Alessandro Perelli

Alessandro Perelli

Lecturer in Biomedical Engineering

University of Dundee

My research interests include computational imaging, and machine learning, in particular for polyenergetic 3D/4D X-ray CT reconstruction and my specialism and passionate interests span to randomized linear algebra, stochastic iterative algorithms, iterative sketching, probabilistic graphical models and variational inference.

I am dealing with all aspects of the image reconstruction, from the theoretical mathematical modelling, through the algorithm development, to the handling and processing of experimental CT data together with physical modeling.

Interests
  • Computational Imaging
  • Deep Learning
  • Medical Applications
  • Computed Tomography
  • Magnetic Resonance Imaging
Education

  • PhD in Electrical Engineering and Computer Science, 2014

    University of Bologna (Italy)

  • MSc in Electrical Engineering, 2010

    Università Politecnica delle Marche (Italy)

  • BSc in Electrical Engineering, 2007

    Università Politecnica delle Marche (Italy)

News

 
 
 
 
 
China Scholarship Council (CSC) PhD research projects on Medical Image Computing
December 2023 – January 2024
Different PhD projects related to medical image computing are available for the Chinese national through the CSC scholarship programme: 1) Multimodal Medical Image Reconstruction and Fusion using Generative Deep Learning, 2) Deep Learning Cardiac MRI reconstruction with motion compensation, 3) Deep Learning for Computational Imaging in Spectral Computed Tomography. Deadline 31 January 2024. Talented students can enquiry: please email me (aperelli001@dundee.ac.uk).
 
 
 
 
 
Open applications for a self-funded PhD research project on Medical Imaging
December 2023 – February 2024
A self-funded PhD research project on “Robust Deep Learning for Medical Image Reconstruction” is advertised on FindAPhD. Deadline 28 February 2024. For enquiries please email me (aperelli001@dundee.ac.uk).
 
 
 
 
 
IEEE Transactions on Radiation and Plasma Medical Sciences
New Paper: Uconnect Synergistic Spectral CT Reconstruction With U-Nets Connecting the Energy Bins
IEEE Transactions on Radiation and Plasma Medical Sciences
November 2023 – Present
A novel synergistic method for spectral CT reconstruction, namely Uconnect is published in IEEE TRPMS. It utilizes trained convolutional neural networks to connect the energy bins to a latent image so that the full binned data is used synergistically.
 
 
 
 
 
IEEE Transactions on Radiation and Plasma Medical Sciences
New Paper: Systematic Review on Learning-based Spectral CT
IEEE Transactions on Radiation and Plasma Medical Sciences
October 2023 – Present
In this review, we present the state-of-the-art data-driven techniques for spectral CT. Check out in IEEE TRPMS.
 
 
 
 
 
Opportunities for prospective Postdoc and PhD students
March 2022 – Present
If you are a strong Postdoc or PhD candidate in computational imaging interested in applying for individual fellowship, please contact me to discuss innotavive projects (aperelli001@dundee.ac.uk).
 
 
 
 
 
Physics in Medicine & Biology
New Paper: Multi-channel Convolutional Analysis Operator Learning for Dual-energy CT Reconstruction
Physics in Medicine & Biology
January 2022 – Present
Advanced dictionary learning method for Dual-energy CT reconstruction is published in PMB. The code is now available on GitHub. Check it out.

Projects

*
All Spectral CT Deep Learning Optimization Bayesian Inference
30 Oct 2023
Uconnect: Synergistic Spectral CT Reconstruction With U-Nets Connecting the Energy Bins
Uconnect: Synergistic Spectral CT Reconstruction With U-Nets Connecting the Energy Bins

We propose a novel synergistic method for spectral CT reconstruction, namely Uconnect. It utilizes trained convolutional neural networks to connect the energy bins to a latent image so that the full binned data is used synergistically.

1 Sep 2023
Systematic Review on Learning-based Spectral CT
Systematic Review on Learning-based Spectral CT

Spectral computed tomography (CT) has recently emerged as an advanced version of medical CT and significantly improves conventional (single-energy) CT. Spectral CT offers image improvement, material decomposition, and feature quantification relative to conventional CT.

4 Oct 2022
LRR-CED low-resolution reconstruction-aware convolutional encoder
LRR-CED low-resolution reconstruction-aware convolutional encoder

We propose a direct deep learning-based reconstruction that exploits the information from low-dimensional scout images, to learn the projection-to-image mapping.

25 Jan 2022
Convolutional Operator Learning for Imaging
Convolutional Operator Learning for Imaging

Data-driven and learning-based approaches have gained much interest in recent years for medical image reconstruction. These methods learn representations of images and are used in combination with model-based techniques to perform complex mappings between corrupted measurements and high-quality images.

1 Nov 2021
Regularized Stochastic Optimization by Denoising
Regularized Stochastic Optimization by Denoising

The main challenges of the project are related to the fact that biomedical image reconstruction requires to recover input data lying in an high-dimensional space and to leverage more sophisticated or learned data-dependent priors which are needed since the estimation problem is generally ill-posed.

1 Oct 2020
Bayesian Inference for Inverse Problems
Bayesian Inference for Inverse Problems

In this project, we investigate the use of an emerging reconstruction method from Compressed Sensing, called Approximate Message Passing, for medical image reconstruction. furthermore, we study Expectation-Propagation methods for approximate Bayesian inference.

Experience

 
 
 
 
 
University of Dundee
Lecturer in Biomedical Engineering
University of Dundee
Sep 2021 – Present Dundee, United Kingdom
 
 
 
 
 
Researcher
French National Institute of Health and Medical Research (INSERM)
Dec 2020 – Aug 2021 Brest (France)
Spectral Computed Tomography, Image reconstruction, Deep learning.
 
 
 
 
 
H.C. Ørsted COFUND Postdoctoral Fellow
Technical University of Denmark (DTU)
Dec 2018 – Nov 2020 Kongens Lyngby (Denmark)
Individual Fellowship - Deep and Randomized Imaging For Tomography (DRIFT).
Machine learning and randomized linear algebra methods in sensing, computation for spectral Tomography.
 
 
 
 
 
Research Associate
The University of Edinburgh
Jul 2014 – Nov 2018 Edinburgh (UK)
a) ERC Research Project C-SENSE - Exploiting low dimensional signal models in sensing, computationand signal processing.
b) Compressed Sensing for advanced X-ray Computed Tomography imaging.
Project funded by U.S. Department of Homeland Security with GE Global Research (GRC).

Contact