Our lab, the Quantitative Translational Imaging in Medicine (QTIM) lab, at the Athinoula A. Martinos Center for Biomedical Imaging of the Massachusetts General Hospital (MGH), focuses on developing quantitative imaging biomarkers for cancer and other diseases using advanced imaging techniques and machine learning methods, with a particular focus on applying deep learning methods to a variety of diseases. Our goal is to unite the cutting edges of machine learning, medical oncology, and image analysis into practical clinical applications, and to that end, we span the gamut in terms of specialists – computer science researchers, medical physicists, neuro-oncologists, and MRI technicians, among others. Postdocs and graduate students in our lab come from a wide variety of backgrounds and departments, including Harvard/MIT Health Sciences and Technology, MIT Electrical Engineering and Computer Science, and Harvard Biophysics, in addition to MD-PhD students and high-school students/undergraduates from around the world!
Some of our recent works have demonstrated that the use of saliency maps in medical imaging is unreliable and do not meet specific defined criteria of trustworthiness, and that detection or segmentation models trained directly for localization tasks give higher utility than saliency maps. Additionally, we have shown that a distributed deep learning approach can outperform models trained in a single institution, in terms of testing accuracy, thereby highlighting the benefits of collaboration among multiple institutions in the context of deep learning without the need to share data in a common repository, relevant to ProCAncer-I’s work. We are also currently involved in two multicenter clinical trials, which incorporate advance MR imaging to answer clinically and biologically relevant questions pertaining to brain metastases.
When we are not seeing patients in clinic or scripting away in lab, we are also an incredibly sociable (and active!) bunch. As a lab, we have participated in activities including but not limited to picnics, sightseeing, rock climbing, kayaking, apple picking, and the ever-popular Spikeball®, a lab-favorite!
We look forward to being a part of the ProCAncer-I collaboration!