École Nationale Supérieure de
Techniques Avancées

Abstract

Essential Thrombocythemia is a rare type of blood disorder associated with an overproduction of platelets by megakaryocytes, which are large cells in the bone marrow. Its rarity makes it hard to study, as such there is still research being done to better understand this condition. One of the most important being the Primary Thrombocythaemia 1 Trial, which created a large set of both genetic data and tissue samples from a sizable cohort of patients with Essential Thrombocythemia. This led to the discovery of prominent genetic differences within the various patients. And most importantly, the discovery of three common gene mutations, and eight genomic subgroups of patients. The team I worked with, developed an algorithm able to extract information about the morphology of megakaryocytes from images of samples of bone marrow (bone marrow trephines). During this project, I was able to find relevant morphological differences between each group of ET-patients. Moreover, I used various machine learning models to classify patients using the morphological features of their megakaryocytes. I was not able to classify patients by their gene mutations. However, classifying patients by their gene mutations showed great results. It was indeed possible to classify one of the subgroups. However, I did not have the time to develop this track, as my internship was coming to an end.