Personalized diagnosis of non-lesional epilepsy using simultaneous PET/MRI

The vision of personalized medicine is to customize medical diagnosticreatment to the individual characteristics of each patient. While considerable attention in personalized medicine has been paid to the use of genetic tests to guide therapeutic decisions, new advances in medical imaging now can be used to better diagnose patient-specific abnormalities. As a result, image-based therapy approaches can be more finely tuned to the underlying malignancies of the individual patient rather than using a "one-size-fits-all" approach. Traditionally, diagnosis and therapy planning are based on clinical imaging modalities, such as Computed Tomography (CT), Magnetic Resonance Tomography (MRI) and nuclear medicine techniques such as Positron Emission Tomography (PET). Here, CT provides anatomical image information whereas PET helps quantify tissue metabolism as an indicator of a biological abnormality. MRI is a technique that can provide both anatomical and limited functional information. Until recently, these methods were used independently of each other. Today, combined imaging methods, such as PET/MR have been developed that provide integrated functional and anatomical images of patients based on simultaneously acquired data. The so obtained images can be fused with utmost precision, and rapidly acquired MR images can be used to correct the PET images that take longer to acquire for misalignments from involuntary patient motion. Hence, PET images from the PET/MR can in theory be of higher quality than PET images from a standalone PET system. Both, PET and MRI are modalities of choice for a number of neurological imaging questions. We hypothesize that the combined use of these modalities can help improve patient management in neurology, specifically patients with epilepsy and non-specified epileptic lesions. Accordingly, the objective of this proposal is the development of a simultaneous PET/MR imaging protocol that takes advantage of the complimentary nature of structural and functional information in order to create a personalized imaging protocol for each patient. Specifically, we hypothesize that patient-specific molecular imaging information provided by PET will guide high-resolution MR imaging of specific areas of an individual patients brain that are crucial for the clinical interpretation of the underlying pathophysiology, thereby adapting the imaging protocol to the unique disease characteristics of the individual subjects. In turn, this will maximize the obtained clinical information within one imaging session and we anticipate that this novel approach will improve surgical outcome in patients with clinically problematic non-lesional epilepsy where scalp EEG alone does not help.

In this research project, we have developed a non-invasive methodology to quantify how active an individual's brain is. Similar to a smartwatch, that measures the physical activity of an individual, we used Positron Emission Tomography (PET) system to measure the level of neuronal activity of the brain. Brain metabolism is considered as an established proxy for Neuronal activity. In PET brain studies involving metabolism, trace amounts of radioactive sugar (radiotracer), known as fluorodeoxyglucose, is injected to the subject. The radioactive sugar being unstable undergoes annihilation followed by the emission of two photons that travel in the opposite direction to each other. The PET detectors detect the emitted photons, which, in turn, will be used to quantify the distribution of the radiotracer. To accurately quantify brain metabolism, one needs to measure the quantity of non-metabolised radioactive sugar in the blood as a function of time, through an arterial puncture. Since it is an invasive and painful procedure that causes significant discomfort to the patients, the clinicians do not recommend this. In this research, we have developed a smart pipeline, that automatically measures this information non-invasively directly from images from an integrated PET/MRI. The software is fully automatic and modular, requiring no user input. Following the validation of the software, by testing it on ten healthy volunteers. The software was used to calculate the metabolic activity of the brain of non-lesional epilepsy patients. The software played a pivotal part in supporting the initial hypothesis of the clinicians by highlighting the abnormal metabolic activity of the brain. We have made the software open-source in Github: https://github.com/LalithShiyam/AQuaPi, to promote rapid translation to other clinics.