#سیلابس_دوره
Syllabus for practical deep learning for medical image analysis:
• Requirements for participants:
• Basic knowledge about an operating system (Windows, Linux or Mac)
• Basic knowledge about command line interfaces (Command Prompt, Terminal, PowerShell or Bash)
• Basics of Python and common modules (os, numpy, matplotlib, …)
• Optional: General knowledge about Deep Learning or Neural Networks

• IDEs and Technologies will be:
• IDEs: Google Colab and one of python IDEs (PyCharm, VS Code or Spyder)
• Package managers: Anaconda, Pip
• Python frameworks: TensorFlow, Keras, HuggingFace
• Source control: Git and GitHub

• Course Contents:
1. Introduction
i. Introduction to AI, Machine Learning and Deep Learning
ii. Basics of Neural Networks + Practice
iii. Deep learning models and training strategies

2. Mathematics of Neural Networks Training
i. Model training cycle
ii. Optimization algorithms (Gradient descent, …)
iii. Back-Propagation algorithm
iv. Important parameters and how to tune them
v. Discussion on Overfitting, Bias and Variance

3. Introduction to TensorFlow and Keras
i. Types of Data
ii. Layers and Models in Keras
iii. Training process
iv. Keras Callbacks (Model Checkpoint, Logger, Tensor Board, …)
v. Save/Load models

4. Computer Vision in Deep Learning
i. Convolutional Neural Network (CNN) models
• Conv layers
• Subsampling layers
ii. Data Preparation
• Load and visualize 2D images (using OpenCV, Pillow, Matplotlib, …)
• Pre-processing / Augmentation / Label preparation
• Prepare Train / Validation / Test datasets
iii. Applications
• Image Classification
• Auto-Encoders
• Object Detection
• Segmentation
• Super-resolution
iv. How to use Kaggle datasets
v. Model evaluation and improvement methods
• Transfer Learning, Data augmentation, Regularization, Dropout, …

5. Work with 3D Medical Images
i. Understanding medical image formats
ii. Loading & visualization of medical images (using SimpleITK, Nibabel, Numpy, …)
iii. Medical images pre-processing
iv. Adjusting Deep CNN models for medical images

6. Implementation of existing and pre-trained models
i. Using public codes and models
ii. Using Keras Applications
iii. Using TensorFlow Hub

7. New trends in Deep Learning
i. Transformer models
ii. HuggingFace models and framework

8. Additional topics and discussions