAL RELEVANCE: Increasing tube current or activating MAR tool does not improve EECR diagnosis, which is hampered by the artifacts generated by dental implants.

Test Platform for Developing New Optical Position Tracking Technology towards Improved Head Motion Correction in Magnetic Resonance Imaging.

Optical tracking of head pose via fiducial markers has been proven to enable effective correction of motion artifacts in the brain during magnetic resonance imaging but remains difficult to implement in the clinic due to lengthy calibration and set up times. Advances in deep learning for markerless head pose estimation have yet to be applied to this problem because of the sub-millimetre spatial resolution required for motion correction. In the present work, two optical tracking systems are described for the development and training of a neural network: one marker-based system (a testing platform for measuring ground truth head pose) with high tracking fidelity to act as the training labels, and one markerless deep-learning-based system using images of the markerless head as input to the network. The markerless system has the potential to overcome issues of marker occlusion, insufficient rigid attachment of the marker, lengthy calibration times, and unequal performance across degrees of freedom (DOF), all of which hamper the adoption of marker-based solutions in the clinic. Detail is provided on the development of a custom moiré-enhanced fiducial marker for use as ground truth and on the calibration procedure for both optical tracking systems. Additionally, the development of a synthetic head pose dataset is described for the proof of concept and initial pre-training of a simple convolutional neural network. Results indicate that the ground truth system has been sufficiently calibrated and can track head pose with an error of <1 mm and <1°. Tracking data of a healthy, adult participant are shown. Pre-training results show that the average root-mean-squared error across the 6 DOF is 0.13 and 0.36 (mm or degrees) on a head model included and excluded from the training dataset, respectively. Overall, this work indicates excellent feasibility of the deep-learning-based approach and will enable future work in training and testing on a real dataset in the MRI environment.

Artificial Intelligence-Based Atrial Fibrillation Recognition Method for Motion Artifact-Contaminated Electrocardiogram Signals Preprocessed by Adaptive Filtering Algorithm.

Atrial fibrillation (AF) is a common arrhythmia, and out-of-hospital, wearable, long-term electrocardiogram (ECG) monitoring can help with the early detection of AF. The presence of a motion artifact (MA) in ECG can significantly affect the characteristics of the ECG signal and hinder early detection of AF. Studies have shown that (a) using reference signals with a strong correlation with MAs in adaptive filtering (ADF) can eliminate MAs from the ECG, and (b) artificial intelligence (AI) algorithms can recognize AF when there is no presence of MAs. However, no literature has been reported on whether ADF can improve the accuracy of AI for recognizing AF in the presence of MAs. Therefore, this paper investigates the accuracy of AI recognition for AF when ECGs are artificially introduced with MAs and processed by ADF. In this study, 13 types of MA signals with different signal-to-noise ratios ranging from +8 dB to -16 dB were artificially added to the AF ECG dataset. Firstly, the accuracy of AF recognition using AI was obtained for a signal with MAs. Secondly, after removing the MAs by ADF, the signal was further identified using AI to obtain the accuracy of the AF recognition. We found that after undergoing ADF, the accuracy of AI recognition for AF improved under all MA intensities, with a maximum improvement of 60%.

Narrow band-pass filtered canonical correlation analysis for frequency identification in SSVEP signals.

Steady-state visual evoked potentials (SSVEP) are generated in the parieto-occipital regions, accompanied by background noise and artifacts. A strong pre-processing method is required to reduce this background noise and artifacts. This study proposed a narrow band-pass filtered canonical correlation analysis (NBPFCCA) to recognize frequency components in SSVEP signals. The proposed method is tested on the publicly available 40 stimulus frequencies dataset recorded from 35 subjects and 4 class in-house dataset acquired from 10 subjects. The performance of the proposed NBPFCCA method is compared with the standard canonical correlation analysis (CCA) and the filter bank CCA (FBCCA). The mean frequency detection accuracy of the standard CCA is 86.21% for the benchmark dataset, and it is improved to 95.58% in the proposed method. Results indicate that the proposed method significantly outperforms the standard canonical correlation analysis with an increase of 9.37% and 17% in frequency recognition accuracy of the benchmark and in-house datasets, respectively.