Article
Health Care Sciences & Services
Mohammed Khouy, Younes Jabrane, Mustapha Ameur, Amir Hajjam El Hassani
Summary: Image segmentation is crucial in clinical decision making and has greatly improved medical care. This paper proposes a new approach called GA-UNet that uses genetic algorithms to automatically design a convolutional neural network with good performance and reduced complexity. Experimental results show that GA-UNet achieves competitive performance with smaller architecture and fewer parameters than the original U-Net model.
JOURNAL OF PERSONALIZED MEDICINE
(2023)
Article
Computer Science, Artificial Intelligence
Mehreen Mubashar, Hazrat Ali, Christer Gronlund, Shoaib Azmat
Summary: U-Net is a widely used neural network in the field of medical image segmentation. However, its performance on complex datasets is not satisfactory. To address this issue, several variants such as R2U-Net and UNET++ have been proposed. In this paper, we propose a new U-Net-based medical image segmentation architecture called R2U++, which overcomes the limitations of traditional U-Net by incorporating deeper recurrent residual convolutional blocks and dense skip connections. Experimental results on multiple medical imaging datasets demonstrate that R2U++ achieves significant improvements in IoU and dice scores compared to UNET++ and R2U-Net.
NEURAL COMPUTING & APPLICATIONS
(2022)
Article
Computer Science, Information Systems
Beytullah Sarica, Dursun Zafer Seker, Bulent Bayram
Summary: This study proposes a novel dense residual U-Net model that enhances automatic MS lesion segmentation using 3D MRI sequences. The model combines attention gate (AG), efficient channel attention (ECA), and Atrous Spatial Pyramid Pooling (ASPP) to achieve better results than other state-of-the-art methods. Validation on ISBI2015 and MSSEG2016 challenge datasets shows superior performance.
INTERNATIONAL JOURNAL OF MEDICAL INFORMATICS
(2023)
Article
Medicine, General & Internal
Xiaozhong Tong, Junyu Wei, Bei Sun, Shaojing Su, Zhen Zuo, Peng Wu
Summary: This paper proposed an extended version of U-Net model for skin lesion segmentation using three attention mechanisms. Experimental results demonstrated the method's strong robustness in handling irregular borders, lesions, skin smooth transitions, noise, and artifacts.
Article
Computer Science, Information Systems
V. B. Shereena, G. Raju
Summary: Advances in medical imaging have led to the development of a novel Multi-Residual U-Net model for accurate segmentation of ultrasound medical images. Deep learning methods and optimized pre-processing techniques were utilized to address the challenges of noise and class imbalance. Experimental results demonstrated that the proposed model outperformed other deep models in segmenting tumor regions from ultrasound images.
MULTIMEDIA TOOLS AND APPLICATIONS
(2023)
Article
Chemistry, Multidisciplinary
Xiaohu Zhang, Haifeng Huang
Summary: Given the significance of convolutional neural network-based attention models in road maintenance, a PSNet method with a Parallel Convolution Module (PCM) and Self-Gated Attention Block (SGAB) was proposed. Experimental results demonstrated competitive segmentation performance for crack detection, showing significant improvements over traditional attention models.
APPLIED SCIENCES-BASEL
(2023)
Article
Chemistry, Multidisciplinary
Priscilla Benedetti, Mauro Femminella, Gianluca Reali
Summary: Convolutional neural networks (CNNs) are popular in medical Image Segmentation, and U-Net is widely used, providing cutting-edge results. However, U-Net's performance can be affected by factors like dataset size, performance metrics, image quality, and organ shape and size. This study analyzes the performance using publicly available images and proposes a solution for improving segmentation performance with sparse binary masks without affecting learning process cost.
APPLIED SCIENCES-BASEL
(2023)
Article
Remote Sensing
Mohammad Aghdami-Nia, Reza Shah-Hosseini, Amirhossein Rostami, Saeid Homayouni
Summary: Sea-land segmentation (SLS) is a crucial task in remote sensing for various coastal and environmental studies. This study proposes a modified SUN model and an automatic coastline extraction framework to improve SLS performance. By analyzing different input images and optimizing the SUN architecture, the proposed modifications enhance the accuracy of segmentation results. Experimental results show significant improvements in SLS performance, particularly in datasets from China and the Caspian Sea region.
INTERNATIONAL JOURNAL OF APPLIED EARTH OBSERVATION AND GEOINFORMATION
(2022)
Article
Mathematical & Computational Biology
Meera Srikrishna, Rolf A. Heckemann, Joana B. Pereira, Giovanni Volpe, Anna Zettergren, Silke Kern, Eric Westman, Ingmar Skoog, Michael Scholl
Summary: Brain tissue segmentation is crucial for analyzing brain scans, with CT being a more accessible modality compared to MRI. The study developed and compared 2D and 3D deep learning models for brain tissue classification in CT scans, finding that 2D models performed better than 3D models.
FRONTIERS IN COMPUTATIONAL NEUROSCIENCE
(2022)
Article
Computer Science, Interdisciplinary Applications
Robert Arnar Karlsson, Sveinn Hakon Hardarson
Summary: This paper proposes an automated retinal vessel segmentation method based on convolutional neural networks, which can simultaneously segment the vessels and classify them as arteries or veins. The method achieves good performance on two public datasets and outperforms the state-of-the-art methods in artery vein classification. The appropriateness and applicability of the proposed method are demonstrated through ablation experiments.
COMPUTER METHODS AND PROGRAMS IN BIOMEDICINE
(2022)
Article
Geochemistry & Geophysics
Guang Li, Xiaohui Zhou, Chaojian Chen, Linan Xu, Feng Zhou, Fusheng Shi, Jingtian Tang
Summary: This study proposes a novel intelligent geomagnetic signal denoising method based on the modified U-Net. The improved U-Net combines the advantages of DnCNN and U-Net and achieves a high-precision denoising model through training. Experimental results demonstrate that the improved U-Net can effectively remove various types of noise and outperforms existing denoising methods in real geomagnetic data processing.
IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING
(2023)
Article
Engineering, Multidisciplinary
Roba Gamal, Hoda Barka, Mayada Hadhoud
Summary: Multiple sclerosis is an autoimmune disease that affects the brain and nervous system, with an estimated 2.8 million people worldwide living with the condition. The disease has a pooled incidence rate of 2.1 per 100,000 persons per year across 75 reporting countries, and the average age of diagnosis is 32 years. A novel deep learning model called GAU-U-net, inspired by the widely used U-Net architecture for medical image segmentation, is proposed in this study. By using the GAU-unet architecture, the Dice coefficient increased from 64% to 72% compared to using 3D-Unet, and also showed improvement compared to the Unet-attention network while using fewer model parameters.
ALEXANDRIA ENGINEERING JOURNAL
(2023)
Article
Computer Science, Artificial Intelligence
Jinghua Zhang, Chen Li, Sergey Kosov, Marcin Grzegorzek, Kimiaki Shirahama, Tao Jiang, Changhao Sun, Zihan Li, Hong Li
Summary: In this paper, a novel Low-cost U-Net (LCU-Net) is proposed for environmental microorganism image segmentation task. The LCU-Net addresses the memory cost issue of traditional U-Net and demonstrates effectiveness and potential in practical application.
PATTERN RECOGNITION
(2021)
Article
Computer Science, Artificial Intelligence
Yufei Chen, Chang Xu, Weiping Ding, Shichen Sun, Xiaodong Yue, Hamido Fujita
Summary: In this paper, a target-aware U-Net model with fuzzy skip connection and target attention mechanism is proposed for pancreas segmentation. The method achieves better results compared to other state-of-the-art models on two datasets.
APPLIED SOFT COMPUTING
(2022)
Article
Engineering, Biomedical
Abdullah F. Al-Battal, Imanuel R. Lerman, Truong Q. Nguyen
Summary: In this paper, a multi-path decoder U-Net architecture trained on bounding box segmentation maps is proposed for detecting and localizing anatomical structures in ultrasound scans. This method eliminates the need for expensive and labor-intensive pixel-wise annotations, allowing for training on small datasets and reducing the cost and time needed for deployment. The proposed architecture offers improved performance compared to other architectures with minimal increase in parameters.
COMPUTERIZED MEDICAL IMAGING AND GRAPHICS
(2023)
Review
Radiology, Nuclear Medicine & Medical Imaging
Federico Bruno, Domenico Albano, Andrea Agostini, Massimo Benenati, Roberto Cannella, Damiano Caruso, Michaela Cellina, Diletta Cozzi, Ginevra Danti, Federica De Muzio, Francesco Gentili, Giuliana Giacobbe, Salvatore Gitto, Giulia Grazzini, Irene Grazzini, Carmelo Messina, Anna Palmisano, Pierpaolo Palumbo, Alessandra Bruno, Francesca Grassi, Roberta Grassi, Roberta Fusco, Vincenza Granata, Andrea Giovagnoni, Vittorio Miele, Antonio Barile
Summary: Metabolic and overload disorders are rare but important diseases that affect different organs and tissues. Imaging plays a crucial role in early detection and accurate diagnosis, especially in specific organs involved in metabolic pathways. MRI is particularly useful due to its multiparametric properties, but advanced imaging techniques may also be required for accurate characterization and quantification. This review aims to describe the various alterations resulting from these disorders and their imaging findings.
JAPANESE JOURNAL OF RADIOLOGY
(2023)
Review
Oncology
Vincenza Granata, Roberta Fusco, Sergio Venanzio Setola, Roberta Galdiero, Nicola Maggialetti, Lucrezia Silvestro, Mario De Bellis, Elena Di Girolamo, Giulia Grazzini, Giuditta Chiti, Maria Chiara Brunese, Andrea Belli, Renato Patrone, Raffaele Palaia, Antonio Avallone, Antonella Petrillo, Francesco Izzo
Summary: Pancreatic cancer is one of the deadliest cancers, and late diagnosis is the main reason for its high mortality rate. Surgical resection is the only curative treatment, so early diagnosis is crucial for improving survival. Therefore, it is appropriate to stratify patients based on familial and genetic risk and develop screening protocols using minimally invasive diagnostic tools.
Review
Medicine, General & Internal
Vincenza Granata, Roberta Fusco, Sergio Venanzio Setola, Igino Simonetti, Carmine Picone, Ester Simeone, Lucia Festino, Vito Vanella, Maria Grazia Vitale, Agnese Montanino, Alessandro Morabito, Francesco Izzo, Paolo Antonio Ascierto, Antonella Petrillo
Summary: Immunotherapy is a significant change in oncological treatment, but only a minority of patients benefit from it. The efficacy of immunotherapy is affected by factors such as genetic features and intra-tumor heterogeneity. Classic imaging assessment methods like CT or MRI have limited role in immunotherapy due to different response patterns and the need to assess immunotherapy-related toxic effects promptly.
Review
Medicine, General & Internal
Vincenza Granata, Roberta Fusco, Valeria D'Alessio, Igino Simonetti, Francesca Grassi, Lucrezia Silvestro, Raffaele Palaia, Andrea Belli, Renato Patrone, Mauro Piccirillo, Francesco Izzo
Summary: The aim of this study was to analyze the use of Electrochemotherapy (ECT) in treating primary and secondary liver tumors in different locations and with different histologies. Other Local Ablative Therapies (LAT) were also discussed. The analysis of these papers shows that ECT is safe and effective for treating large lesions, regardless of their histology. Compared to other thermal ablation techniques, ECT performs better in lesions larger than 6 cm and can be safely used for lesions located near vital structures. ECT spares vessels and bile ducts, can be repeated, and can be performed between chemotherapeutic cycles.
Review
Health Care Sciences & Services
Michela Gabelloni, Lorenzo Faggioni, Roberta Fusco, Igino Simonetti, Federica De Muzio, Giuliana Giacobbe, Alessandra Borgheresi, Federico Bruno, Diletta Cozzi, Francesca Grassi, Mariano Scaglione, Andrea Giovagnoni, Antonio Barile, Vittorio Miele, Nicoletta Gandolfo, Vincenza Granata
Summary: Due to the rich vascularization and lymphatic drainage of the pulmonary tissue, lung metastases (LM) are not uncommon in patients with cancer. Radiomics, an active research field aimed at extracting quantitative data from diagnostic images, has potential applications in lesion characterization, treatment planning, and prognostic assessment of patients with LM. This article provides a systematic review of the literature to illustrate the current applications, strengths, and weaknesses of radiomics in this field.
JOURNAL OF PERSONALIZED MEDICINE
(2023)
Review
Health Care Sciences & Services
Carmen Cutolo, Roberta Fusco, Igino Simonetti, Federica De Muzio, Francesca Grassi, Piero Trovato, Pierpaolo Palumbo, Federico Bruno, Nicola Maggialetti, Alessandra Borgheresi, Alessandra Bruno, Giuditta Chiti, Eleonora Bicci, Maria Chiara Brunese, Andrea Giovagnoni, Vittorio Miele, Antonio Barile, Francesco Izzo, Vincenza Granata
Summary: Liver resection is the most effective treatment for primary liver malignancies and metastatic disease. The type of resection depends on various factors, including the type of malignancy, tumor size, and relation with blood and biliary vessels. Imaging, such as ultrasonography, computed tomography, and magnetic resonance imaging, plays a critical role in postoperative assessment and diagnosing complications.
JOURNAL OF PERSONALIZED MEDICINE
(2023)
Review
Radiology, Nuclear Medicine & Medical Imaging
Antonio Galluzzo, Sofia Boccioli, Ginevra Danti, Federica De Muzio, Michela Gabelloni, Roberta Fusco, Alessandra Borgheresi, Vincenza Granata, Andrea Giovagnoni, Nicoletta Gandolfo, Vittorio Miele
Summary: Gastrointestinal stromal tumours, originating from Cajal cells, are rare neoplasms in the gastroenteric tract. Diagnosis is mainly done through endoscopy, echoendoscopy, computed tomography, magnetic resonance imaging, and positron emission tomography. Radiomics, an emerging technique, can extract invisible medical imaging information and convert it into quantitative data, improving diagnosis, treatment, and prognosis of these tumors.
JAPANESE JOURNAL OF RADIOLOGY
(2023)
Review
Oncology
Vincenza Granata, Roberta Fusco, Sergio Venanzio Setola, Roberta Galdiero, Nicola Maggialetti, Renato Patrone, Alessandro Ottaiano, Guglielmo Nasti, Lucrezia Silvestro, Antonio Cassata, Francesca Grassi, Antonio Avallone, Francesco Izzo, Antonella Petrillo
Summary: In this narrative review, the role of radiomics in assessing prognostic features for liver metastases patients is discussed. Radiomics analysis allows the assessment of textural characteristics in radiological images, which can provide biological data without invasive procedures. However, issues such as poor standardization, reproducibility, and clinical study results hamper the translation of radiomics analysis into clinical practice.
INFECTIOUS AGENTS AND CANCER
(2023)
Review
Medicine, General & Internal
Francesca Grassi, Vincenza Granata, Roberta Fusco, Federica De Muzio, Carmen Cutolo, Michela Gabelloni, Alessandra Borgheresi, Ginevra Danti, Carmine Picone, Andrea Giovagnoni, Vittorio Miele, Nicoletta Gandolfo, Antonio Barile, Valerio Nardone, Roberta Grassi
Summary: The role of radiotherapy in the treatment of lung neoplasms, along with surgery and systemic therapies, has become essential. The focus has shifted towards improving survival outcomes, quality of life, treatment compliance, and management of side effects. Imaging plays a crucial role in evaluating treatment efficacy and identifying rare effects, especially when multiple treatments are involved. Radiation recall pneumonitis, a rare complication, needs to be recognized and characterized accurately, requiring prompt identification and the best therapeutic strategy for minimal disruption of ongoing cancer treatment. Artificial intelligence could play a critical role in this regard, provided a larger patient dataset is available.
JOURNAL OF CLINICAL MEDICINE
(2023)
Review
Medicine, General & Internal
Federica De Muzio, Roberta Fusco, Carmen Cutolo, Giuliana Giacobbe, Federico Bruno, Pierpaolo Palumbo, Ginevra Danti, Giulia Grazzini, Federica Flammia, Alessandra Borgheresi, Andrea Agostini, Francesca Grassi, Andrea Giovagnoni, Vittorio Miele, Antonio Barile, Vincenza Granata
Summary: Rectal cancer is a highly lethal malignancy and surgery is the most common treatment option. The choice of surgical approach aims to maximize function while minimizing the risk of recurrence, and is determined by a multidisciplinary team assessing patient and tumor characteristics. Total mesorectal excision, including both low anterior resection and abdominoperineal resection, remains the standard of care for rectal cancer. Rating: 8 out of 10.
JOURNAL OF CLINICAL MEDICINE
(2023)
Review
Medicine, General & Internal
Carmine Picone, Roberta Fusco, Michele Tonerini, Salvatore Claudio Fanni, Emanuele Neri, Maria Chiara Brunese, Roberta Grassi, Ginevra Danti, Antonella Petrillo, Mariano Scaglione, Nicoletta Gandolfo, Andrea Giovagnoni, Antonio Barile, Vittorio Miele, Claudio Granata, Vincenza Granata
Summary: In modern clinical practice, imaging techniques are increasingly used in emergencies, leading to a higher frequency of examinations and increased radiation exposure. The management of pregnant women is particularly critical as they are at higher risk. While ultrasound and magnetic resonance imaging are preferred, computed tomography remains necessary in certain cases. Protocol optimization is crucial in reducing risks. This review aims to evaluate different diagnostic tools and protocols to control radiation dose in emergency conditions involving abdominal pain and trauma.
JOURNAL OF CLINICAL MEDICINE
(2023)
Article
Oncology
Umberto Committeri, Simona Barone, Giovanni Salzano, Antonio Arena, Gerardo Borriello, Francesco Giovacchini, Roberta Fusco, Luigi Angelo Vaira, Alfonso Scarpa, Vincenzo Abbate, Lorenzo Ugga, Pasquale Piombino, Franco Ionna, Luigi Califano, Giovanni Dell'Aversana Orabona
Summary: This study aimed to improve the effectiveness of pre-surgical diagnosis by using a machine learning tool to analyze inflammatory biomarkers and radiomic metrics extracted from MRI images to differentiate between benign and malignant salivary gland tumors. The results showed that both inflammatory biomarkers and radiomic features were capable of supporting a differential diagnosis between different types of tumors.
Review
Medicine, General & Internal
Orlando Catalano, Roberta Fusco, Federica De Muzio, Igino Simonetti, Pierpaolo Palumbo, Federico Bruno, Alessandra Borgheresi, Andrea Agostini, Michela Gabelloni, Carlo Varelli, Antonio Barile, Andrea Giovagnoni, Nicoletta Gandolfo, Vittorio Miele, Vincenza Granata
Summary: Breast ultrasound has made significant technological advancements, transitioning from a low-resolution grayscale technique to a high-performing, multiparametric modality. This review covers the range of commercially available technical tools, including microvasculature imaging, high-frequency transducers, extended field-of-view scanning, elastography, contrast-enhanced ultrasound, and more. It also discusses the expanded applications of breast ultrasound in clinical scenarios, such as primary ultrasound, complementary ultrasound, and second-look ultrasound. The review concludes by acknowledging the ongoing limitations and challenges in breast ultrasound.
Review
Imaging Science & Photographic Technology
Francesca Angelone, Alfonso Maria Ponsiglione, Carlo Ricciardi, Giuseppe Cesarelli, Mario Sansone, Francesco Amato
Summary: In addition to their use for obtaining 3D digital dental models, intraoral scanners (IOSs) have recently shown promise as tools for oral health diagnostics. This review examined the literature on the applications of IOSs as detection systems for oral cavity pathologies, highlighting their potential in areas such as tooth wear, caries, plaques, periodontal defects, and other complications. However, there is limited clinical evidence for the use of IOSs as oral health probes, and further validation is needed.
JOURNAL OF IMAGING
(2023)
Article
Oncology
Mario Sansone, Roberta Fusco, Francesca Grassi, Gianluca Gatta, Maria Paola Belfiore, Francesca Angelone, Carlo Ricciardi, Alfonso Maria Ponsiglione, Francesco Amato, Roberta Galdiero, Roberta Grassi, Vincenza Granata, Roberto Grassi
Summary: The study aims to develop mammographic image processing techniques for the extraction of indicators indicative of breast density (BD) risk factors. The results show that machine learning techniques can classify breasts, with the best classifier being the Support Vector Machine (SVM) with an accuracy of 93.55%, true positive rate of 94.44%, and true negative rate of 92.31%. The SVM with 7 selected features by a wrapper method achieves an accuracy of 0.95, sensitivity of 0.96, and specificity of 0.90 in the external validation cohort. Conclusion: Radiomics analysis and machine learning approach can objectively identify breast density.