Pub Date : 2023-10-01DOI: 10.23736/s2723-9284.23.00247-9
Michele AVANZO, Giovanni PIRRONE, Joseph STANCANELLO
Radiomics is a growing field where hundreds or thousands of quantitative features are extracted from a contoured region in a medical image in order to describe the image properties of a lesion or tissue. The radiomic features are then used for building an artificial intelligence-based model that can perform a diagnosis or characterization of tissues and organs. In this article we have defined the field of radiomics, its workflow and tools and describe some of the results achieved in studies applying radiomics. We also want to discuss its main limitations and strengths, in particular when compared with other artificial intelligence technique applied to imaging.
{"title":"Radiomics for the radiologist: opportunities and challenges","authors":"Michele AVANZO, Giovanni PIRRONE, Joseph STANCANELLO","doi":"10.23736/s2723-9284.23.00247-9","DOIUrl":"https://doi.org/10.23736/s2723-9284.23.00247-9","url":null,"abstract":"Radiomics is a growing field where hundreds or thousands of quantitative features are extracted from a contoured region in a medical image in order to describe the image properties of a lesion or tissue. The radiomic features are then used for building an artificial intelligence-based model that can perform a diagnosis or characterization of tissues and organs. In this article we have defined the field of radiomics, its workflow and tools and describe some of the results achieved in studies applying radiomics. We also want to discuss its main limitations and strengths, in particular when compared with other artificial intelligence technique applied to imaging.","PeriodicalId":369070,"journal":{"name":"Journal of Radiological Review","volume":"38 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"135759973","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}
Pub Date : 2023-10-01DOI: 10.23736/s2723-9284.23.00256-0
Elisa BARATELLA, Pierluca MINELLI, Antonio SEGALOTTI, Maria A. COVA
Artificial intelligence (AI) has its earliest roots in ancient history and during the modern age the assumption that a human process could be mechanized was furtherly developed by Western philosophers. The term was coined for the first time in 1956, and in 1976 CASNET - a causal-associational network - was introduced in clinical practice as one of the very first prototypes of AI applied to medicine. The technological progress in the last three decades brought new interest and a significant development in the Artificial Intelligence field, which currently includes computational algorithms that can perform tasks once considered exclusive to human intelligence. Nowadays, there are several methods of Artificial Intelligence, above all machine learning - in which a training stage is needed by the algorithm to recognize specific features - and deep learning - in which algorithms form artificial neural networks in order to simulate the performances of neural networks of the human brain. There is currently an increasing application of AI in radiology and chest imaging is crucially involved in this topic: the aim of this narrative review is thus to describe all the possible applications of different methods of AI in thoracic radiology, regarding diagnostic imaging as well as interventional procedures.
{"title":"Application of artificial intelligence in chest radiology","authors":"Elisa BARATELLA, Pierluca MINELLI, Antonio SEGALOTTI, Maria A. COVA","doi":"10.23736/s2723-9284.23.00256-0","DOIUrl":"https://doi.org/10.23736/s2723-9284.23.00256-0","url":null,"abstract":"Artificial intelligence (AI) has its earliest roots in ancient history and during the modern age the assumption that a human process could be mechanized was furtherly developed by Western philosophers. The term was coined for the first time in 1956, and in 1976 CASNET - a causal-associational network - was introduced in clinical practice as one of the very first prototypes of AI applied to medicine. The technological progress in the last three decades brought new interest and a significant development in the Artificial Intelligence field, which currently includes computational algorithms that can perform tasks once considered exclusive to human intelligence. Nowadays, there are several methods of Artificial Intelligence, above all machine learning - in which a training stage is needed by the algorithm to recognize specific features - and deep learning - in which algorithms form artificial neural networks in order to simulate the performances of neural networks of the human brain. There is currently an increasing application of AI in radiology and chest imaging is crucially involved in this topic: the aim of this narrative review is thus to describe all the possible applications of different methods of AI in thoracic radiology, regarding diagnostic imaging as well as interventional procedures.","PeriodicalId":369070,"journal":{"name":"Journal of Radiological Review","volume":"26 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"135761593","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}
Pub Date : 2023-10-01DOI: 10.23736/s2723-9284.23.00249-9
Stefania MONTEMEZZI, Carlo CAVEDON
Artificial intelligence (AI) is a fast-moving technology that enables machines to perform tasks that could previously be done only by humans. The current debate is now whether machines will outperform humans, and therefore substitute them in critical tasks. In this paper, an attempt will be made to identify the most used AI techniques in diagnostic imaging, providing examples and identifying potential pitfalls.
{"title":"Tools for quantitative radiology: natural and artificial intelligence together","authors":"Stefania MONTEMEZZI, Carlo CAVEDON","doi":"10.23736/s2723-9284.23.00249-9","DOIUrl":"https://doi.org/10.23736/s2723-9284.23.00249-9","url":null,"abstract":"Artificial intelligence (AI) is a fast-moving technology that enables machines to perform tasks that could previously be done only by humans. The current debate is now whether machines will outperform humans, and therefore substitute them in critical tasks. In this paper, an attempt will be made to identify the most used AI techniques in diagnostic imaging, providing examples and identifying potential pitfalls.","PeriodicalId":369070,"journal":{"name":"Journal of Radiological Review","volume":"160 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"135759974","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}
Pub Date : 2023-10-01DOI: 10.23736/s2723-9284.23.00244-9
Umberto ROZZANIGO, Giulia CASAGRANDA, Marianna MOCHEN, Mauro FERRARI
Recent International and National Guidelines for management of Acute Ischemic Stroke recommend the use of automated perfusion software that calculates core/penumbra maps to support clinical decision-making. Artificial intelligence (AI) algorithms requires human expertise for the interpretation and real-life implementation, challenging the Radiologist to be responsible for the performance of this new diagnostic AI-based tool. We illustrate our experience introducing an automated computed tomography brain-perfusion software in the critical setting of the Emergency Radiology of a hub hospital, showing advantages and limitations of the currently available AI technology.
{"title":"Artificial intelligence in neuroradiology: brain CT perfusion imaging for acute ischemic stroke management","authors":"Umberto ROZZANIGO, Giulia CASAGRANDA, Marianna MOCHEN, Mauro FERRARI","doi":"10.23736/s2723-9284.23.00244-9","DOIUrl":"https://doi.org/10.23736/s2723-9284.23.00244-9","url":null,"abstract":"Recent International and National Guidelines for management of Acute Ischemic Stroke recommend the use of automated perfusion software that calculates core/penumbra maps to support clinical decision-making. Artificial intelligence (AI) algorithms requires human expertise for the interpretation and real-life implementation, challenging the Radiologist to be responsible for the performance of this new diagnostic AI-based tool. We illustrate our experience introducing an automated computed tomography brain-perfusion software in the critical setting of the Emergency Radiology of a hub hospital, showing advantages and limitations of the currently available AI technology.","PeriodicalId":369070,"journal":{"name":"Journal of Radiological Review","volume":"44 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"135759978","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}
Pub Date : 2023-10-01DOI: 10.23736/s2723-9284.23.00245-9
Riccardo DE ROBERTIS, Flavio SPOTO, Francesca PASQUAZZO, Mirko D’ONOFRIO
The evaluation of treatment response and the noninvasive prognostic stratification of cancer patients are the most interesting and ambitious applications of radiomics and artificial intelligence, with potentially relevant clinical implications. Several studies reported promising results at this regard, even though their scientific quality is low and large-scale validation of the results is necessary. The purpose of this paper was to review systematic reviews and meta-analyses regarding the use of radiomics and artificial intelligence for prognostic stratification and evaluation of treatment response in cancer patients.
{"title":"Clinical applications of radiomics and artificial intelligence: prognostic stratification and response to treatment","authors":"Riccardo DE ROBERTIS, Flavio SPOTO, Francesca PASQUAZZO, Mirko D’ONOFRIO","doi":"10.23736/s2723-9284.23.00245-9","DOIUrl":"https://doi.org/10.23736/s2723-9284.23.00245-9","url":null,"abstract":"The evaluation of treatment response and the noninvasive prognostic stratification of cancer patients are the most interesting and ambitious applications of radiomics and artificial intelligence, with potentially relevant clinical implications. Several studies reported promising results at this regard, even though their scientific quality is low and large-scale validation of the results is necessary. The purpose of this paper was to review systematic reviews and meta-analyses regarding the use of radiomics and artificial intelligence for prognostic stratification and evaluation of treatment response in cancer patients.","PeriodicalId":369070,"journal":{"name":"Journal of Radiological Review","volume":"84 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"135759982","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}
Pub Date : 2023-10-01DOI: 10.23736/s2723-9284.23.00254-0
Vittoria ROSSI, Riccardo DE ROBERTIS, Luisa TOMAIUOLO, Luca GERACI, Mirko D’ONOFRIO
Two branches on which precision medicine is based are radiomics and genomics, in particular the latter analyzes the different molecules. The study of the molecules is the basis of the response to treatment and therefore of the choice of the different therapeutic strategies. Currently, radiomic data are typically not incorporated as part of this data stream; however, this is changing with the adoption of structured radiology reporting. The challenge going forward will be to capture radiomic data as part of the structured report. Based on multiple studies about liver and pancreas neoplasms it is clearly visible what radiomics has brought in terms of preoperative prognostic factors related to survival and prognostic stratification, based on degree of aggressiveness of the lesion, as well as the evaluation of factors associated with presence of metastases or presence of vascular microinvasion. Several studies broadly describe genomic approaches to solve different problems in the context of liver and pancreatic imaging. In particular segmentation, quantification, characterization and improvement of image quality. Artificial intelligence will not be able to replace man, who covers a fundamental role; for example, the radiologist’s experience in manual tumor segmentation. Surely the prospect is to bring help in terms of time consumption.
{"title":"Radiomics, radiogenomics and artificial intelligence in the study of liver and pancreatic tumors","authors":"Vittoria ROSSI, Riccardo DE ROBERTIS, Luisa TOMAIUOLO, Luca GERACI, Mirko D’ONOFRIO","doi":"10.23736/s2723-9284.23.00254-0","DOIUrl":"https://doi.org/10.23736/s2723-9284.23.00254-0","url":null,"abstract":"Two branches on which precision medicine is based are radiomics and genomics, in particular the latter analyzes the different molecules. The study of the molecules is the basis of the response to treatment and therefore of the choice of the different therapeutic strategies. Currently, radiomic data are typically not incorporated as part of this data stream; however, this is changing with the adoption of structured radiology reporting. The challenge going forward will be to capture radiomic data as part of the structured report. Based on multiple studies about liver and pancreas neoplasms it is clearly visible what radiomics has brought in terms of preoperative prognostic factors related to survival and prognostic stratification, based on degree of aggressiveness of the lesion, as well as the evaluation of factors associated with presence of metastases or presence of vascular microinvasion. Several studies broadly describe genomic approaches to solve different problems in the context of liver and pancreatic imaging. In particular segmentation, quantification, characterization and improvement of image quality. Artificial intelligence will not be able to replace man, who covers a fundamental role; for example, the radiologist’s experience in manual tumor segmentation. Surely the prospect is to bring help in terms of time consumption.","PeriodicalId":369070,"journal":{"name":"Journal of Radiological Review","volume":"24 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"135761592","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}
Pub Date : 2023-10-01DOI: 10.23736/s2723-9284.23.00252-0
Giovanni MORANA
With the advent of artificial intelligence (AI) in the field of radiology, a new perspective opens up in terms of diagnosis and management of patients. There is a need to review the way radiologists work so as to rebuild the doctor-patient relationship that has been sidelined over the years to increase our productivity. It is precisely the improvement in productivity that will be made possible by AI that will be able to free the radiology physician from time-consuming activities that add little to the diagnostic value of our work; this “gift of time” will have to be used to have a direct relationship with the patient, who can be followed up directly by the radiology physician, and not just sent by other physicians. This will be all the more necessary since with the new methods of image analysis (deep learning, texture analysis) the radiologist physician will not only have the task of diagnosing a lesion as accurately as possible, but also of indicating its evolution and progression, what makes indispensable a new pact with the patient, who will have to not only “accept” the diagnosis of an existing lesion but, above all, will have to trust the prognosis of that lesion, a trust based on an immaterial datum (the advanced image analysis) but which weighs like a boulder on the psyche of the patient. Only a relationship of great trust with his new physician, the radiologist, can make him follow our directions.
{"title":"Machine learning and big data in precision medicine: what is the role of the Radiologist?","authors":"Giovanni MORANA","doi":"10.23736/s2723-9284.23.00252-0","DOIUrl":"https://doi.org/10.23736/s2723-9284.23.00252-0","url":null,"abstract":"With the advent of artificial intelligence (AI) in the field of radiology, a new perspective opens up in terms of diagnosis and management of patients. There is a need to review the way radiologists work so as to rebuild the doctor-patient relationship that has been sidelined over the years to increase our productivity. It is precisely the improvement in productivity that will be made possible by AI that will be able to free the radiology physician from time-consuming activities that add little to the diagnostic value of our work; this “gift of time” will have to be used to have a direct relationship with the patient, who can be followed up directly by the radiology physician, and not just sent by other physicians. This will be all the more necessary since with the new methods of image analysis (deep learning, texture analysis) the radiologist physician will not only have the task of diagnosing a lesion as accurately as possible, but also of indicating its evolution and progression, what makes indispensable a new pact with the patient, who will have to not only “accept” the diagnosis of an existing lesion but, above all, will have to trust the prognosis of that lesion, a trust based on an immaterial datum (the advanced image analysis) but which weighs like a boulder on the psyche of the patient. Only a relationship of great trust with his new physician, the radiologist, can make him follow our directions.","PeriodicalId":369070,"journal":{"name":"Journal of Radiological Review","volume":"180 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"135759962","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}
Pub Date : 2023-10-01DOI: 10.23736/s2723-9284.23.00255-0
Lorenzo CERESER, Leonardo MONTERUBBIANO, Valeria PERUZZI, Chiara ZUIANI, Rossano GIROMETTI
This review provides an overview of how artificial intelligence (AI) can assist radiologists in evaluating prostate magnetic resonance imaging (MRI). Main tasks include image quality assessment, gland outlining, lesion detection and classification, lesion delineation, and structured reporting. Although the implementation of AI-based systems is still in its early stages, they have demonstrated promising results in improving the accuracy and efficiency of prostate MRI and reducing variability in diagnostic performance. Specifically, AI-based tools have proven effective in image quality evaluation, gland segmentation, and lesion detection and classification. However, improvements are still necessary, particularly for lesion delineation and automatic structured reporting. Indeed, AI-assisted lesion delineation requires larger, uniformly labeled datasets, and automatic structured reporting requires higher-quality linguistic expression generation. Taken as a whole, while AI-based models hold significant potential to support radiologists in various prostate MRI-related tasks, validation through human-driven clinical trials is required before implementing them in clinical practice. High-quality research is warranted to demonstrate the added value of AI compared to radiologists alone to bridge the gap between the current role of supporting tool and the futuristic role of decision-making tool.
{"title":"Clinical applications of artificial intelligence in Radiology: prostate magnetic resonance imaging","authors":"Lorenzo CERESER, Leonardo MONTERUBBIANO, Valeria PERUZZI, Chiara ZUIANI, Rossano GIROMETTI","doi":"10.23736/s2723-9284.23.00255-0","DOIUrl":"https://doi.org/10.23736/s2723-9284.23.00255-0","url":null,"abstract":"This review provides an overview of how artificial intelligence (AI) can assist radiologists in evaluating prostate magnetic resonance imaging (MRI). Main tasks include image quality assessment, gland outlining, lesion detection and classification, lesion delineation, and structured reporting. Although the implementation of AI-based systems is still in its early stages, they have demonstrated promising results in improving the accuracy and efficiency of prostate MRI and reducing variability in diagnostic performance. Specifically, AI-based tools have proven effective in image quality evaluation, gland segmentation, and lesion detection and classification. However, improvements are still necessary, particularly for lesion delineation and automatic structured reporting. Indeed, AI-assisted lesion delineation requires larger, uniformly labeled datasets, and automatic structured reporting requires higher-quality linguistic expression generation. Taken as a whole, while AI-based models hold significant potential to support radiologists in various prostate MRI-related tasks, validation through human-driven clinical trials is required before implementing them in clinical practice. High-quality research is warranted to demonstrate the added value of AI compared to radiologists alone to bridge the gap between the current role of supporting tool and the futuristic role of decision-making tool.","PeriodicalId":369070,"journal":{"name":"Journal of Radiological Review","volume":"15 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"135759975","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}
Pub Date : 2023-10-01DOI: 10.23736/s2723-9284.23.00260-2
Mirco CLEVA, Lorenzo ZULIANI, Antonio PINTO, Ennio BRUSCHI, Mariachiara CIRILLO, Massimo VALENTINO
Generative artificial intelligence (AI) refers to algorithms that can be used to produce new content, such as text, regulations, images, videos and audio. The background of generative AI in the field of scientific research and publication has significantly shifted with the appearance of generative large language models, such as ChatGPT. ChatGPT represents an AI language model that can produce text close to human writing, making it appropriate for tasks such as summarizing literature and producing statistical studies. The release of ChatGPT has made a substantial influence in the academic world and it has become comprehensible that such technology will notably impact the method of working of researchers. Since its beginning, ChatGPT has been mentioned in several preprints and published articles with authorship credits. This has spawned a large discussion about the function of AI tools in published literature and whether they need to be recognized as authors among journal editors, academics and publishers. Specific guidelines will support the correct use of generative AI, which may be able of future activities such as experiment design and peer review, and facilitate the distribution of important scientific information through publications avoiding any types of scientific misconduct.
{"title":"Lights and shadows of new models in the health-care research: artificial intelligence's role in crafting healthcare research papers","authors":"Mirco CLEVA, Lorenzo ZULIANI, Antonio PINTO, Ennio BRUSCHI, Mariachiara CIRILLO, Massimo VALENTINO","doi":"10.23736/s2723-9284.23.00260-2","DOIUrl":"https://doi.org/10.23736/s2723-9284.23.00260-2","url":null,"abstract":"Generative artificial intelligence (AI) refers to algorithms that can be used to produce new content, such as text, regulations, images, videos and audio. The background of generative AI in the field of scientific research and publication has significantly shifted with the appearance of generative large language models, such as ChatGPT. ChatGPT represents an AI language model that can produce text close to human writing, making it appropriate for tasks such as summarizing literature and producing statistical studies. The release of ChatGPT has made a substantial influence in the academic world and it has become comprehensible that such technology will notably impact the method of working of researchers. Since its beginning, ChatGPT has been mentioned in several preprints and published articles with authorship credits. This has spawned a large discussion about the function of AI tools in published literature and whether they need to be recognized as authors among journal editors, academics and publishers. Specific guidelines will support the correct use of generative AI, which may be able of future activities such as experiment design and peer review, and facilitate the distribution of important scientific information through publications avoiding any types of scientific misconduct.","PeriodicalId":369070,"journal":{"name":"Journal of Radiological Review","volume":"44 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"135759983","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}
Pub Date : 2023-10-01DOI: 10.23736/s2723-9284.23.00246-9
Calogero ZARCARO, Paola CLAUSER
Breast cancer is the most diagnosed cancer in women worldwide, causing significant morbidity and mortality. Imaging techniques play a pivotal role in the early detection of breast cancer; digital mammography (DM) and digital breast tomosynthesis are commonly used for screening average-risk women, while magnetic resonance imaging is employed for high-risk women. Although several progresses have been made in early diagnosis, the number of breast cancer-related deaths remains high, especially among younger women and those diagnosed at advanced stages. To address this problem, new tools are needed that can enable personalized screening or new early diagnosis strategies. Artificial intelligence (AI)-base techniques can assist radiographers and radiologists in various aspects of breast cancer management, including image quality optimization, breast density evaluation, risk assessment and lesion characterization. The level of maturity of the AI technologies currently available in breast imaging is variable. Computer-aided detection (CADe) and computer-aided diagnosis (CADx) were the first AI models introduced to aid radiologists in interpreting DM; CADe marked suspicious areas, while CADx assisted in characterizing findings. However, large-scale studies revealed limited utility and potential negative impacts on mammography interpretation. Conventional CAD systems suffered from low specificity and frequent false positives, failing to address human image perception limitations. The new generation of AI algorithms aims to overcome these limitations and assist radiologists in identifying hidden lesions. This review provides an overview of the current contributions of AI in breast cancer diagnosis, focusing on achieved results, potential objectives, and limitations in clinical practice application.
{"title":"Artificial intelligence clinical applications in breast diagnostic imaging","authors":"Calogero ZARCARO, Paola CLAUSER","doi":"10.23736/s2723-9284.23.00246-9","DOIUrl":"https://doi.org/10.23736/s2723-9284.23.00246-9","url":null,"abstract":"Breast cancer is the most diagnosed cancer in women worldwide, causing significant morbidity and mortality. Imaging techniques play a pivotal role in the early detection of breast cancer; digital mammography (DM) and digital breast tomosynthesis are commonly used for screening average-risk women, while magnetic resonance imaging is employed for high-risk women. Although several progresses have been made in early diagnosis, the number of breast cancer-related deaths remains high, especially among younger women and those diagnosed at advanced stages. To address this problem, new tools are needed that can enable personalized screening or new early diagnosis strategies. Artificial intelligence (AI)-base techniques can assist radiographers and radiologists in various aspects of breast cancer management, including image quality optimization, breast density evaluation, risk assessment and lesion characterization. The level of maturity of the AI technologies currently available in breast imaging is variable. Computer-aided detection (CADe) and computer-aided diagnosis (CADx) were the first AI models introduced to aid radiologists in interpreting DM; CADe marked suspicious areas, while CADx assisted in characterizing findings. However, large-scale studies revealed limited utility and potential negative impacts on mammography interpretation. Conventional CAD systems suffered from low specificity and frequent false positives, failing to address human image perception limitations. The new generation of AI algorithms aims to overcome these limitations and assist radiologists in identifying hidden lesions. This review provides an overview of the current contributions of AI in breast cancer diagnosis, focusing on achieved results, potential objectives, and limitations in clinical practice application.","PeriodicalId":369070,"journal":{"name":"Journal of Radiological Review","volume":"41 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"135761591","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：