Image Properties Research Articles

A novel automatic annotation method for whole slide pathological images combined clustering and edge detection technique

AbstractPixel‐level labeling of regions of interest in an image is a key step in building a labeled training dataset for supervised deep learning networks of images. However, traditional manual labeling of cancerous regions in digital pathological images by doctors is time‐consuming and inefficient. To address this issue, this paper proposes an automatic labeling method for whole slide images, which combines clustering and edge detection techniques. The proposed method utilizes the multi‐level feature fusion model and the Long‐Short Term Memory network to discriminate the cancerous nature of the whole slide images, thereby improving the classification accuracy of the whole slide images. Subsequently, the automatic labeling of cancerous regions is achieved by integrating a density‐based clustering algorithm and an edge point extraction algorithm, both based on the discriminated results of the cancerous properties of whole slide images. The experimental results demonstrate the effectiveness of the proposed method, which offers an efficient and accurate solution to the challenging task of cancerous region labeling in digital pathological images.

Deep learning for head and neck semi-supervised semantic segmentation

Objective. Radiation therapy (RT) represents a prevalent therapeutic modality for head and neck (H&N) cancer. A crucial phase in RT planning involves the precise delineation of organs-at-risks (OARs), employing computed tomography (CT) scans. Nevertheless, the manual delineation of OARs is a labor-intensive process, necessitating individual scrutiny of each CT image slice, not to mention that a standard CT scan comprises hundreds of such slices. Furthermore, there is a significant domain shift between different institutions’ H&N data, which makes traditional semi-supervised learning strategies susceptible to confirmation bias. Therefore, effectively using unlabeled datasets to support annotated datasets for model training has become a critical issue for preventing domain shift and confirmation bias. Approach. In this work, we proposed an innovative cross-domain orthogon-based-perspective consistency (CD-OPC) strategy within a two-branch collaborative training framework, which compels the two sub-networks to acquire valuable features from unrelated perspectives. More specifically, a novel generative pretext task cross-domain prediction (CDP) was designed for learning inherent properties of CT images. Then this prior knowledge was utilized to promote the independent learning of distinct features by the two sub-networks from identical inputs, thereby enhancing the perceptual capabilities of the sub-networks through orthogon-based pseudo-labeling knowledge transfer. Main results. Our CD-OPC model was trained on H&N datasets from nine different institutions, and validated on the four local intuitions’ H&N datasets. Among all datasets CD-OPC achieved more advanced performance than other semi-supervised semantic segmentation algorithms. Significance. The CD-OPC method successfully mitigates domain shift and prevents network collapse. In addition, it enhances the network’s perceptual abilities, and generates more reliable predictions, thereby further addressing the confirmation bias issue.

Self-guided imagery rescripting for worry images: A preliminary experimental investigation.

Mental images of feared events are overactive and intrusive in generalized anxiety disorder (GAD). Imagery rescripting involves integration of positive or neutral imagery and corrective information into images to facilitate emotional processing, reduce imagery intrusions, and re-structure underlying schema. Yet only one known study has applied the technique to treatment of worry. The present study aimed firstto examine the relationship between trait worry and properties of future-oriented worry images, and secondto examine the efficacy of a self-guided imagery rescripting intervention in improving individuals' response to their worries. Participants recruited through Amazon Mechanical Turk (N = 365) identified their major worry and wrote the script of a worst-case scenario mental image. Participants were randomized to three conditions: re-writing the same worry image script (exposure), or writing scripts of either one or three positive alternative future-oriented images (rescripting conditions). In preliminary analyses, trait worry negatively predicted participants' ratings of worry images, including valence and ability to cope, and positively predicted distress, anticipated cost, and belief in their negative meaning. In experimental analyses, linear mixed-effects models revealed anxious response and cognitive appraisal of the threat were significantly lower among participants allocated to rescripting relative to exposure. There was no effect of rescripting type. This investigation demonstrated the impact of a future-oriented imagery rescripting task on anxiety and cognitive biases associated with real worries in an unselected sample. Results may contribute to the development of imagery rescripting interventions for GAD.

The Optimal Model for Copy-Move Forgery Detection in Medical Images.

Digital devices can easily forge medical images. Copy-move forgery detection (CMFD) in medical image has led to abuses in areas where access to advanced medical devices is unavailable. Forgery of the copy-move image directly affects the doctor's decision. The method discussed here is an optimal method for detecting medical image forgery. The proposed method is based on an evolutionary algorithm that can detect fake blocks well. In the first stage, the image is taken to the signal level with the help of a discrete cosine transform (DCT). It is then ready for segmentation by applying discrete wavelet transform (DWT). The low-low band of DWT, which has the most image properties, is divided into blocks. Each block is searched using the equilibrium optimization algorithm. The blocks are most likely to be selected, and the final image is generated. The proposed method was evaluated based on three criteria of precision, recall, and F1 and obtained 90.07%, 92.34%, and 91.56%, respectively. It is superior to the methods studied on medical images. It concluded that our method for CMFD in the medical images was more accurate.

Flashlights: Properties of Highly Magnified Images Near Cluster Critical Curves in the Presence of Dark Matter Subhalos

Dark matter subhalos with extended profiles and density cores, and globular star clusters of mass 106–108 M ⊙ that live near the critical curves in galaxy cluster lenses can potentially be detected through their lensing magnification of stars in background galaxies. In this work, we study the effect such subhalos have on lensed images, and compare to the case of more well-studied microlensing by stars and black holes near critical curves. We find that the cluster density gradient and the extended mass distribution of subhalos are important in determining image properties. Both lead to an asymmetry between the image properties on the positive- and negative-parity sides of the cluster that is more pronounced than in the case of microlensing. For example, on the negative-parity side, subhalos with cores larger than about 50 pc do not generate any images with magnification above ∼100 outside of the immediate vicinity of the cluster critical curve. We discuss these factors using analytical and numerical analysis, and exploit them to identify observable signatures of subhalos: Subhalos create pixel-to-pixel flux variations of ≳0.1 mag on the positive-parity side of clusters. These pixels tend to cluster around (otherwise invisible) subhalos. Unlike in the case of microlensing, signatures of subhalo lensing can be found up to 1″ away from the critical curves of massive clusters.

Simulating image coaddition with the Nancy Grace Roman Space Telescope – I. Simulation methodology and general results

ABSTRACT The upcoming Nancy Grace Roman Space Telescope will carry out a wide-area survey in the near-infrared. A key science objective is the measurement of cosmic structure via weak gravitational lensing. Roman data will be undersampled, which introduces new challenges in the measurement of source galaxy shapes; a potential solution is to use linear algebra-based coaddition techniques such as imcom that combine multiple undersampled images to produce a single oversampled output mosaic with a desired ‘target’ point spread function (PSF). We present here an initial application of imcom to 0.64 square degrees of simulated Roman data, based on the Roman branch of the Legacy Survey of Space and Time (LSST) Dark Energy Science Collaboration (DESC) Data Challenge 2 (DC2) simulation. We show that imcom runs successfully on simulated data that includes features such as plate scale distortions, chip gaps, detector defects, and cosmic ray masks. We simultaneously propagate grids of injected sources and simulated noise fields as well as the full simulation. We quantify the residual deviations of the PSF from the target (the ‘leakage’), as well as noise properties of the output images; we discuss how the overall tiling pattern as well as Moiré patterns appear in the final leakage and noise maps. We include appendices on interpolation algorithms and the interaction of undersampling with image processing operations that may be of broader applicability. The companion paper (‘Paper II’) explores the implications for weak lensing analyses.

Human skin detection: An unsupervised machine learning way

Researchers have been involved for decades in search of an efficient skin detection method. However, current methods have not overcome the significant challenges of skin detection, such as variation of illumination, various skin tones of different ethnic groups, and many others. This research proposed a clustering and region-growing-based skin detection method to overcome these limitations. Together with significant insight, these methods result in a more effective algorithm. The insight concerns the capability to dynamically define the number of clusters in a collection of pixels organized as images. In Clustering for most problem domains, the number of clusters is fixed prior and does not perform effectively over a wide variety of data contents. Therefore, this research paper proposed a skin detection method that validated the above findings. The proposed method assigns the number of clusters based on image properties and ultimately allows freedom from manual thresholds or other manual operations. The dynamic determination of clustering outcomes allows for greater automation of skin detection when dealing with uncertain real-world conditions.

Troubleshooting and Quality Assurance in Hyperpolarized Xenon Magnetic Resonance Imaging: Tools for High-Quality Image Acquisition.

Hyperpolarized (HP) xenon magnetic resonance imaging (129Xe MRI) is a recently federal drug administration (FDA)-approved imaging modality that produces high-resolution images of an inhaled breath of xenon gas for investigation of lung function. However, implementing 129Xe MRI is uniquely challenging as it requires specialized hardware and equipment for hyperpolarization, procurement of xenon imaging coils and coil software, development and compilation of multinuclear MR imaging sequences, and reconstruction/analysis of acquired data. Without proper expertise, these tasks can be daunting, and failure to acquire high-quality images can befrustrating, and expensive. Here, we present some quality control (QC) protocols, troubleshooting practices, and helpful tools for129Xe MRI sites, which may aid in the acquisition of optimized, high-quality data and accurate results. The discussion will begin with an overview of the process for implementing HP 129Xe MRI, including requirements for a hyperpolarizer lab, the combination of 129Xe MRI coil hardware/software, data acquisition and sequence considerations, data structures, k-space and image properties, and measured signal and noise characteristics. Within each of these necessary steps lies opportunities for errors, challenges, and unfavorable occurrences leading to poor image quality or failed imaging, and this presentation aims to address some of the more commonly encountered issues. In particular, identification and characterization of anomalous noise patterns in acquired data are necessary to avoid image artifacts and low-quality images; examples will be given, and mitigation strategies will be discussed. We aim to make the 129Xe MRI implementation process easier for new sites while providing some guidelines and strategies for real-time troubleshooting.

Edge roughness analysis in nanoscale for single-molecule localization microscopy images

Abstract The recent advances in super-resolution fluorescence microscopy, including single-molecule localization microscopy (SMLM), has enabled the study of previously inaccessible details, such as the organization of proteins within cellular compartments and even nanostructures in nonbiological nanomaterials, such as the polymers and semiconductors. With such developments, the need for the development of various computational nanostructure analysis methods for SMLM images is also increasing; however, this has been limited to protein cluster analysis. In this study, we developed an edge structure analysis method for pointillistic SMLM images based on the line edge roughness and power spectral density analyses. By investigating the effect of point properties in SMLM images, such as the size, density, and localization precision on the roughness measurement, we successfully demonstrated this analysis method for experimental SMLM images of actual samples, including the semiconductor line patterns, cytoskeletal elements, and cell membranes. This systematic investigation of the effect of each localization rendering parameter on edge roughness measurement provides a range for the optimal rendering parameters that preserve the relevant nanoscale structure of interest. These new methods are expected to expand our understanding of the targets by providing valuable insights into edge nanoscale structures that have not been previously obtained quantitatively.

The Morphology of Brain Neurons: The Box-Counting Method in the Quantitative Analysis of 2D Images.

This chapter begins by showing the difference between fractal geometry and fractal analysis. The text shows the difference between mathematical and natural fractals and how they are best defined by explaining the concept of fractal analysis. Furthermore, the text presents the most famous technique of fractal analysis: the box-counting method. Defining this method and showing the methodology that leads to the precise value of the fractal (i.e., the box) dimension is done by demonstrating the images of human dentate neurons. A more detailed explanation of the methodology was presented in the previous version of this chapter.This version promotes the notion of monofractal analysis and shows how three types of the same neuronal images can quantify four image properties. The results showed that monofractal parameters successfully quantified four image properties in three nuclei of the cerebellum. Finally, the author discusses the results of this chapter and previously published conclusions. The results show how the monofractal parameters discriminate images of neurons from the three nuclei of the human cerebrum. These outcomes are discussed along with the results of previous studies.

Robust CNN for facial emotion recognition and real-time GUI

<abstract> <p>Computer vision is witnessing a surge of interest in machines accurately recognizing and interpreting human emotions through facial expression analysis. However, variations in image properties such as brightness, contrast, and resolution make it harder for models to predict the underlying emotion accurately. Utilizing a robust architecture of a convolutional neural network (CNN), we designed an efficacious framework for facial emotion recognition that predicts emotions and assigns corresponding probabilities to each fundamental human emotion. Each image is processed with various pre-processing steps before inputting it to the CNN to enhance the visibility and clarity of facial features, enabling the CNN to learn more effectively from the data. As CNNs entail a large amount of data for training, we used a data augmentation technique that helps to enhance the model's generalization capabilities, enabling it to effectively handle previously unseen data. To train the model, we joined the datasets, namely JAFFE and KDEF. We allocated 90% of the data for training, reserving the remaining 10% for testing purposes. The results of the CCN framework demonstrated a peak accuracy of 78.1%, which was achieved with the joint dataset. This accuracy indicated the model's capability to recognize facial emotions with a promising level of performance. Additionally, we developed an application with a graphical user interface for real-time facial emotion classification. This application allows users to classify emotions from still images and live video feeds, making it practical and user-friendly. The real-time application further demonstrates the system's practicality and potential for various real-world applications involving facial emotion analysis.</p> </abstract>

Topologies for Finite Words: Compatibility with the Cantor Topology

Infinite words are often considered as limits of finite words. As topological methods have been proved to be useful in the theory of ω-languages it seems to be providing to include finite and infinite words into one (topological) space. In most cases this results in a poor topological structure induced on the subspace of finite words. In the present paper we investigate the possibility to link topologies in the space of finite words with a topology in the space of infinite words via a natural mapping. A requirement in this linking of topologies consists in the compatibility of the topological properties (openness, closedness etc.) of images with preimages and vice versa. Here, we show that choosing for infinite words the natural topology of the CANTOR space and the δ-limit as linking mapping there are several natural topologies on the space of finite words compatible with the topology of the CANTOR space. It is interesting to observe that besides the well-known prefix topology there are at least two more whose origin is from language theory – centers and supercenters of languages. We show that several of these topologies on the space of finite words fit into a class of ℒ-topologies and exhibit their special properties w.r.t. to the compatibility with the CANTOR topology.

Overview of Essential Components in deep learning reference-based super resolution methods

Reference-based super resolution (RefSR) aims to recover the lost details in a low-resolution image and generate a high-resolution result, guided by a high-resolution reference image with similar contents or textures. In contrast to the traditional single-image super-resolution, which focuses on the intrinsic properties of the single low-resolution image, the challenge of RefSR lies in matching and aggregating highly-related but misaligned reference images with low-resolution images. Several effective but complex designs have been proposed to address this challenge, which poses difficulties in implementing RefSR in real-world applications. In order to better understand the working mechanism of RefSR and design a more efficient and lightweight architecture, we provide a review about the essential components of the existing deep learning-based RefSR. We decompose and classify the common pipeline into four submodules according to the functionalities. Then, we summarize and describe the implementation details of the commonly-adopted approaches in each submodule. Finally, we discuss the challenges and promising research directions.

Multilabel SegSRGAN-A framework for parcellation and morphometry of preterm brain in MRI.

Magnetic resonance imaging (MRI) is a powerful tool for observing and assessing the properties of brain tissue and structures. In particular, in the context of neonatal care, MR images can be used to analyze neurodevelopmental problems that may arise in premature newborns. However, the intrinsic properties of newborn MR images, combined with the high variability of MR acquisition in a clinical setting, result in complex and heterogeneous images. Segmentation methods dedicated to the processing of clinical data are essential for obtaining relevant biomarkers. In this context, the design of quality control protocols for the associated segmentation is a cornerstone for guaranteeing the accuracy and usefulness of these inferred biomarkers. In recent work, we have proposed a new method, SegSRGAN, designed for super-resolution reconstruction and segmentation of specific brain structures. In this article, we first propose an extension of SegSRGAN from binary segmentation to multi-label segmentation, leading then to a partitioning of an MR image into several labels, each corresponding to a specific brain tissue/area. Secondly, we propose a segmentation quality control protocol designed to assess the performance of the proposed method with regard to this specific parcellation task in neonatal MR imaging. In particular, we combine scores derived from expert analysis, morphometric measurements and topological properties of the structures studied. This segmentation quality control can enable clinicians to select reliable segmentations for clinical analysis, starting with correlations between perinatal risk factors, regional volumes and specific dimensions of cognitive development. Based on this protocol, we are investigating the strengths and weaknesses of SegSRGAN and its potential suitability for clinical research in the context of morphometric analysis of brain structure in preterm infants, and to potentially design new biomarkers of neurodevelopment. The proposed study focuses on MR images from the EPIRMEX dataset, collected as part of a national cohort study. In particular, this work represents a first step towards the design of 3-dimensional neonatal brain morphometry based on segmentation. The (free and open-source) code of multilabel SegSRGAN is publicly available at the following URL: https://doi.org/10.5281/zenodo.12659424.

Local work function on graphene nanoribbons.

Graphene nanoribbons show exciting electronic properties related to the exotic nature of the charge carriers and to local confinement as well as atomic-scale structural details. The local work function provides evidence for such structural, electronic, and chemical variations at surfaces. Kelvin prove force microscopy can be used to measure the local contact potential difference (LCPD) between a probe tip and a surface, related to the work function. Here we use this technique to map the LCPD of graphene nanoribbons grown on a Au(111) substrate. The LCPD data shows charge transfer between the graphene nanoribbons and the gold substrate. Our results are corroborated with density functional theory calculations, which verify that the maps reflect the doping of the nanoribbons. Our results help to understand the relation between atomic structure and electronic properties both in high-resolution images and in the distance dependence of the LCPD.

A Fast Iris Liveness Detection for Embedded Systems using Textural Feature Level Fusion Algorithm

Iris recognition is a widely used biometric authentication technique due to its high accuracy and uniqueness. However, iris recognition systems are susceptible to attacks using fake or synthetic iris images, causing a serious security threat. To address this issue, this paper presents a fast iris liveness detection method specifically designed for embedded systems. The proposed method utilizes a textural feature level fusion algorithm using Local Binary Pattern (LBP) and Gray-Level Co-Occurrence Matrix (GLCM) to distinguish between live and printed iris images. LBP captures texture information, while GLCM characterizes the statistical properties of the iris images. By combining these complementary features, the proposed method enhances the discrimination capability and robustness against presentation attacks. Furthermore, to enable real-time and efficient implementation, the proposed liveness detection is optimized and implemented for embedded systems. Experimental results on benchmark datasets demonstrate the effectiveness of the proposed method in accurately detecting iris liveness. The proposed fast iris liveness detection is implemented and optimized on C++ which can be complied and deployed in various embedded devices for iris recognition systems on real-world applications, such as access control, biometric authentication, and surveillance systems.

HPCDNet: Hybrid position coding and dual-frquency domain transform network for low-light image enhancement.

Low-light image enhancement (LLIE) improves lighting to obtain natural normal-light images from images captured under poor illumination. However, existing LLIE methods do not effectively utilize positional and frequency domain image information. To address this limitation, we proposed an end-to-end low-light image enhancement network called HPCDNet. HPCDNet uniquely integrates a hybrid positional coding technique into the self-attention mechanism by appending hybrid positional codes to the query and key, which better retains spatial positional information in the image. The hybrid positional coding can adaptively emphasize important local structures to improve modeling of spatial dependencies within low-light images. Meanwhile, frequency domain image information lost under low-light is recovered via discrete wavelet and cosine transforms. The resulting two frequency domain feature types are weighted and merged using a dual-attention module. More effective use of frequency domain information enhances the network's ability to recreate details, improving visual quality of enhanced low-light images. Experiments demonstrated that our approach can heighten visibility, contrast and color properties of low-light images while better preserving details and textures than previous techniques.

Modular GAN: positron emission tomography image reconstruction using two generative adversarial networks.

The reconstruction of PET images involves converting sinograms, which represent the measured counts of radioactive emissions using detector rings encircling the patient, into meaningful images. However, the quality of PET data acquisition is impacted by physical factors, photon count statistics and detector characteristics, which affect the signal-to-noise ratio, resolution and quantitative accuracy of the resulting images. To address these influences, correction methods have been developed to mitigate each of these issues separately. Recently, generative adversarial networks (GANs) based on machine learning have shown promise in learning the complex mapping between acquired PET data and reconstructed tomographic images. This study aims to investigate the properties of training images that contribute to GAN performance when non-clinical images are used for training. Additionally, we describe a method to correct common PET imaging artefacts without relying on patient-specific anatomical images. The modular GAN framework includes two GANs. Module 1, resembling Pix2pix architecture, is trained on non-clinical sinogram-image pairs. Training data are optimised by considering image properties defined by metrics. The second module utilises adaptive instance normalisation and style embedding to enhance the quality of images from Module 1. Additional perceptual and patch-based loss functions are employed in training both modules. The performance of the new framework was compared with that of existing methods, (filtered backprojection (FBP) and ordered subset expectation maximisation (OSEM) without and with point spread function (OSEM-PSF)) with respect to correction for attenuation, patient motion and noise in simulated, NEMA phantom and human imaging data. Evaluation metrics included structural similarity (SSIM), peak-signal-to-noise ratio (PSNR), relative root mean squared error (rRMSE) for simulated data, and contrast-to-noise ratio (CNR) for NEMA phantom and human data. For simulated test data, the performance of the proposed framework was both qualitatively and quantitatively superior to that of FBP and OSEM. In the presence of noise, Module 1 generated images with a SSIM of 0.48 and higher. These images exhibited coarse structures that were subsequently refined by Module 2, yielding images with an SSIM higher than 0.71 (at least 22% higher than OSEM). The proposed method was robust against noise and motion. For NEMA phantoms, it achieved higher CNR values than OSEM. For human images, the CNR in brain regions was significantly higher than that of FBP and OSEM (p < 0.05, paired t-test). The CNR of images reconstructed with OSEM-PSF was similar to those reconstructed using the proposed method. The proposed image reconstruction method can produce PET images with artefact correction.

A Fast Horizon Detector and a New Annotated Dataset for Maritime Video Processing

Accurate and fast sea horizon detection is vital for tasks in autonomous navigation and maritime security, such as video stabilization, target region reduction, precise tracking, and obstacle avoidance. This paper introduces a novel sea horizon detector from RGB videos, focusing on rapid and effective sea noise suppression while preserving weak horizon edges. Line fitting methods are subsequently employed on filtered edges for horizon detection. We address the filtering problem by extracting line segments with a very low edge threshold, ensuring the detection of line segments even in low-contrast horizon conditions. We show that horizon line segments have simple and relevant properties in RGB images, which we exploit to suppress noisy segments. Then we use the surviving segments to construct a filtered edge map and infer the horizon from the filtered edges. We propose a careful incorporation of temporal in- formation for horizon inference and experimentally show its effectiveness. We address the computational constraint by providing a vectorized implementation for efficient CPU execution, and leveraging image downsizing with minimal loss of accuracy on the original size. Moreover, we contribute a public horizon line dataset to enrich existing data resources. After extensive tests, we report the following major findings: 1) thanks to its filter, our algorithm accurately detects horizon lines with low or weak edge response, 2) the vectorized filter takes no more than 1.71% of the overall computations, while most of the computations are taken by the Line Segment Detection (LSD) algorithm we integrated into our pipeline, 3) our strategy of incorporating the temporal information avoids outlier detections, mitigates the effect of strong noisy lines, and exhibits high robustness when using incorrect detections as a temporal reference. Our algorithm’s performance is rigorously evaluated against state-of-the-art methods, and its core components are validated through ablation experiments.

Data Augmentation for Medical Image Classification based on Gaussian Laplacian Pyramid Blending with a Similarity Measure.

Breast cancer is a devastating disease that affects women worldwide, and computer-aided algorithms have shown potential in automating cancer diagnosis. Recently Generative Artificial Intelligence (GenAI) opens new possibilities for addressing the challenges of labeled data scarcity and accurate prediction in critical applications. However, a lack of diversity, as well as unrealistic and unreliable data, have a detrimental impact on performance. Therefore, this study proposes an augmentation scheme to address the scarcity of labeled data and data imbalance in medical datasets. This approach integrates the concepts of the Gaussian-Laplacian pyramid and pyramid blending with similarity measures. In order to maintain the structural properties of images and capture inter-variability of patient images of the same category similarity-metric-based intermixing has been introduced. It helps to maintain the overall quality and integrity of the dataset. Subsequently, deep learning approach with significant modification, that leverages transfer learning through the usage of concatenated pre-trained models is applied to classify breast cancer histopathological images. The effectiveness of the proposal, including the impact of data augmentation, is demonstrated through a detailed analysis of three different medical datasets, showing significant performance improvement over baseline models. The proposal has the potential to contribute to the development of more accurate and reliable approach for breast cancer diagnosis.