TY - JOUR

T1 - Challenges of Artificial Intelligence in Cancer Diagnosis: An Update on Future and Prospects

AU - Rashidi, Bahare

AU - Zohori, Kiarash

AU - Aminzadeh, Seyyedeh Samin

AU - Abachi, Seyedeh Fatemeh

AU - Behzad, Farnaz

AU - Omranzadeh, Alireza

PY - 2025/11

Y1 - 2025/11

N2 - Background: Artificial intelligence (AI) has become increasingly prominent in the medical field, particularly in the diagnosis of cancer. Objectives: This comprehensive review was conducted to review the challenges of AI in cancer diagnosis. Methods: This comprehensive review was conducted through a systematic search of major scientific databases, including PubMed, Scopus, Web of Science, and IEEE Xplore, utilizing a combination of keywords and Medical Subject Headings (MeSH) terms such as “artificial intelligence,” “machine learning,” “deep learning,” “neural networks,” “cancer diagnosis,” “oncological imaging,” “pathology,” “biomarkers,” and “precision oncology,” covering the period from January 2019 to December 2024 to capture the most relevant and impactful studies in this rapidly evolving field. The inclusion criteria were focused on peer-reviewed original research articles, significant review papers, and high-impact conference proceedings that demonstrated a direct application of AI algorithms in diagnostic procedures, while exclusion criteria encompassed non-English publications, studies with insufficient methodological detail, articles not focused on diagnostic applications, and editorials or opinion pieces without original data, ensuring a robust and evidence-based analysis of the current landscape. Results: The challenges in the widespread utilization of this technology in clinical settings are discussed. Deep learning algorithms, especially convolutional neural networks (CNN), can identify suspicious areas in mammograms, CT scans, and MRI images that doctors may easily overlook. These capabilities improve accuracy and reduce human errors in cancer diagnosis. In addition to image analysis, AI can also analyze patients' molecular and genetic data. Using genomic and proteomic data, this technology can identify gene mutations and specific biological markers of cancer. As a result, early diagnosis and selection of targeted patient treatments are carried out with greater accuracy. However, despite significant progress in this field, several challenges remain, including the accurate interpretation of data, the need for substantial training data, and the ability to generalize algorithms to diverse populations. Conclusion: In conclusion, AI is fundamentally augmenting the field of cancer diagnostics, moving from a theoretical promise to a powerful clinical tool. The evidence demonstrates that AI algorithms, particularly deep learning models, offer significant and measurable benefits.

AB - Background: Artificial intelligence (AI) has become increasingly prominent in the medical field, particularly in the diagnosis of cancer. Objectives: This comprehensive review was conducted to review the challenges of AI in cancer diagnosis. Methods: This comprehensive review was conducted through a systematic search of major scientific databases, including PubMed, Scopus, Web of Science, and IEEE Xplore, utilizing a combination of keywords and Medical Subject Headings (MeSH) terms such as “artificial intelligence,” “machine learning,” “deep learning,” “neural networks,” “cancer diagnosis,” “oncological imaging,” “pathology,” “biomarkers,” and “precision oncology,” covering the period from January 2019 to December 2024 to capture the most relevant and impactful studies in this rapidly evolving field. The inclusion criteria were focused on peer-reviewed original research articles, significant review papers, and high-impact conference proceedings that demonstrated a direct application of AI algorithms in diagnostic procedures, while exclusion criteria encompassed non-English publications, studies with insufficient methodological detail, articles not focused on diagnostic applications, and editorials or opinion pieces without original data, ensuring a robust and evidence-based analysis of the current landscape. Results: The challenges in the widespread utilization of this technology in clinical settings are discussed. Deep learning algorithms, especially convolutional neural networks (CNN), can identify suspicious areas in mammograms, CT scans, and MRI images that doctors may easily overlook. These capabilities improve accuracy and reduce human errors in cancer diagnosis. In addition to image analysis, AI can also analyze patients' molecular and genetic data. Using genomic and proteomic data, this technology can identify gene mutations and specific biological markers of cancer. As a result, early diagnosis and selection of targeted patient treatments are carried out with greater accuracy. However, despite significant progress in this field, several challenges remain, including the accurate interpretation of data, the need for substantial training data, and the ability to generalize algorithms to diverse populations. Conclusion: In conclusion, AI is fundamentally augmenting the field of cancer diagnostics, moving from a theoretical promise to a powerful clinical tool. The evidence demonstrates that AI algorithms, particularly deep learning models, offer significant and measurable benefits.

KW - Artificial intelligence

KW - Cancer diagnosis

KW - Convolutional neural network

KW - Machine learning

KW - Medical image analysis

UR - https://www.scopus.com/pages/publications/105027855336

UR - https://www.mendeley.com/catalogue/200efe50-30f1-379a-9256-dafd569bbca5/

U2 - 10.30483/rjm.2025.254628.1408

DO - 10.30483/rjm.2025.254628.1408

M3 - Article

VL - 13

SP - 1

EP - 16

JO - Razavi International Journal of Medicine

JF - Razavi International Journal of Medicine

SN - 2345-6426

IS - 4

ER -