Article
Computer Science, Artificial Intelligence
Kazu Mishiba
Summary: This paper proposes a new method for faster calculation of a weighted median filter. The use of a weight kernel based on the guided filter avoids gradient reversal artifacts and achieves real-time processing and filtering of multidimensional, multichannel, and high precision data. The algorithm based on a linked list reduces the memory requirements and computational cost of updating histograms.
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2023)
Article
Computer Science, Information Systems
Obed Appiah, Michael Asante, James Benjamin Hayfron-Acquah
Summary: Median filter is a predominant filter used to suppress impulse noise with variations focusing on generating quality outputs or reducing running time. The proposed Improved Approximation Median Filtering Algorithms show advantages in both output quality and running time.
JOURNAL OF KING SAUD UNIVERSITY-COMPUTER AND INFORMATION SCIENCES
(2022)
Article
Radiology, Nuclear Medicine & Medical Imaging
Sungho Yun, Uijin Jeong, Donghyeon Lee, Hyeongseok Kim, Seungryong Cho
Summary: In this study, a dual-domain network is proposed to reduce the bowtie-filter-induced artifacts in cone-beam CT (CBCT) images. The network compensates for the filter-induced beam-hardening in the projection domain and further reduces the remaining cupping artifacts associated with scatter in the image domain.
Article
Multidisciplinary Sciences
Lukasz Malinski, Krystian Radlak, Bogdan Smolka
Summary: This study describes a substantial improvement in the efficiency of switching filters for removing impulsive noise within color images. Various noisy pixel detection and replacement techniques were evaluated, showing that the use of more robust detection techniques, combined with novel methods of corrupted pixel restoration, can achieve better image denoising performance. Additionally, the application of a convolutional network surpassed current filtering methods in improving image denoising efficiency.
Article
Radiology, Nuclear Medicine & Medical Imaging
Giovanni Foti, Alessandro Fighera, Antonio Campacci, Simone Natali, Massimo Guerriero, Claudio Zorzi, Giovanni Carbognin
Summary: The study compared the diagnostic performance of dual-energy CT and conventional radiography in detecting hip prosthesis loosening. Dual-energy CT showed higher sensitivity and specificity compared to conventional radiography, with better interobserver agreement between readers.
Article
Engineering, Biomedical
Ruisen Huang, Kunqiang Qing, Dalin Yang, Keum-Shik Hong
Summary: Functional near-infrared spectroscopy (fNIRS) is a technique for non-invasive brain-computer interface (BCI) that requires precise brain signals. This paper proposes a real-time filtering technique to remove motion artifact (MA) and low-frequency drift in fNIRS signals. Experimental results demonstrate that the newly proposed dual-stage median filter method outperforms other compared methods in attenuating MAs and signal distortion.
BIOMEDICAL SIGNAL PROCESSING AND CONTROL
(2022)
Article
Computer Science, Hardware & Architecture
Gilles Perrot, Stephane Domas, Raphael Couturier
Summary: This research aims to demonstrate that a separable approximation of a 2D median filtering is often stronger than its full 2D implementation, which is confirmed by experiments. In addition, a GPU implementation of 2D separable median filters is proposed, achieving the fastest median filtering solution to date.
JOURNAL OF SUPERCOMPUTING
(2022)
Article
Oncology
Hiroaki Kunogi, I-Chow Hsu, Nanae Yamaguchi, Soshi Kusunoki, Keiko Nakagawa, Yayoi Sugimori, Kazunari Fujino, Yasuhisa Terao, Daiki Ogishima, Ryoichi Yoshimura, Keisuke Sasai
Summary: In this study, CT-guided interstitial HDR pelvic nodal brachytherapy showed promising results in treating bulky pelvic nodes in cervical or endometrial cancer patients. The treatment was well tolerated with excellent local control and minimal toxicities, suggesting it as a useful therapeutic option for patients with unresected bulky pelvic nodes.
FRONTIERS IN ONCOLOGY
(2021)
Article
Radiology, Nuclear Medicine & Medical Imaging
Taly Gilat Schmidt, Barbara A. Sammut, Rina Foygel Barber, Xiaochuan Pan, Emil Y. Sidky
Summary: The cOSSCIR algorithm effectively tackles metal artifacts in computed tomography images caused by beam hardening, noise, and photon starvation, by directly estimating basis material maps from photon-counting data. It outperforms a two-step decomposition followed by reconstruction approach in terms of stability and accuracy.
Article
Computer Science, Information Systems
Jiayi Chen, Wentao Zuo, Yinwei Zhan
Summary: An adaptive equidistant median filter was proposed for image restoration in the presence of impulse noise, using a circular neighborhood for noise detection and removal processing. This method significantly improved noise detection and removal performance, outperforming state-of-the-art filters.
MULTIMEDIA TOOLS AND APPLICATIONS
(2022)
Article
Engineering, Biomedical
Shuli Guo, Guowei Wang, Lina Han, Xiaowei Song, Wentao Yang
Summary: This paper presents an improved CT image denoising algorithm to address the issue of confusion between COVID-19 lesions and noise. A median filtering algorithm based on adaptive two-stage threshold and an adaptive weighted median filter image denoising method based on hybrid genetic algorithm are proposed. Simulation results demonstrate the superior performance of the improved algorithm under different noise densities.
BIOMEDICAL SIGNAL PROCESSING AND CONTROL
(2022)
Article
Engineering, Electrical & Electronic
David B. Tay
Summary: Graph wavelet transforms are effective in representing signals over irregular domains by encoding salient features. Nonlinear graph wavelets based on the median operator are proposed to construct more modular and edge-aware transforms. This nonlinear transform has similar spectral characteristics to the linear counterpart and can separate low-pass and high-pass features in a signal. It does not require preprocessing for edge detection and can be applied to nonlinear approximation and denoising.
IEEE TRANSACTIONS ON SIGNAL PROCESSING
(2023)
Article
Radiology, Nuclear Medicine & Medical Imaging
Wenfeng Jin, Yifei Ma, Dan Han, Xiaojie Xie, Weiyuan Zhang, Yan Wu, Guozhi Zhang
Summary: This study aimed to evaluate the effect of using hybrid iterative reconstruction (HIR) and an adaptive filter (AF) on reducing streak artifacts and improving the image quality of neck-and-shoulder computed tomography (CT). The results showed that the joint application of HIR and AF significantly reduced noise and improved signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR). Additionally, it improved the visibility of suspicious lesions, decreased streaking artifacts, and enhanced soft-tissue contrast and visualization of small structures.
JOURNAL OF APPLIED CLINICAL MEDICAL PHYSICS
(2022)
Article
Engineering, Multidisciplinary
Houwang Zhang, Yuan Zhu, Hanying Zheng
Summary: A novel algorithm called NAMF is proposed to remove salt-and-pepper noise from corrupted images. By using a window detector and a nonlocal mean filter, the algorithm achieves better image restoration quality under different levels of SAP noise.
MATHEMATICAL PROBLEMS IN ENGINEERING
(2021)
Article
Oncology
Chao Wang, Jin Mao, Siyu Yang, Huanhuan Xie, Shan Wang, Ling Hu
Summary: This study investigates the impact of adjacent large vessels on the attenuation values of thymic cysts on contrast-enhanced CT. The size, CT attenuation of cysts and adjacent large vessels were measured, and the patients were classified into different groups based on the density changes of the cysts. The study found that the density of thymic cysts is affected by the adjacent large vessels, and understanding this phenomenon can help avoid misdiagnosis and unnecessary thymectomy.
FRONTIERS IN ONCOLOGY
(2022)
Article
Veterinary Sciences
Parminder S. Basran, Jonathon Gao, Scott Palmer, Heidi L. Reesink
Summary: This study developed a radiomics platform to compare features from micro-CTs of proximal sesamoid bones (PSBs) in horses with catastrophic fractures. Radiomics features were highly correlated with previously published data, and imperceptible image features were found to be significantly different between cases and controls using radiomics. The study demonstrated the potential of radiomics in identifying and reproducing differences in image features in PSBs.
EQUINE VETERINARY JOURNAL
(2021)
Editorial Material
Radiology, Nuclear Medicine & Medical Imaging
Clive Baldock, Parminder S. Basran, Habib Zaidi
Article
Agriculture, Dairy & Animal Science
I. R. Porter, M. Wieland, P. S. Basran
Summary: Deep learning technology allows for digital assessment of teat-end condition in dairy cows, providing a systematic and efficient method for evaluation.
JOURNAL OF DAIRY SCIENCE
(2021)
Editorial Material
Engineering, Biomedical
Parminder Basran, Giuseppe Palma, Clive Baldock
PHYSICAL AND ENGINEERING SCIENCES IN MEDICINE
(2021)
Article
Veterinary Sciences
Parminder S. Basran, Ryan B. Appleby
Summary: This article provides insights into the current application and potential of artificial intelligence in veterinary medicine. It highlights the importance of AI in improving veterinary work efficiency and providing new insights for animal treatment.
AMERICAN JOURNAL OF VETERINARY RESEARCH
(2022)
Article
Veterinary Sciences
Ryan B. Appleby, Parminder S. Basran
Summary: Artificial intelligence (AI), a branch of computer science, is reshaping day-to-day life and has significant implications for the field of veterinary medicine. This article discusses the essential elements of AI for veterinary practitioners and emphasizes the importance of veterinary expertise in ensuring data quality and meeting professional needs.
JAVMA-JOURNAL OF THE AMERICAN VETERINARY MEDICAL ASSOCIATION
(2022)
Article
Agriculture, Dairy & Animal Science
Youshan Zhang, Ian R. Porter, Matthias Wieland, Parminder S. Basran
Summary: The health of dairy cows is crucial for milk quality and mammary gland health. Traditional manual assessment of teat-end hyperkeratosis is time-consuming and expensive. In this study, a computer-vision approach using a convolutional neural network is proposed to classify teat-end hyperkeratosis, achieving a substantial improvement in accuracy. Using image-based machine learning to routinely monitor teat-end hyperkeratosis on commercial dairy farms is of great importance.
Article
Veterinary Sciences
Lauren K. Luedke, Phoebe Ilevbare, Kira J. Noordwijk, Pablo M. Palomino, Sean P. McDonough, Scott E. Palmer, Parminder S. Basran, Eve Donnelly, Heidi L. Reesink
Summary: The study aimed to investigate the differences in microdamage and fracture toughness of proximal sesamoid bones (PSBs) between Thoroughbred racehorses with PSB fractures and control group. The results showed that microdamage was mainly detected in the articular region of PSBs, but there was no significant difference in microdamage between PSB fracture and control group. The fracture toughness of PSB flexor cortices also did not differ between the two groups. Therefore, although microdamage is uncommon in Thoroughbred racehorses, it does not affect the fracture toughness of PSBs.
VETERINARY SURGERY
(2022)
Article
Veterinary Sciences
Jasmine Chang, Ian R. Porter, Marnin A. Forman, Natalya Shcherban, Parminder S. Basran
Summary: This study aimed to evaluate the repeatability and agreement of ultrasound measurements of intestinal wall thicknesses in cats. The results showed that there was higher interobserver agreement in segmentations of small intestines compared to measurements of intestinal wall thicknesses.
VETERINARY RADIOLOGY & ULTRASOUND
(2023)
Review
Veterinary Sciences
Del Leary, Parminder S. S. Basran
Summary: Artificial intelligence (AI) systems have the potential to revolutionize veterinary radiation oncology with applications such as treatment simulation, automated segmentation, and automated treatment planning. However, caution must be exercised when adopting AI technologies due to the limited understanding of their efficacy.
VETERINARY RADIOLOGY & ULTRASOUND
(2022)
Review
Veterinary Sciences
Adrien-Maxence Hespel, Youshan Zhang, Parminder S. S. Basran
Summary: This article introduces the essential definitions of AI with medical images and compares common machine learning methods. It provides a detailed description of convolutional neural networks commonly used in deep learning classification and regression models, as well as the utility of natural language processing in machine learning. The goal is to provide veterinarians, veterinary radiologists, and radiation oncologists with the necessary background to understand and comprehend AI-focused research projects and publications.
VETERINARY RADIOLOGY & ULTRASOUND
(2022)
Article
Agriculture, Dairy & Animal Science
Parminder S. Basran, Sean McDonough, Scott Palmer, Heidi L. Reesink
Summary: By analyzing radiomic features in PSBs of horses and using machine learning models, fracture risk can be predicted. Different machine learning algorithms show varying performance in predicting PSB fractures, with Support Vector Machine, Random Forest showing better performance.
Article
Agriculture, Dairy & Animal Science
Kira J. Noordwijk, Leyi Chen, Bianca D. Ruspi, Sydney Schurer, Brittany Papa, Diana C. Fasanello, Sean P. McDonough, Scott E. Palmer, Ian R. Porter, Parminder S. Basran, Eve Donnelly, Heidi L. Reesink
Summary: Identifying imaging features associated with racehorse catastrophic musculoskeletal injuries could improve both jockey and racehorse welfare. This study compared bone mineral properties in cadaver limb specimens obtained from horses sustaining catastrophic bone fracture and controls using multiple imaging modalities. Total high-speed furlong exercise was strongly predictive of bone density and pathologic features. Efforts are underway to investigate diagnostic modalities that could help identify racehorses at increased risk of fracture.
Article
Veterinary Sciences
Parminder S. Basran, Natalya Shcherban, Marnin Forman, Jasmine Chang, Sophie Nelissen, Benjamin K. Recchia, Ian R. Porter
Summary: This study aimed to model and predict feline intestinal diseases using machine-learning approaches, including segmentation of small intestine ultrasound images, complete blood count, and serum biochemical profile data. The average performance of the models for all classifications ranged from 0.504 to 0.886, suggesting high accuracy can be achieved. Including CBC and biochemistry data with US radiomics data did not significantly improve accuracy in the models.
VETERINARY RADIOLOGY & ULTRASOUND
(2023)
Article
Computer Science, Information Systems
Youshan Zhang, Matthias Wieland, Parminder S. Basran
Summary: A novel method using image-based analysis and deep learning is proposed for monitoring the health of dairy cows in large-scale dairy farms. Key frames extracted from videos are used to classify teat-end hyperkeratosis, allowing for efficient assessment of infections and diseases.