Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Online learning for X-ray, CT or MRI

Mosabbir Bhuiyan, MD Abdullah Al Nasim, Sarwar Saif, Kishor Datta Gupta, Md Jahangir Alam, Sajedul Talukder

TL;DR

The paper addresses the challenge of early, accurate disease detection from X-ray, CT, and MRI images by surveying AI-enabled approaches. It outlines a methodology pipeline spanning image processing, handcrafted feature-based ML, and deep learning, with emphasis on custom CNNs, transfer learning, and CNN-ML hybrids. Key contributions include a structured synthesis of techniques and performance metrics, highlighting that deep learning provides automatic feature extraction and superior diagnostic accuracy, as evidenced by metrics like ROC/AUC and F1. The work underscores the practical impact of computer-aided diagnosis in radiology, enabling faster, more reliable interpretation and potential privacy-preserving collaborations across institutions.

Abstract

Medical imaging plays an important role in the medical sector in identifying diseases. X-ray, computed tomography (CT) scans, and magnetic resonance imaging (MRI) are a few examples of medical imaging. Most of the time, these imaging techniques are utilized to examine and diagnose diseases. Medical professionals identify the problem after analyzing the images. However, manual identification can be challenging because the human eye is not always able to recognize complex patterns in an image. Because of this, it is difficult for any professional to recognize a disease with rapidity and accuracy. In recent years, medical professionals have started adopting Computer-Aided Diagnosis (CAD) systems to evaluate medical images. This system can analyze the image and detect the disease very precisely and quickly. However, this system has certain drawbacks in that it needs to be processed before analysis. Medical research is already entered a new era of research which is called Artificial Intelligence (AI). AI can automatically find complex patterns from an image and identify diseases. Methods for medical imaging that uses AI techniques will be covered in this chapter.

Online learning for X-ray, CT or MRI

TL;DR

The paper addresses the challenge of early, accurate disease detection from X-ray, CT, and MRI images by surveying AI-enabled approaches. It outlines a methodology pipeline spanning image processing, handcrafted feature-based ML, and deep learning, with emphasis on custom CNNs, transfer learning, and CNN-ML hybrids. Key contributions include a structured synthesis of techniques and performance metrics, highlighting that deep learning provides automatic feature extraction and superior diagnostic accuracy, as evidenced by metrics like ROC/AUC and F1. The work underscores the practical impact of computer-aided diagnosis in radiology, enabling faster, more reliable interpretation and potential privacy-preserving collaborations across institutions.

Abstract

Medical imaging plays an important role in the medical sector in identifying diseases. X-ray, computed tomography (CT) scans, and magnetic resonance imaging (MRI) are a few examples of medical imaging. Most of the time, these imaging techniques are utilized to examine and diagnose diseases. Medical professionals identify the problem after analyzing the images. However, manual identification can be challenging because the human eye is not always able to recognize complex patterns in an image. Because of this, it is difficult for any professional to recognize a disease with rapidity and accuracy. In recent years, medical professionals have started adopting Computer-Aided Diagnosis (CAD) systems to evaluate medical images. This system can analyze the image and detect the disease very precisely and quickly. However, this system has certain drawbacks in that it needs to be processed before analysis. Medical research is already entered a new era of research which is called Artificial Intelligence (AI). AI can automatically find complex patterns from an image and identify diseases. Methods for medical imaging that uses AI techniques will be covered in this chapter.
Paper Structure (11 sections, 7 equations, 15 figures, 2 tables)

This paper contains 11 sections, 7 equations, 15 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Related Works
  3. Methodology
  4. Image Processing
  5. Machine Learning
  6. Deep Learning
  7. Custom CNN
  8. Transfer Learning
  9. CNN-ML
  10. Performance Analysis
  11. Conclusion

Figures (15)

  • Figure 1: An X-ray photo of a one-year-old girl who swallowed a sewing pin 100.
  • Figure 2: Modern CT scanner located at the Lochotín University Hospital in Pilsen, Czech Republic 109.
  • Figure 3: Examples of T1-weighted, T2-weighted, and PD-weighted MRI scans 112.
  • Figure 4: Ultrasound image (sonogram) of a fetus in the womb, viewed at 12 weeks of pregnancy (bidimensional scan) 13.
  • Figure 5: Image denoising abd2002real.
  • ...and 10 more figures