Raman Imaging allows the discrimination of several molecular compounds of biological samples, exploiting their unique vibrational spectrum as a source of contrast. It is of particular interest in cancer diagnosis, as it provides detailed maps with precise chemical specificity and high spatial resolution. In this thesis, a 76-channel broadband Stimu lated Raman imaging system is described, introducing the development of a new FPGA firmware which controls the system and implements 76 high-speed digital acquisition chains. The Raman signal is distributed across two sets of 38-channel photodiode ar rays. The processing is performed through a mixed, analogue/digital, acquisition chain. The system utilises the lock-in technique. The carrier frequency (1-10MHz) is chosen taking into account the noise spectral distribution of the employed laser. After the pho todiodes capture the signal, it is directed to ten custom ASICs, each managing eight differential channels. The signal undergoes preamplification, the signal and reference are gain-balanced, subsequently subtracted and finally demodulated via a passive mixer before being acquired by ten ADCs, one for each chip. Once digitised, the signals are transmitted to the FPGA through serial data interface (SDI). Two lines per ADC are em ployed for a total of twenty lines. The newly developed digital acquisition chain processes the data in parallel and applies appropriate filtering. Data processing can occur either through a single demodulation, using the analogue mixer, or through a more sophisticated technique known as 2-step frequency down conversion. This requires an additional de modulation stage implemented in the digital domain. The demodulating wave is a digital sine generated by a direct digital synthesizer (DDS). To eliminate unwanted harmonics, a low-pass filter and a discrete time integrator (DTI) are employed as anti-aliasing and as a comb filter respectively. The processed data is transmitted via USB 3.0 to a computer to be reconstructed by software into a hyperspectral image. In addition to the digital ac quisition chain, this thesis presents the design and simulation of auxiliary firmware blocks that enable the FPGA to control and configure the entire system, allowing for flexible operation. Finally, experimental results are provided, both in the absence and presence of an input optical signal incident on the photodiodes. The results favourably compare with theoretically predicted performance.
Il Raman Imaging permette di discriminare diverse molecole in campioni biologici utiliz zando il loro spettro in frequenza come firma. La tecnica è di particolare interesse in con testi clinici di diagnosi tumorale grazie alla possibilità di produrre mappe di distribuzione di specie chimiche con alta precisione spaziale. In questa tesi viene descritto un sistema di spettroscopia Raman a banda larga a 76 canali e introdotto il nuovo firmware FPGA re sponsabile del controllo del dispositivo e dell’implementazione di 76 catene di acquisizione digitale. Il segnale Raman a banda larga viene diffuso su due coppie di array di fotodiodi a catodo comune da 38 canali per poi essere processato da una catena di acquisizione mista analogico/digitale. Il sistema adopera la tecnica lock-in per trasportare l’informazione ad una frequenza che sia il più possibile vantaggiosa considerata la distribuzione spet trale del rumore del laser impiegato: 1-10MHz. Una volta acquisito il segnale ottico dai fotodiodi la catena di acquisizione procede per 10 ASIC custom che gestiscono 8 canali differenziali l’uno preamplificando, blinanciando e sottraendo reference da segnale per poi essere demodulati da un mixer passivo e acquisiti da 10 ADC. Una volta nel dominio digitale i segnali vengono trasmessi all’FPGA attraverso 2 connessioni SDI per ADC, per un totale di 20 linee. La catena di acquisizione digitale elabora i dati completamente in parallelo e è dotata di filtraggi ad hoc. La elaborazione dei dati può essere eseguita con una singloa demodulazione effettuata dal mixer analogico o con una tecnica più raffinata: una 2-step down conversion. Nel caso si attivi l’ulteriore demodulazione verrà eseguita nel dominio digitale impiegando una sinusoide digitale generata da un ”direct digital syn thesizer” (DDS). Per rimuovere armoniche indesiderate viene utilizzato un primo LPF che agisce da filtraggio anti-alias e un ”decrete time integrator” (DTI) utilizzato come filtro a pettine per rimuovere completamente anche le armoniche a bassa frequenza. I dati così elaborati vengono trasmessi tramite USB superspeed ad un computer che ricostruirà le informazioni iperspettrali in un immagine. La tesi inoltre illustra il progetto e la simu lazione di blocchi firmware ausiliari e accessori che permettono all’FPGA di controllare e configurare l’intero dispositivo e di renderlo flessibile nel suo utilizzo. Infine vengono mostrati i risultati dei test effettuati sia senza segnale ottico incidente sui fotodiodi che con segnale ottico presente e vengono confermate le previsioni teoriche.
FPGA firmware solution for 76 channels Raman imaging system
D'ONOFRIO, ALBERTO
2023/2024
Abstract
Raman Imaging allows the discrimination of several molecular compounds of biological samples, exploiting their unique vibrational spectrum as a source of contrast. It is of particular interest in cancer diagnosis, as it provides detailed maps with precise chemical specificity and high spatial resolution. In this thesis, a 76-channel broadband Stimu lated Raman imaging system is described, introducing the development of a new FPGA firmware which controls the system and implements 76 high-speed digital acquisition chains. The Raman signal is distributed across two sets of 38-channel photodiode ar rays. The processing is performed through a mixed, analogue/digital, acquisition chain. The system utilises the lock-in technique. The carrier frequency (1-10MHz) is chosen taking into account the noise spectral distribution of the employed laser. After the pho todiodes capture the signal, it is directed to ten custom ASICs, each managing eight differential channels. The signal undergoes preamplification, the signal and reference are gain-balanced, subsequently subtracted and finally demodulated via a passive mixer before being acquired by ten ADCs, one for each chip. Once digitised, the signals are transmitted to the FPGA through serial data interface (SDI). Two lines per ADC are em ployed for a total of twenty lines. The newly developed digital acquisition chain processes the data in parallel and applies appropriate filtering. Data processing can occur either through a single demodulation, using the analogue mixer, or through a more sophisticated technique known as 2-step frequency down conversion. This requires an additional de modulation stage implemented in the digital domain. The demodulating wave is a digital sine generated by a direct digital synthesizer (DDS). To eliminate unwanted harmonics, a low-pass filter and a discrete time integrator (DTI) are employed as anti-aliasing and as a comb filter respectively. The processed data is transmitted via USB 3.0 to a computer to be reconstructed by software into a hyperspectral image. In addition to the digital ac quisition chain, this thesis presents the design and simulation of auxiliary firmware blocks that enable the FPGA to control and configure the entire system, allowing for flexible operation. Finally, experimental results are provided, both in the absence and presence of an input optical signal incident on the photodiodes. The results favourably compare with theoretically predicted performance.File | Dimensione | Formato | |
---|---|---|---|
Alberto_DO_Executive_Summary_.pdf
accessibile in internet per tutti
Descrizione: Executive Summary Alberto D'Onofrio, versione finale
Dimensione
1.98 MB
Formato
Adobe PDF
|
1.98 MB | Adobe PDF | Visualizza/Apri |
AlbertoDO_Master_thesis.pdf
accessibile in internet per tutti
Descrizione: Alberto D'Onofrio, tesi magistrale versione finale
Dimensione
8.45 MB
Formato
Adobe PDF
|
8.45 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/10589/226936