The Role of Artificial Intelligence for Image Analysis in Surgical Pathology
| Autor: | Mariana Deacu, Lucian Cristian Petcu, Mariana Aşchie, Nicolae Dobrin, Anca Chisoi, Ionut Burlacu, Liliana Steriu, Ionut Eduard Iordache, Gabriela Izabela Baltatescu, Anca Antonela Nicolau |
|---|---|
| Rok vydání: | 2020 |
| Předmět: |
medicine.medical_specialty
whole slide images lcsh:HB71-74 Computer science Materials Science (miscellaneous) lcsh:Economic theory. Demography ComputingMethodologies_IMAGEPROCESSINGANDCOMPUTERVISION lcsh:Economics as a science artificial intelligence lcsh:QA75.5-76.95 Industrial and Manufacturing Engineering lcsh:HB1-3840 Surgical pathology ComputingMethodologies_PATTERNRECOGNITION machine learning medicine Medical physics lcsh:Electronic computers. Computer science Business and International Management surgical pathology |
| Zdroj: | Annals of Dunarea de Jos University. Fascicle I : Economics and Applied Informatics, Vol 26, Iss 2, Pp 41-48 (2020) |
| ISSN: | 2344-441X 1584-0409 |
| Popis: | Nowadays, artificial intelligence (AI) is an important part of our life and it is a field which continues to grow, to develop and to be implemented in many aspects of our daily tasks. In addition, it continues to conquer many domains of health care with huge progress maid especially in medical image analysis, like radiography, computer tomography, magnetic resonance imaging, digital breast tomosynthesis, positron emission tomography scans or retinal images. The purpose of our work is to analyze the current state of AI systems and software in the practice of pathology. The digitalization process of pathology was possible due to development and improvement of several whole slide images systems in the last decade which provide a vast and rich image data to pathologists and researchers. The algorithmic base of AI is represented by machine learning (ML) and the basis of AI software reside in different statistical model and methods on large set of data. Initially, different mathematical models were used as toolset for ML like supervised and unsupervised learning, random forest, clustering algorithms or component analysis. In the present, deep learning subfield of ML with its neural networks (NN), artificial and convolutional NN, are frequently used in machine vision field. Other tools used in image analysis of digital slide are data augmentation, probability heat maps, patching and computer-aided diagnosis. All of them lead to a high valuable output from which an algorithm can be extracted and can assist the pathologist to render a final diagnosis. Even if there are a lot of challenges in integrated AI solutions in pathology department, they can streamline the diagnosis process in pathology with improvement of workflow, boosting the performance with a better and quicker diagnosis. Image analysis in pathology continues to expend and new branches emerge (digital pathology and pathology informatics) with huge potential for innovative discoveries. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|