Page 10 - AIH-1-1

Artificial Intelligence in Health AI in prostate cancer detection

tripled in comparison to developing countries. Notably, Artificial intelligence (AI) and machine learning (ML)
prostate cancer ranks as the second leading cause of tools are increasingly employed to improve the diagnostic
mortality in men worldwide. Interestingly, patterns of accuracy of prostate cancer. AI, fundamentally a computer
mortality do not strictly align with those of incidence [1,2] . system and program, facilitates the execution of complex
Factors contributing to the variance between incidence tasks that humans traditionally perform. These tasks
and mortality across different countries include advanced encompass, but are not limited to, pattern recognition,
age, ethnicity, family history, genetic mutations, and translations, speech, and vision. AI helps in task automation,
certain lifestyle factors; nevertheless, these elements only optimizing cost, saving time, improving accuracy, and
offer partial explanations for the observed differences [4-6] . minimizing the risk of human error. Examples of AI tools
Early diagnosis emerges as a potential explanation for the include Google Translate, AI-powered robotic surgeries,
dichotomous pattern between incidence and mortality in and self-driving cars. ML serves as an integral component
developed and developing countries. of overarching AI systems, utilizing data, statistics, and
Serum prostate-specific antigen (PSA) test has been smart algorithms to learn like humans, thereby assisting
used in screening for prostate cancer, primarily in in solving complex problems such as medical diagnosis,
developed countries. This adoption initially resulted in machine failure prediction, and image recognition.
a significant increase in incidence rates, followed by a Various types of ML exist, including supervised learning
subsequent decline. The adoption of this test has gradually for classification, unsupervised for clustering, and
extended to other countries. Between 2000 and 2015, a reinforcement learning for real-time applications. Tran
[18]
discernable rise in the incidence of prostate cancer was et al. characterized the use of AI applications in cancer
observed in developing countries as well [7-9] . Serum PSA research. Several recent studies have explored the utility of
[19]
[20]
[21]
estimation is a simple, relatively inexpensive, and widely AI in conjunction with pathology , MRI , or both .
available test for prostate cancer screening. However, its Review articles discussing AI-based neural networks
widespread availability carries the risk of yielding false- and their current role in the diagnosis of prostate cancer
positive results, prompting unnecessary biopsies, detecting have been published. Furthermore, bibliometric analyses
clinically indolent disease with no impact on the patient’s (BA) on various aspects of prostate cancer are emerging.
history, imposing a burden on healthcare resources, and Guo et al. noted in a BA study that AI in health-care
[22]
inflicting psychological harm on both individuals and holds a promising role and great prospects. For example,
their families . Acknowledging this concern, the United Adam et al. conducted a BA of the top 100 cited articles
[23]
[10]
States Preventive Services Task Force and the American published in the field of prostate cancer. In addition, Tang
Cancer Society have revised their guidelines, suggesting et al. carried out a BA of 100 most cited articles on
[24]
the use of PSA screening in men with average risk only prostate cancer brachytherapy. Ma et al. explored trends
[25]
after comprehensive discussions on uncertainties, risks, related to erectile dysfunction and prostate cancer using
and potential benefits [11,12] . BA. Takeshima et al. undertook a BA study examining
[26]
A high serum PSA level prompts a series of prostate cancer among Japanese males, utilizing autopsy
[27]
investigations, such as digital rectal examination, transrectal reports. Mushtaq and Loan explored the research status
ultrasound, and ultrasound-guided biopsy. The biopsy of prostate cancer using BA, particularly focusing on Iran
findings are subsequently interpreted based on the and India.
Gleason score (GS), categorizing the disease as either In this study, we conducted a comprehensive and
clinically insignificant (GS 6) or significant (GS 7 – 10). systematic literature review using BA tools to examine
However, GS is inherently subjective and liable to inter- the use of AI or ML (AI/ML) techniques in the diagnosis
individual variability. In recent developments, magnetic and treatment of prostate cancer. In addition, we propose
resonance imaging (MRI) has been incorporated into the directions for future research avenues to enhance the
diagnostic process before biopsy to enhance the detection application of AI/ML in the realm of prostate cancer cure.
of clinically significant prostate cancer, distinguishing it
from clinically insignificant cases [13-15] . Multiparametric 2. Search criteria and research framework
MRI (mpMRI) is now a recommended component in the 2.1. Search Criteria
guidelines, to be performed before biopsy . However,
[16]
the limited accessibility of mpMRI, increased workload, Established research repositories such as SCOPUS, Web
and the inherent inter-individual variability in reporting of Science (WoS), and Google Scholar were employed to
all underscore the need for more effective methods of identify published articles related to the use of AI/ML in the
diagnosing clinically significant prostate cancer at an early detection and treatment of prostate cancer. The keywords
stage . employed in the searches are as follows: (“prostate” AND
[17]

Volume 1 Issue 1 (2024) 4 https://doi.org/10.36922/aih.1958