Deep Learning in Selected Cancers' Image Analysis-A Survey.

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Oncology

Background:

• Deep learning algorithms are increasingly preferred for medical image analysis, including cancer detection.
• Applications span various cancers like breast, cervical, brain, colon, and lung, utilizing diverse imaging modalities.
• Significant research exists in developed nations, but African research remains limited.

Purpose of the Study:

• To survey and review deep learning approaches for cancer image analysis.
• To highlight the state-of-the-art performance of deep learning in tumor detection, segmentation, and classification.
• To identify disparities in research focus, particularly the underrepresentation of African contributions.

Main Methods:

• Systematic review of deep learning applications in medical imaging for cancer.
• Analysis of different deep learning implementation modes: training from scratch, transfer learning, and architectural modification.
• Comparative assessment of research focus across geographical regions.

Main Results:

• Deep learning methods achieve state-of-the-art results in tumor detection, segmentation, feature extraction, and classification.
• Three primary deep learning implementation strategies were identified: training from scratch, transfer learning, and network architecture modification.
• Research is concentrated in economically developed countries, with limited attention in Africa.

Conclusions:

• Deep learning is a powerful tool for cancer image analysis, achieving superior performance.
• Standardized approaches and broader adoption of deep learning are crucial for global cancer diagnostics.
• Urgent attention and investment are needed to bridge the research gap in Africa for cancer imaging analysis.