Machine learning and deep learning approach for medical image analysis: diagnosis to detection

Area of Science:

• Medical Imaging and Diagnostics
• Artificial Intelligence in Healthcare
• Computational Pathology

Background:

• Early disease detection via medical imaging is crucial for reducing cancer and tumor mortality rates.
• Machine Learning (ML) faces limitations with large datasets, while Deep Learning (DL) excels with varying data volumes.
• DL utilizes multilayered neural networks for enhanced data analysis compared to traditional ML.

Purpose of the Study:

• To systematically review the applications of ML and DL in detecting and classifying multiple diseases from medical images.
• To provide an overview of ML/DL approaches, imaging modalities, evaluation tools, and datasets.
• To experimentally compare ML classifiers and DL models using an MRI dataset.

Main Methods:

• Systematic literature review of 40 primary studies published between January 2014 and 2022.
• Analysis of various ML and DL techniques for disease detection and classification.
• Comparative experimental analysis of ML classifiers and DL models on an MRI dataset.

Main Results:

• DL demonstrates superior performance with varying data amounts compared to ML.
• The review categorizes different ML/DL approaches, imaging modalities, and evaluation metrics.
• Experimental results provide a direct comparison of ML and DL model efficacy on MRI data.

Conclusions:

• DL is an enhanced technique of ML, adept at handling complex medical imaging data for disease detection.
• This study offers guidance for healthcare professionals to select optimal diagnostic techniques.
• The findings aim to reduce diagnosis time and improve accuracy in clinical practice.