Types Of Images Research Articles

Research and design of image style transfer technology based on multi‐scale convolutional neural network feature fusion

AbstractTo reduce the occurrence of information loss and distortion in image style transfer, a method is proposed for researching and designing image style transfer technology based on multi‐scale convolutional neural networks (CNNs) feature fusion. Initially, the VGG19 model is designed for coarse and fine‐scale networks to achieve multi‐scale CNN feature extraction of target image information. Subsequently, while setting the corresponding feature loss function, an additional least‐squares penalty parameter is introduced to balance the optimal total loss function. Finally, leveraging the characteristics of stochastic gradient descent iteration, image features are fused and reconstructed to obtain better style transfer images. Experimental evaluations utilize peak signal‐to‐noise ratio (PSNR), structural similarity index (SSIM), information entropy (IE), and mean squared error (MSE) as metrics for assessing the transferred images, comparing them with three typical image style transfer methods. Results demonstrate that the proposed method achieves optimal performance across all metrics, realizing superior image style transfer effects.

Focal haemorrhage as atypical feature on MRI in progressive multifocal leukoencephalopathy

Progressive multifocal leukoencephalopathy (PML) is a severe progressive demyelinating disorder caused by the suppression of cell-mediated immunity and is diagnosed by typical imaging features on MRI and confirmed by the identification of the JC virus on CSF. We present a case of HIV positive adult male who presented with progressive hemiparesis and spastic dysarthria over 3 months duration, and MRI showed white matter T2 hyper, T1 isointense and peripherally diffusion restricting lesion in the right frontoparietal region with an additional focal haemorrhage within the lesion, which misled us away from considering PML. However, repeat imaging showed progression of the lesion and CSF analysis showed John Cunningham (JC) virus, confirming the diagnosis.

Is 18F-FDG-PET/CT an Optimal Imaging Modality for Detecting Immune-Related Adverse Events after Immune-Checkpoint Inhibitor Therapy? Pros and Cons.

Immunotherapy with immune checkpoint inhibitors (ICIs) has revolutionized contemporary oncology, presenting efficacy in various solid tumors and lymphomas. However, ICIs may potentially overstimulate the immune system, leading to immune-related adverse events (irAEs). IrAEs may affect multiple organs, such as the colon, stomach, small intestine, kidneys, skin, lungs, joints, liver, lymph nodes, bone marrow, brain, heart, and endocrine glands (e.g., pancreas, thyroid, or adrenal glands), exhibiting autoimmune inflammation. 18F-fluorodeoxyglucose positron emission tomography/computed tomography (18F-FDG PET/CT) is commonly used in oncology for staging and assessment of therapy responses, but it may also serve as a tool for detecting irAEs. This review aims to present various patterns of metabolic activation associated with irAEs due to ICI treatment, identifiable through 18F-FDG PET/CT. It describes the advantages of early detection of irAEs, but also presents the challenges in differentiating them from tumor progression. It also delves into aspects of molecular response assessment within the context of pseudoprogression and hyperprogression, along with typical imaging findings related to these phenomena. Lastly, it summarizes the role of functional PET imaging in oncological immunotherapy, speculating on its future significance and limitations.

#3124 How well do risk assessment guidelines perform for ADPKD?

Abstract Background and Aims The approval of Tolvaptan for treatment of ADPKD heralds a new era when mechanism-based therapy is now possible. However, Tolvaptan is an expensive drug associated with potentially serious side effects and is currently reserved for patients at high-risk for progression to ESKD. Two sets of risk assessment guidelines for ADPKD are now available based on the consensus of two panels of nephrologists from Canada and Europe. However, how well do these guidelines perform in risk assessment has not been formally assessed. Method We conducted a prospective study in 474 patients with typical imaging pattern of ADPKD by MRI who also had detailed clinical and laboratory data. We used age- and height-adjusted total kidney volume to derive the Mayo Clinic Imaging Class as a “gold-standard” for risk assessment (i.e. low-risk: 1A-1B; high-risk: 1C-1E). We then applied the revised Canadian guidelines (Can J Kidney Health Dis. 2018) and the updated European guidelines (NDT 2022) to our patient cohort to assess their performance. Results The study cohort consisted of 286/474 (60%) high-risk patients with MCIC 1C-1E. Applying the updated Canadian risk assessment algorithm resulted in exclusion of 245/474 (52%) patients including 86/286 (30%) of the high-risk patients. The resultant cohort (229/474) was enriched with 88% high-risk patients but also included 12% of low-risk patients. The updated European guidelines provide a 3-step hierarchal algorithm, when applied resulted in exclusion of 55% (157/286) of high-risk patients. During the last step, 122 patients could not be classified due to a lack of eGFR slope information. The resultant cohort (180/474) was enriched with 72% high-risk patients but also included 28%(51/180) of low-risk patients. Conclusion Risk assessment in ADPKD is an evolving process that needs to be redefined by new clinical data and test technologies. Clinical guidelines should be evaluated using real-life data and those that enrich high-risk patients while minimizing low-risk patients have the most clinical utility

Neoplastic and Non-neoplastic Soft Tissue Lesions Around the Knee.

Neoplastic and non-neoplastic soft tissue masses around the knee are often incidental findings. Most of these lesions are benign with typical imaging characteristics that allow a confident diagnosis. However, some of these incidental neoplastic masses are characterized by morbidity and potential mortality. This review highlights the typical aspects of these lesions, facilitating a correct diagnosis.

Cubital tunnel syndrome: anatomy, pathology, and imaging.

Cubital tunnel syndrome (CuTS) is the second most common peripheral neuropathy in the upper limb. It occurs due to ulnar nerve compression within the fibro-osseous cubital tunnel at the elbow joint. Although CuTS is typically diagnosed clinically and with electrodiagnostic studies, the importance of imaging in evaluating the condition is growing. Knowing the typical imaging findings of ulnar nerve entrapment is necessary for precise diagnosis and proper treatment. In this article, we focus on the clinical features, workup and complex imaging of the "anatomic" cubital tunnel and relevant pathological entities.

CORONARY ARTERY ANEURYSM IN IGG4–RELATED DISEASE

Abstract Introduction Coronary artery ectasia is a rare disease defined as a local or diffuse dilatation of the coronary artery more than 1.5 times the diameter of the adiacent normal segment. The etiology is diverse and can be rarely associated with immunoglobulin G4 (IgG4)–related disease. The high levels of serum IgG4 may promote the development of low–density plaques, intimal thickening due to pericoronitis and subsequent coronary ectasia. Case Report We present the case of a 72–year–old male patient with arterial hypertension, dyslipidemia, chronic renal insufficiency, and a family history of cardiovascular diseases. In the previous year, he suffered a cerebral hemorrhage attributed to an unspecified cerebral vascular malformation, which was managed with an endovascular approach. The patient was admitted for Non–ST–segment Elevation Myocardial Infarction (NSTEMI). Coronary angiography revealed widespread ectasia of the coronary vessels, a chronically occluded marginal branch, and an aneurysmatic proximal right coronary artery with the presence of a thrombus on the aneurysmal segment, dissolved following administration of intracoronaric heparin. Furthermore a critical stenosis at the crux involving the posterior interventricular branch was detected and angioplasty with drug–eluting stent implantation was performed on the affected vessel. Given the coronary artery ectasia, aortic ectasia revealed by echocardiography, renal insufficiency and persistently elevated amylase and lipase levels, further diagnostic investigation was conducted suspecting IgG4–related disease. Hematological tests revealed elevated IgG and IgG4 subclass levels (882 mg/dL and 1438 mg/L, respectively), C3 117 mg/dL, C4 30 mg/dL, and total IgE 130 KU/L. The diagnosis of IgG4–related disease was confirmed and the patient was advised to continue cardiological and rheumatological follow–up. Conclusion When the coronary arteries are involved in the IgG4–related disease, risk of acute events such as heart attack may increase and the prognosis is poor. Coronary thickening and ectasia, typical imaging features, other aortic involvement and inflammation indicators, like elevated serum IgG4, should be considered as red flags for an early investigation and subsequent early intervention.

CADASIL-A RARE PRESENTATION OF STROKE

Objective: Cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy (CADASIL) is a hereditary syndrome caused by heterozygous mutations in the NOTCH3 gene that manifests in adulthood and is characterized by recurrent transient ischemic attacks and strokes, migraine-like headaches, psychiatric disturbance, and progressive dementia. Methods: This is a case report. Results: Current case presented with complain of dizziness and vomiting. The diagnosis was suspected mainly because of the typical brain magnetic resonance imaging (MRI). This shows the importance of brain MRI in the diagnosis of CADASIL. Conclusion: Increased awareness of neurologists and neuroradiologists about the typical MRI features of CADASIL is of vital importance to reach the diagnosis in a timely manner.

Integration and performance analysis of artificial intelligence and computer vision based on deep learning algorithms

This paper focuses on the analysis of the application effectiveness of the integration of deep learning and computer vision technologies. Deep learning achieves a historic breakthrough by constructing hierarchical neural networks, enabling end-to-end feature learning and semantic understanding of images. The successful experiences in the field of computer vision provide strong support for training deep learning algorithms. The tight integration of these two fields has given rise to a new generation of advanced computer vision systems, significantly surpassing traditional methods in tasks such as machine vision image classification and object detection. In this paper, typical image classification cases are combined to analyze the superior performance of deep neural network models while also pointing out their limitations in generalization and interpretability, proposing directions for future improvements. Overall, the efficient integration and development trend of deep learning with massive visual data will continue to drive technological breakthroughs and application expansion in the field of computer vision, making it possible to build truly intelligent machine vision systems. This deepening fusion paradigm will powerfully promote unprecedented tasks and functions in computer vision, providing stronger development momentum for related disciplines and industries.

Ultrasound and CT findings of subcutaneous metastases in trunk and pelvis: a comprehensive analysis.

This study aimed to describe the ultrasound, CT findings, and clinical manifestations of pathologically confirmed metastases involving the subcutaneous fat layer of the trunk and pelvis. We included 30 patients with subcutaneous metastases in the trunk and pelvis, verified by ultrasound-guided biopsy. We comprehensively reviewed ultrasound findings of all 30 patients and contrast-enhanced CT findings of 25 patients obtained before biopsy. Medical records were reviewed, including primary malignancy type, presence of coexisting distant metastasis, and detection method leading to biopsy referral. Most subcutaneous metastases were heterogeneously hypoechoic (86.7%) with well-defined margins (80.0%), lobulated (46.7%) or round-to-oval (40.0%) shape, and vascularity (96.7%). Metastases frequently exhibited no contact (53.3%) or focal contact with deep peripheral fascia, resulting in acute contact angle formation (30.0%). Common CT manifestations included central low attenuation with peripheral rim-like enhancement (60.0%) or well-circumscribed lesion with heterogeneous enhancement (32.0%). Lung cancer (46.7%) was the prevalent primary malignancy. CT was the predominant detection method (56.7%). Coexisting subcutaneous metastases were present in 50.0% of cases, and distant metastases (less subcutaneous metastases) were observed in 90.0% of patients. This study describes typical imaging findings of subcutaneous metastases involving the trunk and pelvis. CT may play a crucial role in their early detection, and our results may assist radiologists in their diagnosis.

Image watermarking based on a ratio of DCT coefficient sums using a gradient-based optimizer

Social networks and various websites provide a lot of tools for publishing and sharing digital images. However, publishing digital content online makes it vulnerable to illegal copying and distribution. An effective method for protecting images is watermarking technology, which embeds a logo or some kind of identifier into an image in an invisible way. If necessary, a watermark can be extracted from an image and used to prove authorship and ownership. A major challenge in designing image watermarking schemes is balancing embedding imperceptibility and resistance to various attacks. In this study, we propose a new robust and blind image watermarking algorithm that set a specific ratio of sum of discrete cosine transform (DCT) coefficient absolute values for each image block. We provide a balance between imperceptibility and robustness using a modern metaheuristic optimization algorithm called gradient-based optimizer, which finds the best vector of changes for mid-frequency DCT coefficients. An objective function of optimization algorithm is designed in such a way as to control the ratio of the sums of DCT coefficients absolute values to increase robustness and reduce a distortion level of an image block during embedding. Experimental results and comparison with some other image watermarking schemes demonstrate the high efficiency of the proposed algorithm in terms of imperceptibility and resistance to typical image processing attacks, such as changing brightness and contrast, sharpening, cropping, and others.

Imaging Diagnosis of Thoracic Elastofibroma Dorsi.

Elastofibroma dorsi (ED) is an uncommon benign tumor that is commonly incidentally discovered on thoracic imaging and at times misinterpreted as a more aggressive lesion. The objective of the study is to characterize the typical cross-sectional imaging findings of elastofibroma dorsi and quantify the risk of masquerading malignancy. Retrospective search of radiology and pathology reports over a 12-year period identified 409 cases of suspected ED. Pertinent imaging was reviewed with a focus on computed tomography (CT) and magnetic resonance imaging (MRI), specifically assessing lesion location, presence of interspersed fat, and appearances on follow-up. Typical imaging appearances of 310 ED, including 10% with pathologic confirmation, were that of a mass deep to the serratus anterior (98%) and near the scapular tip (98%). Intralesional interspersed fat was present in 87% of cases imaged with CT and in 90% of cases imaged with MRI. In the 43 cases imaged with both modalities, 8 (19%) did not have interspersed fat on CT, but 7 (88%) of these did have interspersed fat on MRI. Twelve tumors (benign and malignant) were included, of which only 17% were deep to serratus anterior and 25% were at the scapular tip, P = 0.0001 and P < 0.0001 versus ED. Only a single tumor contained interspersed fat, P < 0.001 versus ED, which had benign pathology on biopsy. Elastofibroma dorsi can be diagnosed with a high degree of certainty in the presence of classic location and imaging characteristics, obviating the need for further imaging or biopsy.

PFMNet: Prototype-based feature mapping network for few-shot domain adaptation in medical image segmentation

Lack of data is one of the biggest hurdles for rare disease research using deep learning. Due to the lack of rare-disease images and annotations, training a robust network for automatic rare-disease image segmentation is very challenging. To address this challenge, few-shot domain adaptation (FSDA) has emerged as a practical research direction, aiming to leverage a limited number of annotated images from a target domain to facilitate adaptation of models trained on other large datasets in a source domain. In this paper, we present a novel prototype-based feature mapping network (PFMNet) designed for FSDA in medical image segmentation. PFMNet adopts an encoder–decoder structure for segmentation, with the prototype-based feature mapping (PFM) module positioned at the bottom of the encoder–decoder structure. The PFM module transforms high-level features from the target domain into the source domain-like features that are more easily comprehensible by the decoder. By leveraging these source domain-like features, the decoder can effectively perform few-shot segmentation in the target domain and generate accurate segmentation masks. We evaluate the performance of PFMNet through experiments on three typical yet challenging few-shot medical image segmentation tasks: cross-center optic disc/cup segmentation, cross-center polyp segmentation, and cross-modality cardiac structure segmentation. We consider four different settings: 5-shot, 10-shot, 15-shot, and 20-shot. The experimental results substantiate the efficacy of our proposed approach for few-shot domain adaptation in medical image segmentation.

Pneumonic-type lung adenocarcinoma with KRAS G12V mutation and sustained response to Afatinib

BackgroundPneumonic-type lung adenocarcinoma (P-ADC) is a rare and challenging subtype of primary lung cancer that can be difficult to distinguish from pneumonia based on radiological images. Furthermore, no drugs are currently available that specifically target KRAS G12V.Case presentationHere we report a case of P-ADC with typical and informative imaging features throughout the course of the disease, including patchy shadows, high-density lesions with aerated bronchus, diffuse ground-glass opacities, and nodular shadows from computed tomography (CT) scan. The KRAS G12V mutation was detected using Next-generation sequencing (NGS). An individualized Afatinib-based therapeutic schedule was prescribed and achieved sustained response after multiple lines of treatment had failed.ConclusionOur case highlights the typical and dynamic changes in imaging features of P-ADC and provides an indicative treatment strategy for KRAS G12V-mutated lung adenocarcinoma.

Apparent Resistivity Variation Imaging Method Based on Magnetic Field Gradient by NGO-LSSVM for the Ground-Airborne Frequency-Domain Electromagnetic Method

Apparent resistivity imaging is a typical rapid imaging method in the ground-airborne frequency-domain electromagnetic method. At present, the apparent resistivity is typically calculated by the measured magnetic field, however, this imaging method exhibits limited capability in recognizing the center of three-dimensional anomalies. Therefore, this paper proposed the calculation of apparent resistivity using magnetic field gradients. To solve the problem of random artificial anomalies that existed during the calculated process, this paper presents a hybrid least square support vector machine (LSSVM) and Northern Goshawk optimization (NGO) to establish the mapping relationship between the magnetic field gradient and apparent resistivity variation. This approach enables accurate prediction of apparent resistivity variations and effectively resolves the challenge of correcting background resistivity. Furthermore, three typical theoretical models and field examples are used to predict the apparent resistivity variation, the imaging results demonstrate that the proposed NGO-LSSVM algorithm is a feasible and efficient tool for predicting the apparent resistivity variation with high accuracy. This study provides a novel and efficient imaging method, which facilitates the application of ground-airborne frequency-domain electromagnetics for high-resolution detection requirements, such as mineral exploration.

Osteopetrosis with Typical Radiological Findings: A Report of a Rare Case

BACKGROUND: Osteopetrosis is a rare inherited disease caused by a lack of osteoclastic bone resorption, resulting in increased bone mass with insufficient mechanical strength. Patients usually present with complications such as pathologic fractures, cranial neuropathies, and bone marrow failure. Diagnosis is based on suggestive clinical and imaging findings, with genetic testing to confirm diagnosis and identify the subtype. CASE PRESENTATION: The patient is an eight-year-old girl who came to our hospital with complaints of bilateral arm swelling and visual disturbances for four years and a history of fracture of the left lower extremity two years before presentation. Physical examination revealed nontender bilateral arm swelling and a shorter left leg. The laboratory tests were within normal limits. A skeletal radiograph showed typical radiographic features of osteopetrosis. CONCLUSION: It is important to consider osteopetrosis in patients presenting with pathologic fractures and cranial neuropathies. Typical imaging findings can provide rapid diagnosis in severe cases.

Comparison of Performance in Diagnosis and Characterization of Breast Lesions: Contrast-Enhanced Mammography Versus Breast Magnetic Resonance Imaging

IntroductionIn contemporary medical practice, magnetic resonance imaging (MRI) is the most sensitive modality for detecting breast cancer. Contrast-enhanced mammography (CEM), a relatively recent technology, represents another contrast-enhanced imaging technique that has the potential to serve as an alternative to breast MRI. Our main goal is to compare the diagnostic accuracy including assessment of sensitivity and specificity of these 2 contrast-enhanced breast imaging methods, CEM and MRI, in the diagnosis and characterization of breast lesions. Material and methodsOur prospective study included patients who were clinically suspected of malignancy and/or had suspicious findings detected by mammography or ultrasound. A total of 116 patients were included, and both CEM and MRI examinations were performed on all patients. All CEM examinations were conducted at our institution, while 56.89% of all MRI examinations were carried out at external centers. While histopathological results were accessible for all malignant lesions, the final diagnosis for 80.5% of benign lesions was established through typical imaging findings and adequate follow-up. ResultsThis study encompassed a total of 219 lesions, with 125 out of 219 (57.07%) malignant lesions and 94 out of 219 (42.92%) benign lesions. The sensitivity and specificity values were 98.40% and 81.91%, respectively, for CEM, and 100% and 75.33%, respectively, for MRI. Moreover, CEM showcased comparable performance to MRI in evaluating women with dense breasts. ConclusionCEM and MRI were compared for breast lesion diagnosis, with MRI showing higher sensitivity and CEM higher specificity; however, the differences were not statistically significant.

Detection and recognition of metal surface corrosion based on CBG-YOLOv5s.

The automatic detection of the degree of surface corrosion on metal structures is of significant importance for assessing structural damage and safety. To effectively identify the corrosion status on the surface of coastal metal facilities, this study proposed a CBG-YOLOv5s model for metal surface corrosion detection, based on the YOLOv5s model. Firstly, we integrated the Convolutional Block Attention Module (CBAM) into the C3 module and developed the C3CBAM module. This module effectively enhanced the channel and spatial attention capabilities of the feature map, thereby improving the feature representation. Second, we introduced a multi-scale feature fusion concept in the feature fusion part of the model and added a small target detection layer to improve small target detection. Finally, we designed a lighter C3Ghost module, which reduced the number of parameters and the computational load of the model, thereby improving the running speed of the model. In addition, to verify the effectiveness of our method, we constructed a dataset containing 6000 typical images of metal surface corrosion and conducted extensive experiments on this dataset. The results showed that compared to the YOLOv5s model and several other commonly used object detection models, our method achieved superior performance in terms of detection accuracy and speed.

AoSRNet: All-in-One Scene Recovery Networks via multi-knowledge integration

Scattering and attenuation of light in no-homogeneous imaging media or inconsistent light intensity will cause insufficient contrast and color distortion in the collected images, which limits the developments such as vision-driven smart urban, autonomous vehicles, and intelligent robots. In this paper, we propose an all-in-one scene recovery network via multi-knowledge integration (termed AoSRNet) to improve the visibility of imaging devices in typical low-visibility imaging scenes (e.g., haze, sand dust, underwater, and low light). It combines gamma correction (GC) and optimized linear stretching (OLS) to create the detail enhancement module (DEM) and color restoration module (CRM). Additionally, we suggest a multi-receptive field extraction module (MEM) to attenuate the loss of image texture details caused by GC nonlinear and OLS linear transformations. Finally, we refine the coarse features generated by DEM, CRM, and MEM through Encoder-Decoder to generate the final restored image. Comprehensive experimental results demonstrate the effectiveness and stability of AoSRNet compared to other state-of-the-art methods. The source code is available at https://github.com/LouisYuxuLu/AoSRNet.

A general image fusion framework using multi-task semi-supervised learning

Existing image fusion methods primarily focus on solving single-task fusion problems, overlooking the potential information complementarity among multiple fusion tasks. Additionally, there has been no prior research in the field of image fusion that explores the mixed training of labeled and unlabeled data for different fusion tasks. To address these gaps, this paper introduces a novel multi-task semi-supervised learning approach to construct a general image fusion framework. This framework not only facilitates collaborative training for multiple fusion tasks, thereby achieving effective information complementarity among datasets from different fusion tasks, but also promotes the (unsupervised) learning of unlabeled data via the (supervised) learning of labeled data. Regarding the specific network module, we propose a so-called pseudo-siamese Laplacian pyramid transformer (PSLPT), which can effectively distinguish information at different frequencies in source images and discriminatively fuse features from distinct frequencies. More specifically, we take datasets of four typical image fusion tasks into the same PSLPT for weight updates, yielding the final general fusion model. Extensive experiments demonstrate that the obtained general fusion model exhibits promising outcomes for all four image fusion tasks, both visually and quantitatively. Moreover, comprehensive ablation and discussion experiments corroborate the effectiveness of the proposed method. The code is available at https://github.com/wwhappylife/A-general-image-fusion-framework-using-multi-task-semi-supervised-learning.