In recent years, discussions have been increasingly emerging in modern scientific research that, in connection with the development of AI technologies, questions arise about the objectivity, plausibility and reliability of knowledge, as well as whether these technologies will not replace the expert figure as the authority that has so far acted as a guarantor of objectivity and the center of decision-making. Modern historians of science Duston L. and Galison P. in their book on the history of scientific objectivity, they talk about the alternation of "epistemic virtues", as one of which objectivity has been established since a certain moment. At the same time, the promotion of one or another virtue regulating the scientific self, i.e. acting as a normative principle for a scientist when choosing one or another way of seeing and one or another scientific practice, depends on making decisions in difficult cases requiring the will and limitation of the self. In this sense, epistemology is combined with ethics: a scientist, guided by certain moral principles, gives preference to one or another way of behavior, choosing, for example, not a more accurate hand-drawn image, but an uncluttered photograph, perhaps fuzzy, but obtained mechanically, which means more objective and free from any admixture of subjectivity. In this regard, the epistemic status of modern AI-based technologies, which increasingly assume the functions of the scientific self, including in terms of influencing final decision-making and obtaining objective knowledge, seems interesting. For example, in the field of medicine, robotic devices already provide significant support, taking over some of the functions, for example, of a first-level doctor to collect and analyze standardized patient data and diagnostics. There is an assumption that AI will take on more and more responsibilities in the near future: data processing, development of new drugs and treatment methods, establishing remote interaction with the patient, etc. But does this mean that the scientific self can be replaced by AI-based algorithms, and another epistemic virtue will replace objectivity, finally breaking the link between ethics and epistemology – this question needs to be investigated.
{"title":"The Epistemic Status of AI in Medical Practices: Ethical Challenges","authors":"Angelina Baeva","doi":"10.17816/dd625319","DOIUrl":"https://doi.org/10.17816/dd625319","url":null,"abstract":"In recent years, discussions have been increasingly emerging in modern scientific research that, in connection with the development of AI technologies, questions arise about the objectivity, plausibility and reliability of knowledge, as well as whether these technologies will not replace the expert figure as the authority that has so far acted as a guarantor of objectivity and the center of decision-making. Modern historians of science Duston L. and Galison P. in their book on the history of scientific objectivity, they talk about the alternation of \"epistemic virtues\", as one of which objectivity has been established since a certain moment. At the same time, the promotion of one or another virtue regulating the scientific self, i.e. acting as a normative principle for a scientist when choosing one or another way of seeing and one or another scientific practice, depends on making decisions in difficult cases requiring the will and limitation of the self. In this sense, epistemology is combined with ethics: a scientist, guided by certain moral principles, gives preference to one or another way of behavior, choosing, for example, not a more accurate hand-drawn image, but an uncluttered photograph, perhaps fuzzy, but obtained mechanically, which means more objective and free from any admixture of subjectivity. In this regard, the epistemic status of modern AI-based technologies, which increasingly assume the functions of the scientific self, including in terms of influencing final decision-making and obtaining objective knowledge, seems interesting. For example, in the field of medicine, robotic devices already provide significant support, taking over some of the functions, for example, of a first-level doctor to collect and analyze standardized patient data and diagnostics. There is an assumption that AI will take on more and more responsibilities in the near future: data processing, development of new drugs and treatment methods, establishing remote interaction with the patient, etc. But does this mean that the scientific self can be replaced by AI-based algorithms, and another epistemic virtue will replace objectivity, finally breaking the link between ethics and epistemology – this question needs to be investigated.","PeriodicalId":34831,"journal":{"name":"Digital Diagnostics","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-03-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140254073","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Alexander Arzamastsev, O. Fabrikantov, Elena Valerievna Kulagina, N. Zenkova
Backgraund. Optical coherence tomography (OCT) is a modern high-tech and informative method for detecting pathology of the retina and preretinal layers of the vitreous body. However, the description and interpretation of the research results require high qualifications and special training of an ophthalmologist, and significant time expenditure for the doctor and the patient. At the same time, the use of mathematical models based on artificial neural networks (ANN- models) currently makes it possible to automate many processes associated with image processing. Therefore, solving problems associated with automating the process of classifying OCT images based on ANN models is actual. Aims. To develop architectures of mathematical (computer) models based on deep learning of convolutional neural networks (CNN) for classification of OCT images of the retina. To compare the results of computational experiments conducted using Python tools in the Google Colaboratory with single-model and multi-model approaches and evaluate classification accuracy. To make conclusions about the optimal architecture of ANN models and the values of the hyperparameters used. Materials and methods. The original dataset, which was anonymized OCT images of real patients, included more than 2000 images obtained directly from the device in a resolution of 1920 × 969 × 24 BPP. The number of image classes is 12. To create the training and validation data sets, a subject area of 1100 × 550 × 24 BPP was “cut out.” Various approaches were studied: the possibility of using pretrained CNNs with transfer learning, techniques for resizing and augmenting images, as well as various combinations of hyperparameters of ANN-models. When compiling the model, the following parameters were used: Adam optimizer, categorical_crossentropy loss function, accuracy metric. All technological processes with images and ANN-models were carried out using Python language tools in Google Colaboratory. Results. Single-model and multi-model principles for classifying OCT images of the retina are proposed. Computational experiments on automated classification of such images obtained from a DRI OCT Triton 3D tomograph using various ANN model architectures showed an accuracy of 98-100% during training and validation and 85% during an additional test, which is a satisfactory result. The optimal architecture of the ANN model - a six-layer convolutional network - was selected and the values of its hyperparameters were determined. Conclusions. The results of deep training of convolutional neural network models with various architectures, their validation and testing showed satisfactory classification accuracy of retinal OCT images. These developments can be used in decision support systems in the field of ophthalmology.
后视镜光学相干断层扫描（OCT）是一种检测视网膜和玻璃体视网膜前层病变的现代高科技信息方法。然而，对研究结果的描述和解释需要眼科医生的高素质和特殊培训，医生和病人也需要花费大量时间。与此同时，目前基于人工神经网络（ANN 模型）的数学模型的使用使得许多与图像处理相关的过程实现了自动化。因此，解决与基于人工神经网络模型的 OCT 图像分类过程自动化相关的问题是切实可行的。目标开发基于卷积神经网络（CNN）深度学习的数学（计算机）模型架构，用于视网膜 OCT 图像分类。使用谷歌实验室中的 Python 工具，比较单一模型和多模型方法的计算实验结果，并评估分类准确性。对 ANN 模型的最佳架构和所使用的超参数值做出结论。材料和方法原始数据集是真实患者的匿名 OCT 图像，包括直接从设备获取的 2000 多张图像，分辨率为 1920 × 969 × 24 BPP。图像类别数量为 12 个。为了创建训练和验证数据集，"切出 "了一个 1100 × 550 × 24 BPP 的主题区域。对各种方法进行了研究：使用带迁移学习的预训练 CNN 的可能性、调整图像大小和增强图像的技术，以及 ANN 模型超参数的各种组合。在编制模型时，使用了以下参数：亚当优化器、分类交叉熵损失函数、准确度指标。所有图像和 ANN 模型的技术处理均使用 Google Colaboratory 中的 Python 语言工具进行。结果提出了视网膜 OCT 图像分类的单模型和多模型原则。使用各种 ANN 模型架构对从 DRI OCT Triton 3D 层析成像仪获得的此类图像进行自动分类的计算实验表明，在训练和验证过程中的准确率为 98%-100%，在附加测试中的准确率为 85%，这是一个令人满意的结果。我们选择了最佳结构的方差网络模型（六层卷积网络），并确定了其超参数值。结论采用不同架构的卷积神经网络模型的深度训练、验证和测试结果表明，视网膜 OCT 图像的分类准确率令人满意。这些进展可用于眼科领域的决策支持系统。
{"title":"Classification of optical coherence tomography images using deep machine learning methods","authors":"Alexander Arzamastsev, O. Fabrikantov, Elena Valerievna Kulagina, N. Zenkova","doi":"10.17816/dd623801","DOIUrl":"https://doi.org/10.17816/dd623801","url":null,"abstract":"Backgraund. Optical coherence tomography (OCT) is a modern high-tech and informative method for detecting pathology of the retina and preretinal layers of the vitreous body. However, the description and interpretation of the research results require high qualifications and special training of an ophthalmologist, and significant time expenditure for the doctor and the patient. At the same time, the use of mathematical models based on artificial neural networks (ANN- models) currently makes it possible to automate many processes associated with image processing. Therefore, solving problems associated with automating the process of classifying OCT images based on ANN models is actual. \u0000Aims. To develop architectures of mathematical (computer) models based on deep learning of convolutional neural networks (CNN) for classification of OCT images of the retina. To compare the results of computational experiments conducted using Python tools in the Google Colaboratory with single-model and multi-model approaches and evaluate classification accuracy. To make conclusions about the optimal architecture of ANN models and the values of the hyperparameters used. \u0000Materials and methods. The original dataset, which was anonymized OCT images of real patients, included more than 2000 images obtained directly from the device in a resolution of 1920 × 969 × 24 BPP. The number of image classes is 12. To create the training and validation data sets, a subject area of 1100 × 550 × 24 BPP was “cut out.” Various approaches were studied: the possibility of using pretrained CNNs with transfer learning, techniques for resizing and augmenting images, as well as various combinations of hyperparameters of ANN-models. When compiling the model, the following parameters were used: Adam optimizer, categorical_crossentropy loss function, accuracy metric. All technological processes with images and ANN-models were carried out using Python language tools in Google Colaboratory. \u0000Results. Single-model and multi-model principles for classifying OCT images of the retina are proposed. Computational experiments on automated classification of such images obtained from a DRI OCT Triton 3D tomograph using various ANN model architectures showed an accuracy of 98-100% during training and validation and 85% during an additional test, which is a satisfactory result. The optimal architecture of the ANN model - a six-layer convolutional network - was selected and the values of its hyperparameters were determined. \u0000Conclusions. The results of deep training of convolutional neural network models with various architectures, their validation and testing showed satisfactory classification accuracy of retinal OCT images. These developments can be used in decision support systems in the field of ophthalmology.","PeriodicalId":34831,"journal":{"name":"Digital Diagnostics","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-03-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140254393","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Background: macular diseases present a large group of pathological conditions leading to vision loss and poor vision. Early diagnosis of such changes plays an important role in the treatment tactics and comes one of the crucial factors in predicting results. Aims: study the possibilities of the artificial intelligence program in the diagnosis of the macular diseases based on the scans of structural OCT analysis. Materials and methods: the study included patients undergoing examination and treatment at the Federal Medico-Biological Agency Federal Research Clinical Center and the Moscow Regional Research Clinical Institute named after M.F. Vladimirsky. 200 eyes with the macular diseases and also eyes without macular pathology were examined. A comparative clinical analysis of structural OCT scans performed on an RTVue XR 110-2 tomograph was carried out. Retina AI software was used to analyze OCT scans. Results: during the OCT scans analysis various pathological structures of the macular were identified, and then a probable diagnosis was defined. The obtained results were compared with the diagnosis of the ophthalmologists. The sensitivity of the method was 95.16%; specificity - 97.76%; accuracy - 97.38%. Conclusions: Retina.AI allows ophthalmologists to successfully perform automated analysis of the OCT scans and identify various pathological conditions of the eye fundus.
背景：黄斑疾病是导致视力下降和视力不佳的一大类病变。对这种病变的早期诊断在治疗策略中起着重要作用，也是预测治疗效果的关键因素之一。目的：研究基于 OCT 结构分析扫描的人工智能程序在黄斑疾病诊断中的可能性。材料与方法：研究对象包括在联邦医学生物局联邦临床研究中心和以 M.F. Vladimirsky 命名的莫斯科地区临床研究学院接受检查和治疗的患者。共检查了 200 只患有黄斑疾病的眼睛和没有黄斑病变的眼睛。对在 RTVue XR 110-2 层析成像机上进行的结构性 OCT 扫描进行了临床对比分析。视网膜 AI 软件用于分析 OCT 扫描。结果：在 OCT 扫描分析期间，确定了黄斑的各种病理结构，然后确定了可能的诊断。获得的结果与眼科医生的诊断进行了比较。该方法的灵敏度为 95.16%；特异性为 97.76%；准确性为 97.38%。结论：Retina.AIRetina.AI让眼科医生能够成功地对OCT扫描进行自动分析，并识别眼底的各种病理情况。
{"title":"Explore the possibilities of the artificial intelligence program in the diagnosis of diabetic macular edema, age-related macular degeneration, central serous choriopathy and vitreomacular interface anomalies on the structural optical coherence tomography scans.","authors":"M. Khabazova","doi":"10.17816/dd624131","DOIUrl":"https://doi.org/10.17816/dd624131","url":null,"abstract":"Background: macular diseases present a large group of pathological conditions leading to vision loss and poor vision. Early diagnosis of such changes plays an important role in the treatment tactics and comes one of the crucial factors in predicting results. \u0000Aims: study the possibilities of the artificial intelligence program in the diagnosis of the macular diseases based on the scans of structural OCT analysis. \u0000Materials and methods: the study included patients undergoing examination and treatment at the Federal Medico-Biological Agency Federal Research Clinical Center and the Moscow Regional Research Clinical Institute named after M.F. Vladimirsky. 200 eyes with the macular diseases and also eyes without macular pathology were examined. A comparative clinical analysis of structural OCT scans performed on an RTVue XR 110-2 tomograph was carried out. Retina AI software was used to analyze OCT scans. \u0000Results: during the OCT scans analysis various pathological structures of the macular were identified, and then a probable diagnosis was defined. The obtained results were compared with the diagnosis of the ophthalmologists. The sensitivity of the method was 95.16%; specificity - 97.76%; accuracy - 97.38%. \u0000Conclusions: Retina.AI allows ophthalmologists to successfully perform automated analysis of the OCT scans and identify various pathological conditions of the eye fundus.","PeriodicalId":34831,"journal":{"name":"Digital Diagnostics","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-03-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140254497","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
E. K. Tsyrenzhapova, O. Rozanova, T. Iureva, Andrey A. Ivanov, Ivan S. Rozanov
Backgraund: A thorough analysis of both the optical and anatomical properties of the cornea in patients after anterior radial keratotomy is of particular importance in choosing the optical strength of an intraocular lens in the surgical treatment of cataracts and other types of optical correction. The variability of the clinical picture of postkeratotomic deformity (PCRD) determines the need to develop its classification and is an important task of modern ophthalmology. Aims: to develop an automated system of classification of corneal PCRD using machine learning and an artificial neural network based on the analysis of topographic maps of the cornea. Materials and methods: depersonalized results of the analysis of medical records of 250 patients aged 59.63±5.95 (from 46 to 76) years were used as the material. The analysis of 500 maps of the relief-topography of the anterior and posterior surfaces of the cornea and 3 stages of machine learning of the PCRD classification were carried out. Results: Stage 1- analysis of the relief topography of the anterior and posterior surfaces of the cornea allowed us to fix the numerical values of the elevation of the anterior and posterior surfaces of the cornea in three ring-shaped zones. At stage 2, in the course of deep machine learning, a direct distribution neural network was selected and created. 8 auxiliary parameters describing the shape of the anterior and posterior surfaces of the cornea were established. Stage 3 was accompanied by obtaining algorithms for the classification of PCRD depending on the ratio of test and training samples, which ranged from 75 to 91%.. Conclusion: The use of artificial neural network algorithms can become a useful tool for automatic classification of postkeratotomic corneal deformity in patients who have previously undergone radial keratotomy.
{"title":"MACHINE LEARNING AND ARTIFICIAL NEURAL NETWORK TECHNOLOGIES IN THE CLASSIFICATION OF POSTKERATOTOMIC CORNEAL DEFORMITY","authors":"E. K. Tsyrenzhapova, O. Rozanova, T. Iureva, Andrey A. Ivanov, Ivan S. Rozanov","doi":"10.17816/dd624022","DOIUrl":"https://doi.org/10.17816/dd624022","url":null,"abstract":"Backgraund: A thorough analysis of both the optical and anatomical properties of the cornea in patients after anterior radial keratotomy is of particular importance in choosing the optical strength of an intraocular lens in the surgical treatment of cataracts and other types of optical correction. The variability of the clinical picture of postkeratotomic deformity (PCRD) determines the need to develop its classification and is an important task of modern ophthalmology. \u0000Aims: to develop an automated system of classification of corneal PCRD using machine learning and an artificial neural network based on the analysis of topographic maps of the cornea. \u0000Materials and methods: depersonalized results of the analysis of medical records of 250 patients aged 59.63±5.95 (from 46 to 76) years were used as the material. The analysis of 500 maps of the relief-topography of the anterior and posterior surfaces of the cornea and 3 stages of machine learning of the PCRD classification were carried out. \u0000Results: Stage 1- analysis of the relief topography of the anterior and posterior surfaces of the cornea allowed us to fix the numerical values of the elevation of the anterior and posterior surfaces of the cornea in three ring-shaped zones. At stage 2, in the course of deep machine learning, a direct distribution neural network was selected and created. 8 auxiliary parameters describing the shape of the anterior and posterior surfaces of the cornea were established. Stage 3 was accompanied by obtaining algorithms for the classification of PCRD depending on the ratio of test and training samples, which ranged from 75 to 91%.. \u0000Conclusion: The use of artificial neural network algorithms can become a useful tool for automatic classification of postkeratotomic corneal deformity in patients who have previously undergone radial keratotomy.","PeriodicalId":34831,"journal":{"name":"Digital Diagnostics","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-03-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140251912","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Yuriy A. Vasilev, A. Vladzymyrskyy, K. Arzamasov, David U Shikhmuradov, Andrey V Pankratov, Илья V Ulyanov, Nikolay B Nechaev
Radiology has undergone several significant changes in recent years. Technologies based on computer vision are being actively introduced and allow improving and accelerating the diagnosis of many diseases, as well as reducing the burden on medical personnel. At the same time, these technologies have already proven their effectiveness in routine practice in the analysis of X-ray studies of the mammary glands and chest organs. Also recently, solutions have appeared for the search, qualitative and quantitative evaluation of such common pathologies as urolithiasis and volumetric formations in the parenchyma of the liver and kidneys with a sufficiently high accuracy. Currently, there are many different architectures of deep learning networks and computer vision algorithms that allow identifying and classify the pathology of the abdominal organs. At the same time, all models can be divided into algorithms that distinguish (segment) pathology and algorithms that allow classification of the pathology of the abdominal organs. This review evaluates the existing computer vision algorithms used in computed tomography of the abdominal organs, determines the main directions of their development, and provides prospects for application in medical organizations.
近年来，放射学经历了几次重大变革。以计算机视觉为基础的技术正在被积极引进，这些技术可以改善和加快许多疾病的诊断，并减轻医务人员的负担。同时，这些技术已在乳腺和胸部器官的 X 射线研究分析的常规实践中证明了其有效性。此外，最近还出现了用于搜索、定性和定量评估常见病症的解决方案，如尿路结石以及肝脏和肾脏实质中的体积形成，并具有足够高的准确性。目前，有许多不同架构的深度学习网络和计算机视觉算法可以对腹部器官的病理进行识别和分类。同时，所有模型可分为区分（分割）病理的算法和对腹部器官病理进行分类的算法。本综述对腹部器官计算机断层扫描中使用的现有计算机视觉算法进行了评估，确定了这些算法的主要发展方向，并展望了其在医疗机构中的应用前景。
{"title":"Prospects for the use of computer vision for abdominal CT","authors":"Yuriy A. Vasilev, A. Vladzymyrskyy, K. Arzamasov, David U Shikhmuradov, Andrey V Pankratov, Илья V Ulyanov, Nikolay B Nechaev","doi":"10.17816/dd515814","DOIUrl":"https://doi.org/10.17816/dd515814","url":null,"abstract":"Radiology has undergone several significant changes in recent years. Technologies based on computer vision are being actively introduced and allow improving and accelerating the diagnosis of many diseases, as well as reducing the burden on medical personnel. At the same time, these technologies have already proven their effectiveness in routine practice in the analysis of X-ray studies of the mammary glands and chest organs. Also recently, solutions have appeared for the search, qualitative and quantitative evaluation of such common pathologies as urolithiasis and volumetric formations in the parenchyma of the liver and kidneys with a sufficiently high accuracy. \u0000Currently, there are many different architectures of deep learning networks and computer vision algorithms that allow identifying and classify the pathology of the abdominal organs. At the same time, all models can be divided into algorithms that distinguish (segment) pathology and algorithms that allow classification of the pathology of the abdominal organs. \u0000This review evaluates the existing computer vision algorithms used in computed tomography of the abdominal organs, determines the main directions of their development, and provides prospects for application in medical organizations.","PeriodicalId":34831,"journal":{"name":"Digital Diagnostics","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-03-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140252799","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Yu. A. Trusov, Victoria S. Chupakhina, Adilya S. Nurkaeva, Natalia A. Yakovenko, Irina V. Ablenina, Roksana F. Latypova, Aleksandra P. Pitke, Anastasiya A. Yazovskih, Artem S. Ivanov, Darya S. Bogatyreva, Ulyana A. Popova, Azat F. Yuzlekbaev
Justification. The incidence of diseases of the circulatory system of the population of the Russian Federation has been steadily increasing over the past two decades, increasing 2,047 times from 2000 to 2019. The process of vascular calcification implies the deposition of calcium salts in the artery wall, leading to remodeling of the vascular wall. Radiation research methods are the gold standard for the diagnosis of vascular calcification. However, due to the need for medical professionals to process a large amount of data for a certain period of time, the number of diagnostic errors inevitably increases, as well as the efficiency of work decreases. The active development and introduction of artificial intelligence (AI) into clinical practice has opened up opportunities for specialists to solve these problems. The purpose of the study. To analyze the domestic and foreign literature devoted to the use of AI in the diagnosis of various types of vascular calcification, as well as to summarize the prognostic value of vascular calcification and evaluate aspects that prevent the diagnosis of vascular calcification without the use of AI. Material and methods. The authors searched for publications in the electronic databases PubMed, Web of Science, Google Scholar and eLibrary. The search was carried out using the following keywords: "artificial intelligence", "machine learning", "vascular calcification", "artificial intelligence", "machine learning", "vascular calcification". The search was carried out in the time interval from the moment of the foundation of the corresponding database until July 2023. Conclusion. AI has proven itself well in the diagnosis of vascular calcification. In addition to improving accuracy and efficiency, the ability to detail surpasses the capabilities of the manual diagnostic method. AI has reached a level that allows doctors to help instrumental diagnostics in the automatic detection of vascular calcification. AI capabilities can contribute to the effective development of radiology in the future.
{"title":"THE USE OF ARTIFICIAL INTELLIGENCE IN THE DIAGNOSIS OF ARTERIAL CALCIFICATION","authors":"Yu. A. Trusov, Victoria S. Chupakhina, Adilya S. Nurkaeva, Natalia A. Yakovenko, Irina V. Ablenina, Roksana F. Latypova, Aleksandra P. Pitke, Anastasiya A. Yazovskih, Artem S. Ivanov, Darya S. Bogatyreva, Ulyana A. Popova, Azat F. Yuzlekbaev","doi":"10.17816/dd623196","DOIUrl":"https://doi.org/10.17816/dd623196","url":null,"abstract":"Justification. The incidence of diseases of the circulatory system of the population of the Russian Federation has been steadily increasing over the past two decades, increasing 2,047 times from 2000 to 2019. The process of vascular calcification implies the deposition of calcium salts in the artery wall, leading to remodeling of the vascular wall. Radiation research methods are the gold standard for the diagnosis of vascular calcification. However, due to the need for medical professionals to process a large amount of data for a certain period of time, the number of diagnostic errors inevitably increases, as well as the efficiency of work decreases. The active development and introduction of artificial intelligence (AI) into clinical practice has opened up opportunities for specialists to solve these problems. \u0000The purpose of the study. To analyze the domestic and foreign literature devoted to the use of AI in the diagnosis of various types of vascular calcification, as well as to summarize the prognostic value of vascular calcification and evaluate aspects that prevent the diagnosis of vascular calcification without the use of AI. \u0000Material and methods. The authors searched for publications in the electronic databases PubMed, Web of Science, Google Scholar and eLibrary. The search was carried out using the following keywords: \"artificial intelligence\", \"machine learning\", \"vascular calcification\", \"artificial intelligence\", \"machine learning\", \"vascular calcification\". The search was carried out in the time interval from the moment of the foundation of the corresponding database until July 2023. \u0000Conclusion. AI has proven itself well in the diagnosis of vascular calcification. In addition to improving accuracy and efficiency, the ability to detail surpasses the capabilities of the manual diagnostic method. AI has reached a level that allows doctors to help instrumental diagnostics in the automatic detection of vascular calcification. AI capabilities can contribute to the effective development of radiology in the future.","PeriodicalId":34831,"journal":{"name":"Digital Diagnostics","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-01-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140486367","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Y. Vasilev, Olga G. Nanova, I. A. Blokhin, R. V. Reshetnikov, A. Vladzymyrskyy, O. Omelyanskaya
Radiomics is the newest and promising direction in modern radiographic diagnostics. The number of head and neck cancer studies with employing radiomics is increasing every year. We performed a systematic review of recent publications (20212023) on computed tomography (CT)-based head and neck malignancies. The search for articles was carried out in the PubMed database. The basic characteristics of the selected articles were extracted and their quality was assessed with RQS 2.0 [3] and the modified QUADAS-CAD questionnaire [17]. We assessed the level of reproducibility of radiomic parameters selected for predictive models in different studies. Eleven articles were selected for our review. In most cases, there was a high risk of systematic error associated with the data imbalance in terms of demographic parameters and level of pathologies. The range of RQS 2.0 scores for the included articles varies from 19.44% to 50.00% of the maximum possible score. The main problems leading to researches quality decreasing are the lack of external validation of the results (73% of the analyzed articles) and the lack of data accessibility and transparency (82%). Inter-study reproducibility of radiomics parameters is low due to the wide variety of techniques used for image acquisition, image post-processing, extraction and statistical processing of radiomics parameters. The basic block of the stable radiomics parameters should be created for the method introducing into clinical practice. The radiomics methods standardization and creating an open radiomics database creation is necessary for this purpose.
{"title":"Priority radiomics parameters for computed tomography in head and neck malignancies: a systematic review","authors":"Y. Vasilev, Olga G. Nanova, I. A. Blokhin, R. V. Reshetnikov, A. Vladzymyrskyy, O. Omelyanskaya","doi":"10.17816/dd623240","DOIUrl":"https://doi.org/10.17816/dd623240","url":null,"abstract":"Radiomics is the newest and promising direction in modern radiographic diagnostics. The number of head and neck cancer studies with employing radiomics is increasing every year. We performed a systematic review of recent publications (20212023) on computed tomography (CT)-based head and neck malignancies. The search for articles was carried out in the PubMed database. The basic characteristics of the selected articles were extracted and their quality was assessed with RQS 2.0 [3] and the modified QUADAS-CAD questionnaire [17]. We assessed the level of reproducibility of radiomic parameters selected for predictive models in different studies. Eleven articles were selected for our review. In most cases, there was a high risk of systematic error associated with the data imbalance in terms of demographic parameters and level of pathologies. The range of RQS 2.0 scores for the included articles varies from 19.44% to 50.00% of the maximum possible score. The main problems leading to researches quality decreasing are the lack of external validation of the results (73% of the analyzed articles) and the lack of data accessibility and transparency (82%). Inter-study reproducibility of radiomics parameters is low due to the wide variety of techniques used for image acquisition, image post-processing, extraction and statistical processing of radiomics parameters. The basic block of the stable radiomics parameters should be created for the method introducing into clinical practice. The radiomics methods standardization and creating an open radiomics database creation is necessary for this purpose.","PeriodicalId":34831,"journal":{"name":"Digital Diagnostics","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-01-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140488217","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
E. S. Samsonova, I. Mikhailov, V. V. Omelyanovsky, M. V. Avksentieva, I. Zheleznyakova, G. G. Lebedenko
Background: Unified system for assessing the results and real contribution of of telemedicine consultations (TMC) to improving the quality of medical care in the healthcare system of the Russian Federation has not yet been developed. Aim: Development of a system of indicators for differentiated assessment of the need for TMC in the provision of medical care of various profiles. Materials and methods: At the first stage we analyzed reports on the results of on-site activities of NMRCs in regions of the Russian Federation and annual public reports of NMRCs (2020-2022) to identify indicators that determine the need for TMC. The identified indicators were clarified and validated in an open interview with representatives of the NMRCs. At the second stage we determined the value of each indicator on the bases of expert survey. 18 experts had to score each indicator from 1 to 5. Then we calculated the weight coefficient of each indicator for their subsequent use in planning TMC volumes. Results: Three groups of indicators have been identified. The first group includes indicators of mortality, disability and hospital mortality, the frequency of emergency/urgent consultations and the frequency of consultations of intensive care patients. The second group includes indicators for assessing the effectiveness and efficiency of TMC, subjective (satisfaction with the results of TMC) and objective (the number of positive and negative disease and hospitalization cases outcomes for which TMC was performed). The third group includes indicators characterizing the validity of requests for TMС: patients examination completeness of before TMC, diagnosis accuracy. The weight coefficients of the indicators of the first group ranged from 0.05 to 1.61 and were different for different profiles. Conclusion: A system of indicators has been proposed for differentiated assessment of the need for TMC when providing medical care of different profiles.
{"title":"Identification of indicators used to assess needs for telemedical consultations in various profiles of medical care","authors":"E. S. Samsonova, I. Mikhailov, V. V. Omelyanovsky, M. V. Avksentieva, I. Zheleznyakova, G. G. Lebedenko","doi":"10.17816/dd622846","DOIUrl":"https://doi.org/10.17816/dd622846","url":null,"abstract":"Background: Unified system for assessing the results and real contribution of of telemedicine consultations (TMC) to improving the quality of medical care in the healthcare system of the Russian Federation has not yet been developed. \u0000Aim: Development of a system of indicators for differentiated assessment of the need for TMC in the provision of medical care of various profiles. \u0000Materials and methods: At the first stage we analyzed reports on the results of on-site activities of NMRCs in regions of the Russian Federation and annual public reports of NMRCs (2020-2022) to identify indicators that determine the need for TMC. The identified indicators were clarified and validated in an open interview with representatives of the NMRCs. At the second stage we determined the value of each indicator on the bases of expert survey. 18 experts had to score each indicator from 1 to 5. Then we calculated the weight coefficient of each indicator for their subsequent use in planning TMC volumes. \u0000Results: Three groups of indicators have been identified. The first group includes indicators of mortality, disability and hospital mortality, the frequency of emergency/urgent consultations and the frequency of consultations of intensive care patients. The second group includes indicators for assessing the effectiveness and efficiency of TMC, subjective (satisfaction with the results of TMC) and objective (the number of positive and negative disease and hospitalization cases outcomes for which TMC was performed). The third group includes indicators characterizing the validity of requests for TMС: patients examination completeness of before TMC, diagnosis accuracy. The weight coefficients of the indicators of the first group ranged from 0.05 to 1.61 and were different for different profiles. \u0000Conclusion: A system of indicators has been proposed for differentiated assessment of the need for TMC when providing medical care of different profiles.","PeriodicalId":34831,"journal":{"name":"Digital Diagnostics","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2024-01-29","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140486278","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
V. Zinchenko, K. Arzamasov, A. Vladzymyrskyy, Yuriy A. Vasilev
Background: Technological defects are critical to deciding on the practical applicability and clinical value of software based AI. Aims: Technological defects, their analysis, structuring is an important task when deciding on the possibility of working in medical organizations, the effectiveness and safety of software based AI. Materials and methods: Monitoring of technological parameters for all participating solutions at the testing stage and at the trial operation stage carried out in an experiment on the use of innovative technologies in the field of computer vision for the analysis of medical images and further application in the Moscow healthcare system. Results: The article presents graphical information on the average number of technological defects for the direction of preventive research in 2021. This period is the most significant and characterize by active development, an increase in the technical stability of the software. Conclusions: The analysis allows us to trace the trend towards a decrease in the number of technological defects. This indicates an increase in the quality of the software based AI due to periodic monitoring.
{"title":"Technological defects of software based on artificial intelligence technology.","authors":"V. Zinchenko, K. Arzamasov, A. Vladzymyrskyy, Yuriy A. Vasilev","doi":"10.17816/dd501759","DOIUrl":"https://doi.org/10.17816/dd501759","url":null,"abstract":"Background: Technological defects are critical to deciding on the practical applicability and clinical value of software based AI. \u0000Aims: Technological defects, their analysis, structuring is an important task when deciding on the possibility of working in medical organizations, the effectiveness and safety of software based AI. \u0000Materials and methods: Monitoring of technological parameters for all participating solutions at the testing stage and at the trial operation stage carried out in an experiment on the use of innovative technologies in the field of computer vision for the analysis of medical images and further application in the Moscow healthcare system. \u0000Results: The article presents graphical information on the average number of technological defects for the direction of preventive research in 2021. This period is the most significant and characterize by active development, an increase in the technical stability of the software. \u0000Conclusions: The analysis allows us to trace the trend towards a decrease in the number of technological defects. This indicates an increase in the quality of the software based AI due to periodic monitoring.","PeriodicalId":34831,"journal":{"name":"Digital Diagnostics","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-12-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"138596316","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Yuriy A. Vasilev, O. Omelyanskaya, Anastasia A. Nasibullina, D. Leonov, Julia V. Bulgakova, D. A. Akhmedzyanova, Y. Shumskaya, R. V. Reshetnikov
Phantoms are used to validate diagnostic imaging methods or develop skills of medical professionals. For instance, they allow conducting an unlimited number of imaging studies during medical training, assessing the image quality, optimizing the radiation dose, and testing novel techniques and equipment. Researchers in breast imaging utilize anthropomorphic models to validate, assess, and optimize new methods for diagnosing breast diseases. Such models also facilitate control over the quality of diagnostic systems, help to optimize clinical protocols and improve image reconstruction algorithms. Realistic simulation of organ tissue is essential for addressing these challenges in breast phantoms. The goal of this review is to describe what breast phantoms for diagnostic imaging are currently available on the market and how they are fabricated.Building an accurate breast model with X-ray imaging requires detailed knowledge of its anatomy and radiological features. The breast has a heterogeneous structure composed of glandular and adipose tissues, skin, and appendages, as well as other structures such as vessels and ligaments. In this literature review, we screened PubMed and Google Scholar for the relevant articles. 72 articles and 13 conference papers were included.There are two major types of breast phantoms: computational and physical. Specifically, the computational phantoms are classified into sub-groups depending on what data they use. These include mathematical models, tissue samples, and medical images of the breast. The physical phantoms, on the other hand, are classified based on their composition: molds, 3D printed, or paper-based with contrast inclusions. The main advantage of computational phantoms is the ability to generate large amounts of virtual data, while physical phantoms allow to perform an unlimited number of radiological studies.
{"title":"ANTHROPOMORPHIC BREAST PHANTOMS FOR RADIOLOGY IMAGING: LITERATURE REVIEW","authors":"Yuriy A. Vasilev, O. Omelyanskaya, Anastasia A. Nasibullina, D. Leonov, Julia V. Bulgakova, D. A. Akhmedzyanova, Y. Shumskaya, R. V. Reshetnikov","doi":"10.17816/dd623341","DOIUrl":"https://doi.org/10.17816/dd623341","url":null,"abstract":"Phantoms are used to validate diagnostic imaging methods or develop skills of medical professionals. For instance, they allow conducting an unlimited number of imaging studies during medical training, assessing the image quality, optimizing the radiation dose, and testing novel techniques and equipment. Researchers in breast imaging utilize anthropomorphic models to validate, assess, and optimize new methods for diagnosing breast diseases. Such models also facilitate control over the quality of diagnostic systems, help to optimize clinical protocols and improve image reconstruction algorithms. Realistic simulation of organ tissue is essential for addressing these challenges in breast phantoms. \u0000The goal of this review is to describe what breast phantoms for diagnostic imaging are currently available on the market and how they are fabricated.Building an accurate breast model with X-ray imaging requires detailed knowledge of its anatomy and radiological features. The breast has a heterogeneous structure composed of glandular and adipose tissues, skin, and appendages, as well as other structures such as vessels and ligaments. \u0000In this literature review, we screened PubMed and Google Scholar for the relevant articles. 72 articles and 13 conference papers were included.There are two major types of breast phantoms: computational and physical. Specifically, the computational phantoms are classified into sub-groups depending on what data they use. These include mathematical models, tissue samples, and medical images of the breast. The physical phantoms, on the other hand, are classified based on their composition: molds, 3D printed, or paper-based with contrast inclusions. The main advantage of computational phantoms is the ability to generate large amounts of virtual data, while physical phantoms allow to perform an unlimited number of radiological studies.","PeriodicalId":34831,"journal":{"name":"Digital Diagnostics","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-12-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"138595763","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：