At Intelligent Medical Informatics Computing Systems (IMICS) Lab at The Hospital for Sick Children and University of Toronto, our research focuses on investigating and developing advanced Artificial Intelligence (AI) algorithms to revolutionize precision medicine through the integration of diverse healthcare data. By leveraging medical imaging modalities such as MRI and CT, radiology reports, and clinical informatics like patient history, we aim to push the boundaries of disease management and personalized care.
Our core objective is to create, implement, and validate cutting-edge AI models that deliver robust capabilities for early diagnosis, precise risk stratification for prognostication, and accurate assessment of treatment response across a wide range of diseases. These tools are designed to guide clinicians in making informed decisions about the type, timing, and intensity of therapy, ultimately optimizing patient outcomes and enhancing healthcare delivery.
To achieve this, we are committed to developing human-centered and compassionate AI tools. These solutions prioritize explainability and transparency, empowering clinicians to understand, trust, and effectively interact with AI systems. Moreover, our work emphasizes using AI as a vehicle for equitable healthcare, addressing disparities by enabling access to high-quality diagnostics and care for underserved populations. By aligning technological innovation with the principles of empathy and inclusivity, we aspire to transform healthcare into a system that benefits all, ensuring no one is left behind.
News
-
This paper reviews how artificial intelligence (AI) will play an increasingly important role in pediatric neuroradiology in the future. A safe, transparent, and human-centric AI is needed to tackle the quadruple aim of improved health outcomes, enhanced patient and family experience, reduced costs, and improved well-being of the healthcare team in pediatric neuroradiology. Published in American Journal of Neuroradiology.
-
In this work, we use an Eye-Gaze dataset along with radiology reports to exhaustively study the impact of adding radiology reports and Eye-Gaze data to X-ray images on the performance and explainability of CNN-based classification models. Additionally, we introduce an explainability metric to quantitatively evaluate the alignment of model attention with radiologist-specified regions of interest (ROIs). Published in MICCAI Workshop Proceedings.
-
In this work, we have introduced a novel end-to-end pipeline that enhances classification by incorporating ROIs generated by a weakly supervised method as an auxiliary task in a deep multitask learning framework. We have demonstrated that our approach outperforms conventional methods that only rely on classification. By using weakly supervised segmentation, we are able to leverage pixel-wise information, which ultimately leads to improved classification performance. Published in MICCAI Workshop Proceedings.
-
Our latest work on the prediction of the most common molecular subtypes of pediatric low-grade gliomas (pLGG) using MRI-based radiomics, published in European Radiology. This is an important step towards noninvasvie molecular diagnosis of pLGG, with significant potential for prognostication and individualized treatment strategies for patients with pLGG. Also, we have mapped radiomic features onto a nomogram to make it easier for clinicians to use and trust the model.
-
Our latest work on MRI-based end-to-end pediatric low-grade glioma segmentation and classification was published in the Canadian Association of Radiologists Journal.
IMICS Lab Research Is Generously Funded By:
IMICS LAB 