Andreozzi, Emilio (2021) Real-Time Quantum Noise Suppression In Very Low-Dose Fluoroscopy. [Tesi di dottorato]

[img] Text
Visibile a [TBR] Repository staff only

Download (5MB) | Request a copy

Download (8MB) | Preview
[error in script] [error in script]
Item Type: Tesi di dottorato
Resource language: English
Title: Real-Time Quantum Noise Suppression In Very Low-Dose Fluoroscopy
Date: 2 February 2021
Number of Pages: 127
Institution: Università degli Studi di Napoli Federico II
Department: Ingegneria Elettrica e delle Tecnologie dell'Informazione
Dottorato: Information technology and electrical engineering
Ciclo di dottorato: 33
Coordinatore del Corso di dottorato:
Cesarelli, MarioUNSPECIFIED
Date: 2 February 2021
Number of Pages: 127
Keywords: Fluoroscopy; low-dose X-ray imaging; quantum noise; noise characterization; noise suppression; real-time denoising; FPGA
Settori scientifico-disciplinari del MIUR: Area 09 - Ingegneria industriale e dell'informazione > ING-INF/06 - Bioingegneria elettronica e informatica
Date Deposited: 24 Mar 2021 16:32
Last Modified: 07 Jun 2023 10:27

Collection description

Fluoroscopy provides real-time X-ray screening of patient's organs and of various radiopaque objects, which make it an invaluable tool for many interventional procedures. For this reason, the number of fluoroscopy screenings has experienced a consistent growth in the last decades. However, this trend has raised many concerns about the increase in X-ray exposure, as even low-dose procedures turned out to be not as safe as they were considered, thus demanding a rigorous monitoring of the X-ray dose delivered to the patients and to the exposed medical staff. In this context, the use of very low-dose protocols would be extremely beneficial. Nonetheless, this would result in very noisy images, which need to be suitably denoised in real-time to support interventional procedures. Simple smoothing filters tend to produce blurring effects that undermines the visibility of object boundaries, which is essential for the human eye to understand the imaged scene. Therefore, some denoising strategies embed noise statistics-based criteria to improve their denoising performances. This dissertation focuses on the Noise Variance Conditioned Average (NVCA) algorithm, which takes advantage of the a priori knowledge of quantum noise statistics to perform noise reduction while preserving the edges and has already outperformed many state-of-the-art methods in the denoising of images corrupted by quantum noise, while also being suitable for real-time hardware implementation. Different issues are addressed that currently limit the actual use of very low-dose protocols in clinical practice, e.g. the evaluation of actual performances of denoising algorithms in very low-dose conditions, the optimization of tuning parameters to obtain the best denoising performances, the design of an index to properly measure the quality of X-ray images, and the assessment of an a priori noise characterization approach to account for time-varying noise statistics due to changes of X-ray tube settings. An improved NVCA algorithm is also presented, along with its real-time hardware implementation on a Field Programmable Gate Array (FPGA). The novel algorithm provides more efficient noise reduction performances also for low-contrast moving objects, thus relaxing the trade-off between noise reduction and edge preservation, while providing a further reduction of hardware complexity, which allows for low usage of logic resources also on small FPGA platforms. The results presented in this dissertation provide the means for future studies aimed at embedding the NVCA algorithm in commercial fluoroscopic devices to accomplish real-time denoising of very low-dose X-ray images, which would foster their actual use in clinical practice.


Downloads per month over past year

Actions (login required)

View Item View Item