Meet Research Grant Recipient: Khalid Malik, PhD

Dr. Khalid Malik is the recipient of the Shirley Dudek Demmer and Kati Lorge Chair of Research for $35,000.

BAF: Please tell us your background, where you are from, schooling, etc.


KM: I am a Distinguished Associate Professor at the School of Engineering and Computer Science, located at Oakland University in Rochester, MI. I obtained my Ph.D. from Tokyo Institute of Technology in Japan in 2010. In addition to my academic research, I have gained valuable expertise through more than 8 years of working on large-scale industrial R&D projects at Sanyo Electric and DTS Inc in Japan. My research endeavors encompass a wide range of areas, with a primary focus on the creation of intelligent clinical decision support systems. These systems utilize medical image analytics and text analytics based on Neuro-symbolic AI to predict strokes and provide valuable insights. My other research interests include Fake multimedia detection with a focus on deepfake detection and automotive security. My research is funded by the National Science Foundation (NSF), Department of Energy (DoE), US Army, Michigan Translational Research & Commercialization (MTRAC), Brain Aneurysm Foundation (BAF), and other national and international agencies.

BAF: What led you to become involved with brain aneurysm research?


KM: While working on a project relevant to data integration aimed at developing automated decision tree-based guidelines for aneurysm treatment at Henry Ford Health Systems (HFHS) in 2015, I encountered the problems that neurosurgeons face regarding the difficulty of predicting aneurysm rupture and establishing a common consensus among the team regarding borderline cases. This is what motivated me to start working on developing a clinical decision support tool that can assist neurosurgeons in predicting the probability of aneurysm rupture. I am grateful to Dr. Ghaus Malik, Executive Vice Chair of Neurosurgery at HFHS, West Bloomfield, who continuously kept me motivated and further encouraged me to pursue this research challenge. Although my initial goal was to help neurosurgeons develop guidelines for a treatment plan, I realized that nothing is more important than developing a system to assist neurosurgeons in accurately predicting aneurysm rupture.

BAF: In the simplest terms, what is the purpose of your project?


KM: The purpose of the prAI and subarachnoid hemorrhage (SAH) prediction tool is to enable neurosurgeons to make correct and timely treatment decisions, help create consensus among them about treatment plans, mitigate misdiagnosis of life-threatening SAH, save time (of reviewing historical patient data), educating neurology and neurosurgical residents about Human-Centered AI, and reduce costs for unnecessary follow-up studies and treatments. This research will individualize the rupture risk of aneurysm based on its location, (considering 18 arterial locations), 15 clinical risk factors from EHR, and neurosymbolic AI-based morphological & geometrical features of aneurysms derived from MRA/DSA/CTA. Finally, this groundbreaking study will utilize distributed AI, enabling knowledge acquisition from diverse hospital datasets for tool development without compromising data privacy.

BAF: In the simplest terms, what do you hope will change through your research findings?


KM: The treatment decisions of Intracranial Aneurysms are complex: some small (<6 mm) aneurysms at a given location have a high propensity of Subarachnoid Hemorrhage (SAH) while larger aneurysms at the same or other arterial locations do not. Thus, treatment of all aneurysms is not prudent, especially when the benefit of aneurysmal occlusion does not outweigh the risk of observation alone. There is a need to address this complex challenge by developing reliable rupture risk of aneurysm using AI techniques and estimating the appropriate time for treatment. We will extend our previous work using explainable AI-based location-specific and global machine learning models.  The tool will provide help to understand how a group of risk factors compared to other combinations carry less risk (i.e., female, small-sized, Paraclinoid vs. female, tiny-sized, Anterior Communicating Artery). The novelties of project include: 1)Prediction of aneurysm’ growth rate and probability of its rupture using neuro-symbolic and multimodal AI; 2) Performing morphological analysis and topological assessment using neuro-symbolic AI, and geometrical quantification techniques applied to imaging modalities (MRA, DSA); and 3)Information-theoretic based multimodal fusion of feature sets extracted from image modalities and semantic analysis of Electronic Health Record (EHRs).

BAF: Why is the funding you are receiving through the Brain Aneurysm Foundation so important?


KM: The funding of the Brain Aneurysm Foundation is crucial to address research problems that haven’t been explored earlier due to the lack of large-scale EHR and brain imaging data and intradisciplinary team. Essentially, Brain Aneurysm Foundation funding will enable our team to :a)  extend our proposed aneurysm detection algorithms using DSA images and make them applicable to MRA images as well b) predict the growth rate using DSA, and MRA images along with clinical features from EHR (dataset of 2700 patients) of Neuroscience Institute, Henry Ford Hospital, MI c)  study the surface features of the aneurysm using MRA and DSA and their correlation with aneurysm rupture prediction d) to calculate geometrical features of the aneurysm from MRA and DSA e) fuse useful clinical, morphological, geometrical features obtained from DSA, MRA and EHR to predict the aneurysm rupture e) to develop and share the data with other researchers to advance the research of brain aneurysm. The obtained results of this work will be used to get MTRAC, NIH, and NSF grants to complete the proposed framework.

Pictured above is Dr. Malik’s research poster, “Integrating Federated and Neurosynbolic AI for Improved Aneurysm Detection and Rupture Prediction”