Hicam is a research project that is under continuous development by leading European institutes which are Department of Electronic Engineering of Polytechnic institute of Milan, Max-Planck Institute, University of Milan, and University College London. The objective of this project is to develop high resolution and compact Gamma camera for clinical and research purposes, with overall spatial resolution of less than 1 mm. The most critical part of the gamma camera is the reconstruction of the interaction of events inside crystal based on the signals comes from the array of SSD detectors. There are different algorithms that could be used for the estimation of the coordinates of interaction of gamma rays inside the scintillator crystal, such as centroid method and Maximum likelihood. But each method has its own advantages and disadvantages in context of field of view, spatial resolution and processing time. This work is based on the investigation of the Artificial Neural Networks architectures especially Pattern Net and Feed Forward Neural Network to provide better estimation of X, Y and Z co-ordinates of gamma rays with good energy resolution at low energies. Neural Networks are very complicated from training point of view but requires little processing time.
Hicam è un progetto di ricerca che è in continua evoluzione da importanti istituti europei che sono Dipartimento di Ingegneria Elettronica del Politecnico di Milano, Max-Planck Institute, Università di Milano, e la University College di Londra. L'obiettivo di questo progetto è di sviluppare alta risoluzione e compatto Gamma camera per scopi clinici e di ricerca, con risoluzione spaziale complessiva inferiore a 1 mm. La parte più critica della gamma camera è la ricostruzione dell'interazione di eventi all'interno cristallo base ai segnali proviene dalla matrice di rivelatori SSD. Ci sono diversi algoritmi che potrebbero essere utilizzati per la stima delle coordinate di interazione dei raggi gamma all'interno del cristallo scintillatore, come metodo centroide e massima verosimiglianza. Ma ogni metodo ha i suoi vantaggi e svantaggi nel contesto del campo visivo, risoluzione spaziale e tempo di elaborazione. Questo lavoro si basa sullo studio delle architetture di reti neurali artificiali in particolare modello di rete e Feed Forward Neural Network per fornire una migliore stima di X, Y e Z coordinate di raggi gamma con una buona risoluzione energetica a basse energie. Le reti neurali sono molto complicati dal punto di vista formativo, ma richiede poco tempo di elaborazione.
Application of neural networks to anger camera imaging
JANJUA, ASHFAQ AHMAD
2012/2013
Abstract
Hicam is a research project that is under continuous development by leading European institutes which are Department of Electronic Engineering of Polytechnic institute of Milan, Max-Planck Institute, University of Milan, and University College London. The objective of this project is to develop high resolution and compact Gamma camera for clinical and research purposes, with overall spatial resolution of less than 1 mm. The most critical part of the gamma camera is the reconstruction of the interaction of events inside crystal based on the signals comes from the array of SSD detectors. There are different algorithms that could be used for the estimation of the coordinates of interaction of gamma rays inside the scintillator crystal, such as centroid method and Maximum likelihood. But each method has its own advantages and disadvantages in context of field of view, spatial resolution and processing time. This work is based on the investigation of the Artificial Neural Networks architectures especially Pattern Net and Feed Forward Neural Network to provide better estimation of X, Y and Z co-ordinates of gamma rays with good energy resolution at low energies. Neural Networks are very complicated from training point of view but requires little processing time.File | Dimensione | Formato | |
---|---|---|---|
Neural Networks.pdf
accessibile in internet per tutti
Descrizione: Text file
Dimensione
4.36 MB
Formato
Adobe PDF
|
4.36 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/84902