Thesis Project Form

Histopathological examination of tissue biopsies is a very important diagnostic tool for a wide range of pathologies. Diagnosis and prognosis are typically based on the manual exploration of a pathologist and qualitative parameters depending on prior experience and subjective evaluation.
In recent years, deep learning methods have shown accurate results for the automatic diagnosis of different pathologies based on the analysis of histopathological images, appearing as a promising tool for supporting pathologists in accurately providing diagnosis and prognosis

This project aims to explore the combination of classic optical microscopy images of histological tissue samples with mechanical and topographical features at the nanoscale measured by atomic force microscopy, in an attempt to develop novel models with increased diagnostic accuracy. The candidate is expected to design models capable of combining optic images with AFM features, performing several tests and ablation studies to assess the usefulness of the AFM features in the histopathological diagnosis domain. Specifically, this project will focus on the analysis of histopathological images of meninges tissues, and the candidate will analyze available histopathological datasets acquired for this project to realize accurate diagnosis of meningiomas, which are brain tumors growing from the meninges.

You will learn/use
experimental methodology: Atomic force microscopy, optical microscopy
analytical tools: deep learning techniques (convolutional neural networks)