Livia Schiavinato Eberlin, of Baylor University, will deliver a seminar entitled, "Mass spectrometry, molecular data, and cancer diagnosis: Advances and challenges towards clinical use." Hosted by The Graduate Students, represented by Symara de Melo Silva.
Seminar: Livia Eberlin, Baylor

Add To Calendar

Zoom Link: https://nyu.zoom.us/j/95330445149?pwd=dWJ4SkNZVHErekdKTE9SRXVlQUJ1Zz09
Abstract: Mass spectrometry is a powerful technology to acquire chemical and molecular profiles from biological samples with unparalleled sensitivity and chemical specificity. Using direct analysis by mass spectrometry, hundreds of molecular ions including metabolites, lipids and protein species can be rapidly detected and identified from clinical specimens. In my research group, we develop and employ mass spectrometry technologies to rapidly analyze and diagnose clinical samples with the goal of expediting clinical decision making and improving treatment and outcome for patients. In this presentation, I will highlight our research using desorption electrospray ionization and statistical classifiers to diagnose clinical biopsies of indeterminate thyroid tumors in order to improve preoperative diagnosis for patients. Classification models pinpointing the important role of several diagnostic metabolites correlated to gene expression results will be discussed. I will also describe our research developing the MasSpec Pen technology as a handheld device integrated to a mass spectrometer for detection of rich molecular profiles directly from in vivo and ex vivo tissues on clinically relevant timescales (<15 seconds). The complex molecular data generated by the MasSpec Pen used in conjunction with machine learning methods allows for the development of powerful statistical models capable of distinguishing disease states with high accuracies (92-98%, depending on tissue type). In particular, I will discuss my team's effort translating and testing the MasSpec Pen technology in the operating room, and our findings applying classification models to diagnose intraoperative data. Challenges and opportunities to improve data analysis and statistical classification methods will also be presented.