A very important issue in modern Nuclear Medicine is to reduce the e ect of motion artifacts on reconstructed data: the very high resolutions achieved can be severely lowered by patient's movements, a ecting the precision of the diagnosis. Therefore, in the present work, a method for motion correction has been presented for the Siemens High Resolution Research Tomograph (HRRT), installed at the Neuroscience department of Karolinska Institutet, in Stockholm. As a rst step, we have wanted to prove the e ective need for motion correction on such a high resolution tomograph: an experiment to assess camera spatial resolution has been conducted and the results have been compared with patients' typical movements. A capillar lled with 18F has been positioned in the eld of view to observe how activity point sources are reconstructed in the nal images; the Full Width at Half Maximum (FWHM) of the retrieved Point Spread Functions ranged from 1.5 mm to 3.2 mm, showing the necessity for correction of motion artifacts. An image-driven approach has been developed and implemented following the Multiple-Acquisition-Frame strategy and using NDI Polaris Vicra to obtain motion data, needed for the algorithm. Corrected frames show promising results both in translation (77% of the frames is below the set threshold) and in rotation correction (only 2 frames are above the set threshold). However, the future possibility provided by external trackers to perform online corrections, once raw data can be accessed, gives our algorithm the advantage to manipulate the single events and, subsequently, to obtain accuracies unreachable for post-reconstruction corrections, like the cross-correlation method employed so far, in the department. The role of the manuscript is then to uphold the need to motion correct HRRT studies and to show how o ine corrections, using external trackers, can provide good performances, despite the impossibilty to access the single events; further developments in the eld will therefore help in obtaining the best resolutions and accuracies achievable for high precision investigations.
Uno dei problemi pi u rilevanti nella diagnostica per immagine e rappresentato dalla necessit a di minimizzare l'interferenza esercitata sui dati nali, e quindi sulla diagnosi, dai movimenti dei pazienti sottoposti ad indagine. Ci siamo quindi proposti con questo studio di sviluppare un metodo di motion correction per la High Resolution Research Tomograph (HRRT), impiegata per l'imaging cerebrale presso il Dipartimento di Neuroscienze del Karolinska Institutet, a Stoccolma. In prima istanza si e indagato sulla risoluzione spaziale dello scanner, posizionando nel centro del Field of View della camera un capillare di dimensioni ridotte, riempito con 18F, al ne di studiare come il sistema fosse in grado di riprodurre sorgenti puntiformi di attivit a. A fronte di valori oscillanti tra i 1.5 mm e i 3.2 mm, in termini di Full Width at Half Maximum (FWHM) dell'impulso riprodotto, si e confermata la necessit a assoluta di ricorrere alla motion correction. Si e quindi sviluppato un algoritmo di motion correction su immagini ricostruite, optando per l'adozione di una strategia di correzione Multiple-Acquisition-Frame e di un dispositivo esterno capace di registrare posizione e orientamento dell'oggetto: il Polaris Vicra. Il metodo di motion correction implementato ha mostrato risultati promettenti, nella correzione sia dei movimenti di traslazione (77% dei frame corretti), sia dei movimenti di rotazione (solo 2 dei frame corretti hanno valori oltre la soglia). Inoltre, la ulteriore possibilit a futura di manipolare la singola coincidenza consentir a al nostro algoritmo di e ettuare una correzione su singolo evento e sfruttare cos scenari che saranno preclusi ai metodi di correzione su immagini gi a ricostruite, come il metodo a cross-correlazione no ad ora utilizzato. Con questo studio si e quindi evidenziata la necessit a di sviluppare una metodologia solida di motion correction e si e dimostrato come anche metodi di correzione o ine, che si avvalgono di dispositivi esterni, siano in grado di fornire buone prestazioni, nonostante operino su dati gi a ricostruiti. Lo sviluppo in questo settore e dunque fondamentale per usufruire di indagini diagnostiche dall'alta risoluzione e accuratezza.
Evaluation and implementation of a motion correction method for high resolution tomographs
FERRANTE, MATTEO
2012/2013
Abstract
A very important issue in modern Nuclear Medicine is to reduce the e ect of motion artifacts on reconstructed data: the very high resolutions achieved can be severely lowered by patient's movements, a ecting the precision of the diagnosis. Therefore, in the present work, a method for motion correction has been presented for the Siemens High Resolution Research Tomograph (HRRT), installed at the Neuroscience department of Karolinska Institutet, in Stockholm. As a rst step, we have wanted to prove the e ective need for motion correction on such a high resolution tomograph: an experiment to assess camera spatial resolution has been conducted and the results have been compared with patients' typical movements. A capillar lled with 18F has been positioned in the eld of view to observe how activity point sources are reconstructed in the nal images; the Full Width at Half Maximum (FWHM) of the retrieved Point Spread Functions ranged from 1.5 mm to 3.2 mm, showing the necessity for correction of motion artifacts. An image-driven approach has been developed and implemented following the Multiple-Acquisition-Frame strategy and using NDI Polaris Vicra to obtain motion data, needed for the algorithm. Corrected frames show promising results both in translation (77% of the frames is below the set threshold) and in rotation correction (only 2 frames are above the set threshold). However, the future possibility provided by external trackers to perform online corrections, once raw data can be accessed, gives our algorithm the advantage to manipulate the single events and, subsequently, to obtain accuracies unreachable for post-reconstruction corrections, like the cross-correlation method employed so far, in the department. The role of the manuscript is then to uphold the need to motion correct HRRT studies and to show how o ine corrections, using external trackers, can provide good performances, despite the impossibilty to access the single events; further developments in the eld will therefore help in obtaining the best resolutions and accuracies achievable for high precision investigations.| File | Dimensione | Formato | |
|---|---|---|---|
|
2013_10_Ferrante.pdf
accessibile in internet per tutti
Descrizione: Testo della tesi
Dimensione
10.31 MB
Formato
Adobe PDF
|
10.31 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/10589/84907