Page 16 - AIH-1-1

Artificial Intelligence in Health AI in prostate cancer detection

Figure 7. Historiographical direct citation network.

Feature Extraction Table 5. Historiographical analysis and themes
9% MRI
Metastasis 12% Path theme (s) References related to historiographical
10% analysis
AI in MRI (Red) Wang et al. ; Alkadi et al. ; Bernatz
[65]
[42]
[67]
[66]
Biopsy et al. ; Corradini et al. ; Abdelmaksoud
[68]
3% et al. ; Donisi et al. [69]
Cancer Grading AI in multi-view Leo et al. ; Kwak and Hewitt ; Kwak and
[70]
[71]
24%
[72]
[19]
Gleason score pathology (Yellow) Hewitt ; Pantanowitz et al. ; Han et al. [73]
15%
Radiomics Li et al. ; Liang et al. [75]
[74]
AI in pathology and Li et al. ; Harmon et al. [77]
[76]
radiology (Green)
[78]
Prostate Radiomics advances Hou et al. ; Lim et al. [79]
Segmentation Radiomics (Purple)
7% 20%
Abbreviations: AI: Artificial intelligence; MRI: Magnetic resonance
Figure 8. Prostate cancer diagnostics and treatment areas. imaging.
In the first category, studies employed a number of Table 6. Artificial intelligence or machine learning
contemporary AI/ML tools for predictive modeling techniques: Contemporary research and application
related to prostate cancer diagnosis and potential curative Contemporary No of Application areas No. of
methods, as outlined in Table 6. The table details the types Research publications publications
of AI/ML techniques applied, with deep learning, ANNs, Deep learning 17 MRI 7
and support vector machines (SVMs) emerging as the most Artificial neural 15 Cancer grading 14
commonly used techniques for prostate cancer prediction networks
and classification. The second category encompasses the Naïve Bayes 2 Radiomics 12
application areas of these AI/ML tools (as presented in Support vector 11 Prostate segmentation 4
Table 6), with cancer grading and radiomics emerging as machine
the two most widely studied areas (as depicted in Figure 8). Random forest 7 Gleason score 9
Eight different AI techniques have been employed Regression 5 Biopsy 2
in modern prostate cancer research. As delineated in K-nearest 4 Metastasis 6
Table 6, approximately 75% of research papers utilized neighbors
deep learning, ANNs, and SVMs. Deep learning emerged Fuzzy clustering 1 Features extraction 5
as the most frequently used technique (17 publications). Abbreviation: MRI: Magnetic resonance imaging.
The success of deep learning-based research can be
attributed to its logical structure, closely resembling human intervention. Another commonly employed AI
human analysis, although requiring more data and less technique in the prediction and classification of cancers is

Volume 1 Issue 1 (2024) 10 https://doi.org/10.36922/aih.1958