Denoising Technique Research Articles

An Effective PDE-based Thresholding for MRI Image Denoising and H-FCM-based Segmentation

Image denoising and segmentation play a crucial role in computer graphics and computer vision. A good image-denoising method must effectively remove noise while preserving important boundaries. Various image-denoising techniques have been employed to remove noise, but complete elimination is often impossible. In this paper, we utilize Partial Differential Equation (PDE) and generalised cross-validation (GCV) within Adaptive Haar Wavelet Transform algorithms to effectively denoise an image, with the digital image serving as the input. After denoising, the image is segmented using the Histon-related fuzzy c-means algorithm (H-FCM), with the processed image serving as the output. The proposed method is tested on images exposed to varying levels of noise. The performance of image denoising and segmentation techniques is evaluated using metrics such as Peak Signal-to-Noise Ratio (PSNR) of 77.42, Mean Squared Error (MSE) of 0.0011, and Structural Similarity Index (SSIM) of 0.7848. Additionally, segmentation performance is measured with a sensitivity of 99%, specificity of 98%, and an accuracy of 98%. The results demonstrate that the proposed methods outperform conventional approaches in these metrics. The implementation of the proposed methods is carried out on the MATLAB platform.

Deep learning-based super-resolution and denoising algorithm improves reliability of dynamic contrast-enhanced MRI in diffuse glioma

Dynamic contrast-enhanced MRI (DCE-MRI) is increasingly used to non-invasively image blood-brain barrier leakage, yet its clinical utility has been hampered by issues such as noise and partial volume artifacts. In this retrospective study involving 306 adult patients with diffuse glioma, we applied deep learning-based super-resolution and denoising (DLSD) techniques to enhance the signal-to-noise ratio (SNR) and resolution of DCE-MRI. Quantitative analysis comparing standard DCE-MRI (std-DCE) and DL-enhanced DCE-MRI (DL-DCE) revealed that DL-DCE achieved significantly higher SNR and contrast-to-noise ratio (CNR) compared to std-DCE (SNR, 52.09 vs 27.21; CNR, 9.40 vs 4.71; P < 0.001 for all). Diagnostic performance assessed by the area under the receiver operating characteristic curve (AUROC) showed improved differentiation of WHO grades based on a pharmacokinetic parameter \\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\hbox {V}_{{e}}$$\\end{document} (AUC, 0.88 vs 0.83, P = 0.02), while remaining comparable to std-DCE in other parameters. Analysis of arterial input function (AIF) reliability demonstrated that \\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\hbox {AIF}_{{DL}}$$\\end{document} exhibited superior agreement compared to \\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$\\hbox {AIF}_{{std}}$$\\end{document}, as indicated by mostly higher intraclass correlation coefficients (Time to peak, 0.79 vs 0.43, P < 0.001). In conclusion, DLSD significantly enhances both the image quality and reliability of DCE-MRI in patients with diffuse glioma, while maintaining or improving diagnostic performance.

Complicated thermo-chemical heterogeneity of the mantle transition zone beneath the Philippine Sea Plate revealed by SS precursors investigation

The Philippine Sea Plate (PSP), a region renowned for its intricate history of multi-stage subduction and back-arc extension, encounters difficulties in elucidating its deep seismic structure, primarily due to the sparse distribution of seismic stations. This study uses over 44,086 traces of SS precursors collected from >1,000 seismic stations over 10–20 years period, to image the mantle transition zone (MTZ) beneath the PSP. With the curvelet denoising technique, we provide high-resolution maps of the depths of the 410 km (D410) and 660 km (D660) discontinuities and the thickness of the MTZ. Notably, we demonstrate a 10–35 km MTZ thickening extending from the West Philippine Basin to the Shikoku Basin, which may be associated with the thermal effect and dehydration of stagnated slabs in the MTZ. Furthermore, the reflection gap and multiple reflectors were both observed in D410 and D660 beneath the Mariana Trench, which suggests that the vertically subducted Pacific plate transports substantial water and non-olivine components into the MTZ in this area. Additionally, a ∼10–25 km MTZ thinning with an abnormally shallow D660 has been observed beneath the Parece Vela Basin and Caroline Plate in the southern PSP, suggesting the potential existence of thermal upwelling from a secondary plume, which may be possibly the tree branch of the Caroline mantle plume in MTZ. Our results provide new seismological constraints on present and past mantle dynamics in and around the PSP.

Enhancing diffusion-weighted prostate MRI through self-supervised denoising and evaluation

Diffusion-weighted imaging (DWI) is a magnetic resonance imaging (MRI) technique that provides information about the Brownian motion of water molecules within biological tissues. DWI plays a crucial role in stroke imaging and oncology, but its diagnostic value can be compromised by the inherently low signal-to-noise ratio (SNR). Conventional supervised deep learning-based denoising techniques encounter challenges in this domain as they necessitate noise-free target images for training. This work presents a novel approach for denoising and evaluating DWI scans in a self-supervised manner, eliminating the need for ground-truth data. By leveraging an adapted version of Stein’s unbiased risk estimator (SURE) and exploiting a phase-corrected combination of repeated acquisitions, we outperform both state-of-the-art self-supervised denoising methods and conventional non-learning-based approaches. Additionally, we demonstrate the applicability of our proposed approach in accelerating DWI scans by acquiring fewer image repetitions. To evaluate denoising performance, we introduce a self-supervised methodology that relies on analyzing the characteristics of the residual signal removed by the denoising approaches.

CEEMDAN combined wavelet denoising improved the MW cycle slip detection algorithm

In the preprocessing of high-precision navigation and positioning data, the most widely used MW combination cycle slip detection method is greatly affected by pseudorange noise. It has issues such as missing small cycle slips and failing to promptly reset the recursive averaging process after cycle slip detection failure, which leads to subsequent threshold divergence. This paper proposes an improved MW combination cycle slip detection method based on Complete Ensemble Empirical Mode Decomposition (CEEMDAN), permutation entropy, and wavelet denoising, which uses CEEMDAN to decompose the cycle slip signal into a series of intrinsic modal functions (IMFs) and then selects the IMFs that require denoising through the permutation entropy algorithm, and the wavelet denoising technique is combined to eliminate the residual noise further, so that the noise can be removed more accurately. Experimental results show that compared with the original MW algorithm, the proposed improved method can effectively reduce the influence of pseudo-range noise, and reduce the false detection rate of cycle slip from 1.6% and 6–0%. All small period slips can be successfully detected in complex noise environments, avoiding the missed detection of the original MW algorithm and the related threshold divergence problems.

Electrical Tree Image De-Noising using Threshold Wavelet Transform and Wiener Filter

Electrical treeing occurred in solid dielectric materials, especially in electrical application with high voltage. The occurrence of electrical tree happens when high electric fields applied, causing tiny channels or paths to form. The main issue during the data collection process is the changes of lighting, making it difficult to study the tree's propagation length, fractal dimension, and growth rate due to corrupted images. This research aims to analyse electrical tree structure images in XLPE material using a CCD camera and develop image de-noising techniques to suppress noise on the electrical tree image. The performance was then analysed using the Otsu thresholding algorithm for accurate segmentation. The methodology was divided into four phases: sample preparation, experimental setup, image pre-processing in MATLAB, and testing four de-noising filters: Wiener, median, NLM, and Gaussian. The Wiener filter with higher PSNR, SNR, and RMSE was selected and using superimposed method, both threshold wavelet transforms and wiener was combined to eliminate the noise. Finally, the proposed method of superimposed was tested with the Otsu thresholding method to evaluate accuracy, sensitivity, and specificity of the combination filter. Based on the analysis of PSNR, SNR, and RMSE, the performance of the threshold wavelet and Wiener filter (TWWF) de-noising technique improves the image quality of the electrical tree structure. Thus, for the Otsu thresholding segmentation algorithm analysis, it also had the highest values in terms of accuracy, sensitivity, and specificity.

Study of Acoustic Emission Signal Noise Attenuation Based on Unsupervised Skip Neural Network.

Acoustic emission (AE) technology, as a non-destructive testing methodology, is extensively utilized to monitor various materials' structural integrity. However, AE signals captured during experimental processes are often tainted with assorted noise factors that degrade the signal clarity and integrity, complicating precise analytical evaluations of the experimental outcomes. In response to these challenges, this paper introduces an unsupervised deep learning-based denoising model tailored for AE signals. It juxtaposes its efficacy against established methods, such as wavelet packet denoising, Hilbert transform denoising, and complete ensemble empirical mode decomposition with adaptive noise denoising. The results demonstrate that the unsupervised skip autoencoder model exhibits substantial potential in noise reduction, marking a significant advancement in AE signal processing. Subsequently, the paper focuses on applying this advanced denoising technique to AE signals collected during the tensile testing of steel fiber-reinforced concrete (SFRC), the tensile testing of steel, and flexural experiments of reinforced concrete beam, and it meticulously discusses the variations in the waveform and the spectrogram of the original signal and the signal after noise reduction. The results show that the model can also remove the noise of AE signals.

Enhancing bone scan image quality: an improved self-supervised denoising approach.

Bone scans play an important role in skeletal lesion assessment, but gamma cameras exhibit challenges with low sensitivity and high noise levels. Deep learning (DL) has emerged as a promising solution to enhance image quality without increasing radiation exposure or scan time. However, existing self-supervised denoising methods, such as Noise2Noise (N2N), may introduce deviations from the clinical standard in bone scans. This study proposes an improved self-supervised denoising technique to minimize discrepancies between DL-based denoising and full scan images. &#xD;Retrospective analysis of 351 whole-body bone scan data sets was conducted. In this study, we used N2N and Noise2FullCount (N2F) denoising models, along with an interpolated version of N2N (iN2N). Denoising networks were separately trained for each reduced scan time from 5 to 50%, and also trained for mixed training datasets, which include all shortened scans. We performed quantitative analysis and clinical evaluation by nuclear medicine experts.&#xD;The denoising networks effectively generated images resembling full scans, with N2F revealing distinctive patterns for different scan times, N2N producing smooth textures with slight blurring, and iN2N closely mirroring full scan patterns. Quantitative analysis showed that denoising improved with longer input times and mixed count training outperformed fixed count training. Traditional denoising methods lagged behind DL-based denoising. N2N demonstrated limitations in long-scan images. Clinical evaluation favored N2N and iN2N in resolution, noise, blurriness, and findings, showcasing their potential for enhanced diagnostic performance in quarter-time scans.&#xD;The improved self-supervised denoising technique presented in this study offers a viable solution to enhance bone scan image quality, minimizing deviations from clinical standards. The method's effectiveness was demonstrated quantitatively and clinically, showing promise for quarter-time scans without compromising diagnostic performance. This approach holds potential for improving bone scan interpretations, aiding in more accurate clinical diagnoses.

Image De-noising Based on WMF Technique for Electrical Trees Structure in High Voltage Cable Insulation

Electrical treeing is a common problem during the pre-breakdown phenomenon in solid insulations due to the damage caused by Partial Discharge (PD) that progresses through stressed insulation via chemical degradation, which resembles the shape of a tree root. This resulted in a decrease in performance through degrading the insulation, which became a serious problem while dealing with electrical equipment. Hence, a deep understanding of electrical tree structure is vital to improving the quality of solid insulations. Ergo, optical microscopy is primarily used to examine tree structures, shapes, and fractal dimensions to reconstruct electrical tree structures for morphological study. However, optical microscopy images are frequently degraded by noise from readout procedures or image data acquisition systems, noise caused by occlusion, illumination, non-uniform intensity, destroying potential tree pixels, and a critical loss of information about the electrical tree structures. Therefore, this research proposed the Wiener Median Fusion (WMF) filter for electrical tree study. The performance of the WMF de-noising technique improves the image quality for the precise portrayal of the electrical tree structure based on thresholding segmentation algorithm analysis in terms of accuracy, sensitivity, and false positive rate. Based on the analysis of the thresholding segmentation algorithm, Otsu's thresholding exhibits the highest result compared to Niblack. The Otsu's overall percentage in terms of accuracy is 80.2934%, the sensitivity is 99.1513%, and the false positive rate is 82.6265%.

A novel chaotic time series wind power point and interval prediction method based on data denoising strategy and improved coati optimization algorithm

Wind power prediction plays a pivotal role in increasing the power grid stability and mitigating market transaction risks. To enhance the prediction accuracy of wind power, this study presents a novel chaotic time series wind power point and interval prediction approach, focusing on data processing, as well as an improved coati optimization algorithm. First, the improved wavelet threshold denoising (IWTD) technique is constructed to eliminate the noise from the original wind power data. Then, the maximal information coefficient (MIC) is adopted to determine the optimal inputs of the prediction model. Subsequently, the improved coati optimization algorithm (ICOA) is developed to optimize the hyperparameters of the bidirectional long short-term memory (BiLSTM), deep belief network (DBN), and gated recurrent unit (GRU) models, along with the linear weight coefficients of the combined model, thereby enhancing the point prediction accuracy. Finally, kernel density estimation (KDE) is employed to obtain interval predictions for wind power with different confidence levels. Results show that the proposed prediction method effectively improves the prediction accuracy compared with other popular prediction models.

Comparative Study of Denoising and Segmentation Techniques for Accurate Brain Tumor Detection in MRI

Background: Brain tumor incidence is on the rise each year, with more than 130 identified types. Precise segmentation models play a vital role in the diagnosis and treatment of brain tumors. This study specifically investigates the utilization of diffusion-based denoising techniques and thresholding methods for segmenting brain tumors from MRI images. Objective: The objective of this study is to examine and compare the efficacy of the Perona-malik and Weickert diffusion techniques in denoising brain MRI images. Additionally, the study aims to assess their performance in threshold-based segmentation of brain tumors. Moreover, it also aims to evaluate the compatibility, benefits, and limitations of the Perona-Malik and Weickert diffusion methods in the denoising of brain MRI images and the effect of denoising on segmentation. Methods: In this study, the Perona-Malik and Weickert diffusion methods are employed to denoise brain MRI images. The denoised images are then subjected to thresholding using both binary and fuzzy approaches, utilizing a triangular membership function. The performance of the diffusion techniques is evaluated using metrics, such as Mean Square Error and Peak Signal to Noise Ratio. Additionally, segmentation models are assessed using metrics such as Dice Similarity Coefficient, Jaccard Similarity Coefficient, and Structural Similarity Measurement Index. Results: The Perona-Malik and Weickert diffusion methods exhibit compatibility with various types of noise, each having its own set of advantages and limitations. The Weickert diffusion method excels in preserving image structure and texture during thresholding. Conclusion: The study provides evidence for the effectiveness of diffusion-based denoising techniques in segmenting brain tumors from MRI images. Specifically, the Weickert diffusion method outperforms in preserving essential image characteristics during thresholding. Additionally, fuzzy thresholding proves to be more successful in accurately segmenting brain tumors. These findings contribute to the advancement of precise models for brain tumor segmentation, ultimately enhancing the diagnosis and treatment of these tumors.

A Comparison of Denoising Approaches for Spoken Word Production Related Artefacts in Continuous Multiband fMRI Data

Abstract It is well-established from fMRI experiments employing gradient echo echo-planar imaging (EPI) sequences that overt speech production introduces signal artefacts compromising accurate detection of task-related responses. Both design and post-processing (denoising) techniques have been proposed and implemented over the years to mitigate the various noise sources. Recently, fMRI studies of speech production have begun to adopt multiband EPI sequences that offer better signal-to-noise ratio (SNR) and temporal resolution allowing adequate sampling of physiological noise sources (e.g., respiration, cardiovascular effects) and reduced scanner acoustic noise. However, these new sequences may also introduce additional noise sources. In this study, we demonstrate the impact of applying several noise-estimation and removal approaches to continuous multiband fMRI data acquired during a naming-to-definition task, including rigid body motion regression and outlier censoring, principal component analysis for removal of cerebrospinal fluid (CSF)/edge-related noise components, and global fMRI signal regression (using two different approaches) compared to a baseline of realignment and unwarping alone. Our results show the strongest and most spatially extensive sources of physiological noise are the global signal fluctuations arising from respiration and muscle action and CSF/edge-related noise components, with residual rigid body motion contributing relatively little variance. Interestingly, denoising approaches tended to reduce and enhance task-related BOLD signal increases and decreases, respectively. Global signal regression using a voxel-wise linear model of the global signal estimated from unmasked data resulted in dramatic improvements in temporal SNR. Overall, these findings show the benefits of combining continuous multiband EPI sequences and denoising approaches to investigate the neurobiology of speech production.

Deformation stage identification in steel material using acoustic emission with a hybrid denoising method and artificial neural network

Acoustic emission (AE) is widely used for identifying source mechanisms and the deformation stage of steel material. The effectiveness of this non-destructive monitoring technique heavily depends on the quality of the measured AE signals. However, the AE signals from deformation are easily contaminated by the signals from noise in a noisy environment. This paper presents a hybrid model for deformation stage identification, which combines a self-adaptive denoising technique and an Artificial neural network (ANN). In pursuit of model generality, AE signals were collected from tensile coupon tests with various steel materials and loading speeds. First, a decomposition-based denoising method is applied based on the singular spectral analysis (SSA) and variational mode decomposition (VMD), which is defined as SSA-VMD. Its effectiveness is demonstrated by simulated signals and experimental results. Following the use of the denoising technique, an ANN is constructed to identify the deformation stage of steel materials with the input of features extracted from the filtered AE signals. The results indicate that the ANN achieves a high prediction accuracy of 0.93 in the test set and 0.87 in unseen data. By applying this denoising method, the ANN-based approach enables accurate correlation of the collected AE signals to deformation stages. The finding can be used as the basis for the creation of new methodologies for monitoring structural health status of in-service steel structures.

Singular value decomposition: A useful technique for image denoising

A key function of image processing is picture denoising, which improves the quality of images by eliminating extraneous noise while keeping crucial information in tact. Singular Value Decomposition (SVD) is a linear algebraic technique that reduces the original datas complexity and scale by breaking down the matrices and extracting the important information. With the power of decomposition which utilizes the non-local self-similarity property of an image to achieve satisfactory denoising performance, SVD denoising has become a potent tool in image processing. In this paper, SVD is outlined and its working, applications, and challenges as a denoising technique in image denoising are discussed. The author discovered that Singular Value Decomposition can be a significant factor in image denoising by applying it to the image. As a result, Singular Value Decomposition could be thought as a helpful image denoising approach in the image processing sequence that will raise the images Peak Signal-to-Noise Ration (PSNR) and improve the quality of the image.

Fast, high-quality, and unshielded 0.2 T low-field mobile MRI using minimal hardware resources.

To propose a deep learning-based low-field mobile MRI strategy for fast, high-quality, unshielded imaging using minimal hardware resources. Firstly, we analyze the correlation of EMI signals between the sensing coil and the MRI coil to preliminarily verify the feasibility of active EMI shielding using a single sensing coil. Then, a powerful deep learning EMI elimination model is proposed, which can accurately predict the EMI components in the MRI coil signals using EMI signals from at least one sensing coil. Further, deep learning models with different task objectives (super-resolution and denoising) are strategically stacked for multi-level post-processing to enable fast and high-quality low-field MRI. Finally, extensive phantom and brain experiments were conducted on a home-built 0.2 T mobile brain scanner for the evaluation of the proposed strategy. 20 healthy volunteers were recruited to participate in the experiment. The results show that the proposed strategy enables the 0.2 T scanner to generate images with sufficient anatomical information and diagnostic value under unshielded conditions using a single sensing coil. In particular, the EMI elimination outperforms the state-of-the-art deep learning methods and numerical computation methods. In addition, 2 × super-resolution (DDSRNet) and denoising (SwinIR) techniques enable further improvements in imaging speed and quality. The proposed strategy enables low-field mobile MRI scanners to achieve fast, high-quality imaging under unshielded conditions using minimal hardware resources, which has great significance for the widespread deployment of low-field mobile MRI scanners.

A Review of Enhancement Techniques for Cone Beam Computed Tomography Images

Cone Beam Computed Tomography (CBCT) has emerged as a valuable imaging modality for various medical applications due to its ability to provide three-dimensional information with minimal radiation exposure. However, CBCT images often suffer from inherent limitations, such as increased noise, artifacts, and reduced spatial resolution. This paper presents a comprehensive review of image processing techniques employed to enhance the quality of CBCT images, addressing the challenges posed by acquisition hardware and image reconstruction algorithms. The review covers a range of preprocessing and post-processing methods, including denoising, artifact correction, and resolution improvement techniques. These methods encompass various mathematical algorithms, machine learning approaches, and hybrid models, which aim to mitigate the imperfections present in CBCT data while preserving diagnostically relevant information. Additionally, this paper discusses the application of deep learning methods, convolutional neural networks, and generative adversarial networks in CBCT image enhancement. These advanced techniques have shown promise in tackling the complex nature of CBCT data and optimizing image quality.

Evaluating FPGA-based denoising techniques for improved signal quality in electrocardiograms

Evaluating FPGA-based denoising techniques for improved signal quality in electrocardiograms

Novel Image Denoising Techniques Using AFMF

Background: In this paper, we have proposed a new image-denoising approach, which is a hybrid technique using the self-organizing migration algorithm (SOMA) and adaptive frequency median filter (AFMF). Materials and Methods: The first step in this approach consists of applying (AFMF) to the noisy image in order to have the first version of the denoised image. This first version of the denoised image is considered a clean image, which is then used as an input of an image-denoising system based on SOMA. This denoising system is then applied for denoising the noisy image and then a final version of the denoised image can be obtained. This image denoising system based on SOMA has two inputs, which are the noisy image and the corresponding clean image. However, we have available only the noisy image, and for that reason, we have first applied the AFMF to the noisy image and then obtained the first version of the denoised image as the clean image. In order to improve this proposed denoising technique, we have replaced the denoising system based on SOMA with targeted image denoising (TID), which is a more recent denoising approach. The PSNR (peak-SNR) and SSIM (structural similarity) have been used for evaluating the performance of the image-denoising techniques proposed in this work. Results: The results obtained from the computations of PSNR and SSIM show the performance of these proposed image-denoising techniques. Conclusion: The results obtained from the computations of PSNR and SSIM show that the proposed image-denoising techniques outperform a number of image-denoising approaches existing in the literature and used here for our evaluation.

Quantitative analysis of laser-generated ultrasonic wave characteristics and their correlation with grain size in polycrystalline materials

Abstract The quantitative relationship between nanosecond pulsed laser parameters and the characteristics of laser-generated ultrasonic waves in polycrystalline materials was evaluated. The high energy of the pulsed laser with a large irradiation spot simultaneously generated ultrasonic longitudinal and shear waves at the epicenter under the slight ablation regime. An optimized denoising technique based on wavelet thresholding and variational mode decomposition was applied to reduce noise in shear waves with a low signal-to-noise ratio. An approach for characterizing grain size was proposed using spectral center frequency ratio (SCFR) based on time-frequency analysis. The results have demonstrated that the generation regime of ultrasonic waves is not solely determined by the laser power density; even at high power densities, a high energy with a large spot can generate an ultrasonic waveform dominated by the thermoelastic effect. It is ascribed to the intensification of the thermoelastic effect with the proportional increase in laser irradiation spot area for a given laser power density. Furthermore, both longitudinal and shear wave SCFRs are linearly related to grain size in polycrystalline materials; however, the shear wave SCFR is more sensitive to finer-grained materials. This study holds great significance for evaluating metal material properties using laser ultrasound.

Al-BERT: a semi-supervised denoising technique for disease prediction.

Medical records are a valuable source for understanding patient health conditions. Doctors often use these records to assess health without solely depending on time-consuming and complex examinations. However, these records may not always be directly relevant to a patient's current health issue. For instance, information about common colds may not be relevant to a more specific health condition. While experienced doctors can effectively navigate through unnecessary details in medical records, this excess information presents a challenge for machine learning models in predicting diseases electronically. To address this, we have developed 'al-BERT', a new disease prediction model that leverages the BERT framework. This model is designed to identify crucial information from medical records and use it to predict diseases. 'al-BERT' operates on the principle that the structure of sentences in diagnostic records is similar to regular linguistic patterns. However, just as stuttering in speech can introduce 'noise' or irrelevant information, similar issues can arise in written records, complicating model training. To overcome this, 'al-BERT' incorporates a semi-supervised layer that filters out irrelevant data from patient visitation records. This process aims to refine the data, resulting in more reliable indicators for disease correlationsand enhancing the model's predictive accuracy and utility in medical diagnostics. To discern noise diseases within patient records, especially those resembling influenza-like illnesses, our approach employs a customized semi-supervised learning algorithm equipped with a focused attention mechanism. This mechanism is specifically calibrated to enhance the model's sensitivity to chronic conditions while concurrently distilling salient features from patient records, thereby augmenting the predictive accuracy and utility of the model in clinical settings. We evaluate the performance of al-BERT using real-world health insurance data provided by Taiwan's National Health Insurance. In our study, we evaluated our model against two others: one based on BERT that uses complete disease records, and another variant that includes extra filtering techniques. Our findings show that models incorporating filtering mechanisms typically perform better than those using the entire, unfiltered dataset. Our approach resulted in improved outcomes across several key measures: AUC-ROC (an indicator of a model's ability to distinguish between classes), precision (the accuracy of positive predictions), recall (the model's ability to find all relevant cases), and overall accuracy. Most notably, our model showed a 15% improvement in recall compared to the current best-performing method in the field of disease prediction. The conducted ablation study affirms the advantages of our attention mechanism and underscores the crucial role of the selection module within al-BERT.