A systematic review on deep learning-based automated cancer diagnosis models

Area of Science:

• Artificial Intelligence
• Medical Informatics
• Oncology

Background:

• Deep learning (DL) is increasingly vital across various applications.
• DL models are extensively used by researchers for automated cancer patient diagnosis.
• Cancer diagnosis remains a critical area for advancing healthcare technologies.

Purpose of the Study:

• To systematically review deep learning models for automated cancer diagnosis.
• To analyze DL models applied to breast, lung, liver, brain, and cervical cancers.
• To compare the efficacy of different DL models in early cancer detection.

Main Methods:

• Systematic review of research articles published between 2016 and 2022.
• Identification and categorization of various deep learning models used in cancer diagnosis.
• Comparative analysis of model performance for early-stage cancer detection.

Main Results:

• Convolutional neural network (CNN) models demonstrated appreciable accuracy in automated cancer diagnosis.
• Pretrained models were frequently utilized for enhanced diagnostic performance.
• The review identified several shortcomings in current DL-based automated cancer diagnosis systems.

Conclusions:

• Deep learning, especially CNNs, offers significant potential for accurate and early cancer diagnosis.
• Further research is needed to address existing limitations and improve DL model robustness.
• Future directions include enhancing DL models for more effective automated cancer detection.