A Survey on Human Cancer Categorization Based on Deep Learning.

Area of Science:

• Medical Imaging Analysis
• Artificial Intelligence in Oncology
• Computational Pathology

Background:

• Deep learning (DL) and advanced graphics processing units (GPUs) have accelerated AI in medicine.
• Convolutional Neural Networks (CNNs) are adept at classifying histopathological images for disease diagnosis.
• DL applications are rapidly expanding in medical diagnostics, especially for various cancer types.

Purpose of the Study:

• To assess key deep learning concepts for medical image analysis.
• To survey current applications of DL in cancer categorization and medical imaging.
• To review the evolution of CNN architectures and identify future research trends.

Main Methods:

• Review of deep learning techniques, focusing on CNNs for image classification.
• Analysis of DL applications in object detection, classification, and cancer categorization.
• Examination of CNN architectures from AlexNet to VGG and Google networks.

Main Results:

• Deep learning, especially CNNs, demonstrates significant potential in classifying histopathological images.
• CNNs effectively detect visual features for diagnosing cancers like lung, skin, brain, prostate, and breast.
• The study provides an overview of DL systems, data mining techniques, and popular cancer types.

Conclusions:

• Deep learning offers powerful tools for medical image investigation and cancer diagnosis.
• Continued advancements in CNN architectures promise enhanced diagnostic accuracy and efficiency.
• Future research should address emerging challenges and trends in vision-based DL for medicine.