This course is outdated because it is based on pytorch lightning and alot of thing has been changed since the release of this course. Further some of datasets in this course are no more available for public anymore. So I am not providing support for this course. I want to make this course free, but udemy is not allowing to do so because of content length. The reason why I am not archiving this course, because its still relevant if you want to gain concept of medical imaging competition.
Greetings. This course is not intended for beginners, and it is more practically oriented. Though I tried my best to explain why I performed a particular step, I put little to no effort into explaining basic concepts such as Convolution neural networks, how the optimizer works, how ResNet, DenseNet model was created etc. This course is for those who have worked on CIFAR, MNIST data and want to work in real-life scenarios
My focus was mainly on how to participate in a competition, get data and train a model on that data, and make a submission. In this course PyTorch lightning is used
The course covers the following topics
Binary Classification
Get the data
Read data
Apply augmentation
How data flows from folders to GPU
Train a model
Get accuracy metric and loss
Multi-class classification (CXR-covid19 competition)
Albumentations augmentations
Write a custom data loader
Use publicly pre-trained model on XRay
Use learning rate scheduler
Use different callback functions
Do five fold cross-validations when images are in a folder
Train, save and load model
Get test predictions via ensemble learning
Submit predictions to the competition page
Multi-label classification (ODIR competition)
Apply augmentation on two images simultaneously
Make a parallel network to take two images simultaneously
Modify binary cross-entropy loss to focal loss
Use custom metric provided by competition organizer to get the evaluation
Get predictions of test set
Capstone Project (Covid-19 Infection Percentage Estimation)
How to come up with a solution
Code walk-through
The secret sauce of model ensemble
Semantic Segmentation
Data download and read data from nii.gz
Apply augmentation to image and mask simultaneously
Train model on NIfTI images
Plot test images and corresponding ground truth and predicted masks