Teaching

Institute of Medical Science

Fall 2022

MSC1113H – Radiomics and Machine Learning in Medical Imaging

Session Offered: 20229
Date: November 3, 2022 – December 8, 2022 (Thursdays), Time: 10:00 am -12:00 pm, Location: In person ( Medical Sciences Building MS 3174)

Lecture 1: Introduction to Radiomics-based Precision Medicine and Medical Imaging Modalities, Hypothesis-driven Radiomics Analytics, Medical Image Preprocessing

Lecture 2: First order Radiomic Features, ROC Analysis, Second order Radiomic Features

Lecture 3: 2D Image Filtering and Transformation (local processing algorithms, e.g., correlation, convolution, gradient, wavelet), Radiomics Tabular Data Preprocessing, Unsupervised and Supervised Feature Selection, Logistic Regression and Random Forest Classifier

Lecture 4: Convolutional Neural Networks for Medical Imaging

Lecture 5: Deep Transfer Learning for Radiomics and Medical Imaging

Lecture 6: U-Net for Medical Image Segmentation, Generative Adversarial Networks (GAN) for Anomaly Detection in Medical Imaging

Summer 2021

MSC1113H – Radiomics and Machine Learning in Medical Imaging

Session Offered: 20215
Date: Tuesdays, June 1 – August 10, 2021, Time: 10 am – 12 noon, six sessions alternating weeks; Location: Online

Description:  This is a 6-week modular course (0.25 FCE) designed to provide an understanding of fundamentals of Artificial Intelligence / Machine Learning (AI/ML) for wide applications in medical imaging. Medical Imaging, which is an important specialty in medicine for diagnosis, prognosis, and intervention of different types of diseases including cancer, is increasingly moving toward quantitative approaches. AI/ML algorithms are playing a key role in quantitative medical imaging analytics for disease diagnosis (detection) and prognosis (prediction). With the help of recent advances in AI/ML and computer vision, novel predictive models are capable of diagnosing a disease with high accuracy and consistency, and predicting clinical outcomes (e.g., response to treatment) with an accuracy, which is beyond existing clinical methods.

Lecture 1: Introduction to Radiomics-based Precision Medicine and Medical Imaging Modalities

Lecture 2: Hypothesis-driven Radiomics Analytics, Histogram-based Image Transformation (point processing algorithms, e.g., Gamma correction, histogram equalization), First-order Radiomic Features, Logistic Regression, ROC Analysis

Lecture 3: 2D Image Filtering and Transformation (local processing algorithms, e.g., correlation, convolution, gradient, wavelet), Second-order Radiomic Features

Lecture 4: Radiomics Tabular Data Preprocessing, Unsupervised and Supervised Feature Selection, Random Forest and Support Vectored Machine Classifiers

Lecture 5: Convolutional Neural Networks and Transfer Learning for Radiomics and Medical Imaging

Lecture 6: Radiomics and AI in Medical Imaging: Challenges and Limitations (Repeatability and Reproducibility, Confidence and Reliability, Explainability)