Numerous studies have shown that AI has the potential to enhance the ability of radiologists to detect emergencies, increase clinical confidence, reduce workload, and inform patient diagnostic and treatment strategies. . Therefore, instead of competing with radiologists as once suspected, AI actually improves radiologists in providing optimal patient care. AI has the potential to transform the work of radiologists in image analysis through three main stages: identification, characterization and monitoring.
The term "artificial intelligence" is used to describe machines or programs that mimic intelligence. This means that these machines or programs can understand situations and therefore help us make decisions - or even make their own decisions. Artificial intelligence is built around three components: large data sets and data analysis, algorithms based on scalable computing power and learning ability, and a method called "deep learning" (among others). Yes, in view.
Search
Identification refers to the process of flagging and binding a specific subsection in an image that is likely to have an injury or disorder. Current technologies that help radiologists identify areas of interest are called computer aided detection (CADe). However, current CADE systems are limited by the high rates of false-positives and the high labor requirements, as each flag requires radiologist evaluation.
The potential benefits of artificially intelligent CADE systems can already be seen through basic research in many areas of image screening. Recent studies have highlighted the potential of in-depth practice-based CADE systems to detect pulmonary nodules in CT. Furthermore, the AI CADE system has been shown to have similar performance compared to human mammogram readers. In a 2020 study involving coronary computed tomography angiography imaging, AI applications were able to accurately diagnose coronary artery disease within two minutes, which could help in the future. Radiologists prefer CCTA images with positive results for more detailed reports. The AI CADE system has also been used in neurology to detect intracranial LVO with excellent sensitivity (82%) and specificity (94%); Implementing these systems will help in getting work priority by alerting senior physicians regarding the case. All of this research highlights the use of AI in the future development of high-performance CADE systems.
Characterization
Characterization refers to the identification of specific features of pathological exploration, such as size, scope, and internal morphology. These features can be used to classify lesions into different pathological categories.
For lung nodules, CNNs distinguish between benign and malignant classifications with higher performance than traditional CADX systems due to their ability to withstand high levels of noise. In addition, in a study of patients with non-small cell lung cancer, AI CADX algorithms were able to use CT images to assess which cancers had EGFR mutations compared to gifitinib. Inform about possible treatments. Deep learning algorithms have also been trained to accurately classify prostate cancer on magnetic resonance imaging (MRI), which promotes early treatment and reduces the number of unnecessary prostate biopsy and prostatectomy procedures. Is. An additional study reported an AI system that could use MRI imaging to accurately diagnose brain tumor classification differences beyond human exposure. The algorithm made 91% correct diagnosis of its first three variants, surpassing brain MRI AI by academic neuroradiologists (86%), colleagues (77%), general radiologists (57%) and radiology residents (56%). Did. The algorithm can classify gliomas into molecular subtypes by identifying the imaging properties associated with mutations in IDH1 / IDH2, EGFR, MGMT, and / or chromosomes 1p and 19q (80% sensitivity and 95% specificity) (Bi et al). . Defendants. , This level of characterization is important because it provides an identification of the entire tumor rather than the source of the tumor from the biopsy, which more accurately describes the treatment.
Supervision
Surveillance refers to the longitudinal follow-up of a identified pathology over time to assess changes in natural history or in response to treatment.
AI-based monitoring completes these protocols by capturing a large number of discriminatory features that are not detected by the human eye. The ability to detect these subtle features allows AI-based surveillance systems to provide a clearer picture of tumor development.
AI-based surveillance techniques can accurately detect progression in active tumors and also predict future recurrence areas after tumor dissection. According to the MRI study, the AI algorithm was able to detect new margins of tumor cell infiltration invisible to the human eye on post-contrast images.
You must be logged in to post a comment.