This project is part of the Statistical Methods for Machine Learning course at Università degli Studi di Milano, and it's done in collaboration with EveryWare Lab based on data from Policlinico of Milan. The goal is to detect pathological images by first training a network on image reconstruction of healthy examples.
Given the nature of the project the data cannot be disclosed. The code avaialble in this repository assumes that the reader is authorized to access it by remotely connecting to the cluster of Università degli Studi di Milano, where data is stored. For the same reason the output of pieces of code which display the original ultrasound images has been omitted, but it can be reproduced by running the relative cells.