- Artificial intelligence is expected to play a huge role in transforming radiology practice.
- Therefore, we aim to provide a explanation on the very basics of artificial intelligence.
- Buzzwords of AL/ML/DL.[1]
Fig. 1: Key terms that have been used throughout the text, together with their short definitions.
References: author
Fig. 2: A schematic overview of AI, machine learning and deep learning.
References: Six, O., & Quantib B.V. (n.d.). The ultimate guide to AI in radiology. Retrieved from https://www.quantib.com/the-ultimate-guide-to-ai-in-radiology.
Introduction to deep learning in radiology :
A- How does machine learning works?
Fig. 3: How Machine learning works?
References: https://edhub.ama-assn.org/jn-learning/video-player/16845576
Fig. 4: How machine learning works ?
References: https://edhub.ama-assn.org/jn-learning/video-player/16845576
Fig. 5: How machine learning works?
References: https://edhub.ama-assn.org/jn-learning/video-player/16845576
B-The difference between machine learning and Deep learning:
Fig. 6: Machine learning Vs. Deep learning
References: https://medium.com/m/global-identity?redirectUrl=https%3A%2F%2Ftowardsdatascience.com%2Fcnn-application-on-structured-data-automated-feature-extraction-8f2cd28d9a7e
C- The architecture of deep learning models :
- we introduce the artificial neural network in general and one specific type: the convolutional neural network.
Fig. 7: Building blocks of a typical CNN.
References: Lundervold, Alexander Selvikvåg, and Arvid Lundervold. “An Overview of Deep Learning in Medical Imaging Focusing on MRI.” Zeitschrift Für Medizinische Physik, vol. 29, no. 2, 2019, pp. 102–127., doi:10.1016/j.zemedi.2018.11.002.
Fig. 8: Artificial Neural Networks (ANN) and Convolutional Neural Networks (CNN)
References: Gogul, I. & Kumar, Sathiesh. (2017). Flower species recognition system using convolution neural networks and transfer learning. 1-6. 10.1109/ICSCN.2017.8085675.
Fig. 9: The difference between artificial and convolutional neural network
References: author
Fig. 10: A diagram showing typical architecture of CNN
References: Mazurowski, Maciej & Buda, Mateusz & Saha, Ashirbani & Bashir, Mustafa. (2018). Deep learning in radiology: an overview of the concepts and a survey of the state of the art. Journal of Magnetic Resonance Imaging. 49. 10.1002/jmri.26534.
C-The learning process of CNN
- More sources are needed to enhance the deep learning systems.
- Most straightforward way of training is to start with a random set of weights and train using available data specific to the problem being solved (training from scratch)[3]
Fig. 11: Transfer learning of CNN
References: https://medium.com/m/global-identity?redirectUrl=https%3A%2F%2Ftowardsdatascience.com%2Fa-comprehensive-hands-on-guide-to-transfer-learning-with-real-world-applications-in-deep-learning-212bf3b2f27a