Deyer et al. [21] | 2019 | Editorial | Discussing application of AI to radiology | ANN process allows the continuous progression of the system, learning in a similar way that corresponds with the human brain, but with the benefit of a permanent memory function |
Schier [22] | 2018 | Opinion | Providing an alternative view on radiology practice and AI | The accuracy rates of ANNs currently surpass human radiologists in narrow-based tasks, such as detecting lung nodule features |
Stivaros et al. [23] | 2010 | Review article | Reviews role of decision support systems in radiological assessment | Machine learning permits reliable automated detection of lung nodules in CT scans, and pneumonia in chest x-rays |
Porto Pazos [24] | 2008 | Book | Use of biological process application in advancement of AI | The behaviour of pre-cancerous lesions on CT scans is predicted by means of modeling, or regression, and prevents superfluous invasive examinations, such as biopsy |
Pesapane et al. [25] | 2018 | Narrative review | Discusses the role of practicing radiologist in promoting AI in radiology | AI will help radiologists in being a multidisciplinary team, be more on forefront with patients and add value to tasks |
Kamar [26] | 2016 | Review paper | Considering Human intelligence in AI | Electronic health records could improve the success rate of radiologists |
Fieschi [27] | 2013 | Book | Discusses expert systems of AI in medicne | Radiologists can distinguish themselves by creating AI that does hybrid work through collective intelligence software |
Danforth et al. [28] | 2009 | Project writing | Developed a virtual patient simulations system that will help in medical education | Intelligence augmentation in radiology is where higher levels of accuracy in diagnosis are achieved through the amalgamation of human radiologists and AI to form hybrid intelligence |
Thrall et al. [29] | 2018 | Descriptive study | Discusses opportunities and challenges faced by AI in radiology | Value created by AI will measure its role in radiology i.e., by increasing diagnostic certainty, quicker turnaround, improved patient outcomes, and for radiologists a better quality of work life |
Gillies et al. [30] | 2016 | Special report | Describes the process of radiomics, its potential power to facilitate better clinical decision making and its challenges; especially in cancer patients | It is important to consider aspects such as picture archiving and communication system (PACS), electronic health records, IT environments, and radiology information systems in the implementation phase of AI |
Prevedello et al. [31] | 2017 | Retrospective study | AI algorithm performance was tested on a separate dataset containing 35 with noncritical findings, 15 with suspected acute infarct findings, and 50 with hemorrhage, mass effect, or hydrocephalus findings | AI (deep learning) is promising for detecting noncontrast-enhanced head CT, whereas, to detect suspected acute infarction required dedicated algorithm. Detection of suspected acute infarct had lower sensitivity compared to hemorrhage, mass effect, or hydrocephalus detection, but showed reasonable performance |
Hanson [32] | 2001 | Review article | Databases were searched for articles regarding application of AI in radiographs used in intensive care units | Unsupervised learning is one approach that allows automated data curation |
Kulikowski [33] | 1988 | Review article | A brief history of articles before 1988 on the use of AI in medicine is reviewed in a systematic way | Recent developments in unsupervised learning include different auto-encoders and generative adversarial networks, which are highly effective, where discriminated aspects are learned despite a lack of comprehensive labelling |