Semantic Segmentation of Prostate Cancer Biopsy Tissue Using Deep Learning

We aim to develop a deep learning-based multiclass segmentation model for prostate cancer biopsy whole-slide images (WSIs). Leveraging the PANDA dataset, we will train a model to classify tissue regions into five categories: background, benign epithelium, stroma, malignant grade 3, grade 4, and grade 5. Unlike the original PANDA challenge, which focused on Gleason score classification, our project will concentrate on semantic segmentation to provide fine-grained annotations of tissue structures.

Self-driving Cars: Semantic Segmentation

Scene understanding is a critical capability for autonomous vehicles. It enables them to interpret and respond to dynamic environments. This project focuses on precise identification of objects in dynamic environments using image segmentation. We will study and train semantic segmentation neural networks to identify key objects in driving scenes, such as roads, pedestrians, vehicles, and more.

Natural Language Processing in Cancer: Extracting Diagnostic Insights from Pathology Reports

This project focuses on leveraging natural language processing (NLP) techniques to extract critical information from pathology reports. Participants will gain hands-on experience with a wide range of text classification methods, from traditional tf-idf analysis to cutting-edge LLMs. The Cancer Genome Atlas (TCGA) pathology report corpus will be used for the development of advanced NLP technologies that can ultimately enhance patient diagnosis, treatment selection, and cancer care.

Breast Cancer Prognosis: Leveraging Cancer Registry Data for Survival Prediction

This project focuses on a survival prediction task using cancer registry data. Participants will gain hands-on experience with key topics, including data cleaning, feature selection techniques, and machine learning modeling tailored to time-to-event data.

Graph Neural Networks for Social and Biological Network Analysis

This project introduces the cutting-edge field of graph neural networks (GNNs) with a focus on embedding actors within social or biological/cellular networks. The aim is to understand how GNNs can capture and represent complex relationships in non-Euclidean data, like social networks. We will explore how these embeddings can effectively represent the intricate connections and interactions between individuals or entities for predictive analytics.

Automated Bladder Cancer Screening with Deep Learning Algorithms

This project focuses on enhancing bladder cancer screening methods by developing algorithms to analyze cell imagery from urine specimens. The goal is to distinguish between cell nucleus and cytoplasm, using various image segmentation algorithms, as a quantitative marker of malignancy.

Cancer Subtype Multi-Class Classification in Gene Expression Data

This project deals with a multi-class classification task (5 tumor types) within RNA-seq gene expression data. This project will introduce participants to topics such as data cleaning, clustering, feature selection methods, and machine learning modeling.

Modeling Complex Binary-Class Associations in Simulated Genomic Data

This project deals with binary classification tasks (case/control) in a variety of simulated single nucleotide polymorphism (SNP) datasets. Each dataset has a different form of underlying complex association (e.g. multivariate additive, epistatic, or genetic heterogeneity). This project will introduce participants to topics such as basic data preparation, feature selection methods, machine learning (ML) modeling algorithms, automated machine learning, and model explanation and interpretation.