A survey on gene expression data analysis using deep learning methods for cancer diagnosis

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Gene expression data analysis is crucial for disease diagnosis, particularly in cancer treatment, by identifying variations in gene expression levels.
• DNA microarrays are effective for gene expression classification and cancer prediction.
• Deep learning (DL) is increasingly utilized in healthcare due to computational advancements.

Purpose of the Study:

• To review existing research on deep learning techniques for cancer classification and prediction using gene expression data.
• To highlight the need for data augmentation in gene expression datasets due to high dimensionality (limited samples, numerous features).

Main Methods:

• Review of deep learning techniques including Feed Forward Neural Network (FFN), Convolutional Neural Network (CNN), Autoencoder (AE), and Recurrent Neural Network (RNN).
• Focus on the application of these methods to gene expression data analysis for cancer-related tasks.
• Discussion of data augmentation strategies to address the dimensionality challenge in gene expression datasets.

Main Results:

• Deep learning models demonstrate potential in extracting meaningful patterns from complex gene expression data.
• Various DL architectures are applicable for classifying and predicting cancer types.
• Data augmentation is essential for improving the performance of DL models on high-dimensional gene expression datasets.

Conclusions:

• Deep learning offers powerful tools for analyzing gene expression data for cancer diagnosis and prediction.
• Further research into DL techniques and data augmentation is warranted for advancing precision oncology.
• The reviewed methods provide a foundation for developing more accurate and efficient cancer detection systems.