A Systematic Review of Real-Time Deep Learning Methods for Image-Based Cancer Diagnostics.

Area of Science:

• Medical image processing
• Artificial intelligence in healthcare
• Oncology

Background:

• Current cancer diagnosis methods involve lengthy waiting periods (5-30 days).
• Deep learning (DL) excels at identifying complex patterns in large datasets, making it suitable for medical image analysis.
• Real-time medical diagnosis aims to identify conditions within a set timeframe.

Purpose of the Study:

• To explore the application of deep learning algorithms for real-time cancer diagnosis.
• To evaluate the impact of DL on reducing diagnostic waiting times.
• To assess the accuracy and efficiency of DL in medical imaging for cancer detection.

Main Methods:

• Comprehensive literature review on DL in cancer diagnosis.
• Evaluation of DL model accuracy and turnaround times across various imaging modalities.
• Analysis of infrastructure requirements and costs for real-time DL diagnostics.

Main Results:

• Deep learning models, particularly convolutional neural networks, achieve up to 99.3% accuracy in cancer diagnosis.
• DL has the potential to significantly decrease the waiting period for cancer diagnosis.
• Effectiveness and cost-efficiency of DL infrastructure are assessed.

Conclusions:

• Explainable DL is crucial for overcoming barriers like generalization issues and data variability in clinical trials.
• Implementing DL can enhance the speed and accuracy of cancer diagnosis.
• Explainable AI will facilitate the broader adoption of DL in clinical cancer diagnosis.