Limits...
Comparison of Deconvolution Filters for Photoacoustic Tomography.

Van de Sompel D, Sasportas LS, Jokerst JV, Gambhir SS - PLoS ONE (2016)

Bottom Line: It was found that the Tikhonov filter yielded the most accurate balance of lower and higher frequency content (as measured by comparing the spectra of deconvolved impulse response signals to the ideal flat frequency spectrum), achieved a competitive image resolution and contrast-to-noise ratio, and yielded the greatest robustness to noise.In addition, its robustness to noise was poorer than that of the Tikhonov filter.Overall, the Tikhonov filter was deemed to produce the most desirable image reconstructions.

View Article: PubMed Central - PubMed

Affiliation: Molecular Imaging Program at Stanford (MIPS), Stanford University School of Medicine, Stanford University, Stanford, CA 94305, United States of America.

ABSTRACT
In this work, we compare the merits of three temporal data deconvolution methods for use in the filtered backprojection algorithm for photoacoustic tomography (PAT). We evaluate the standard Fourier division technique, the Wiener deconvolution filter, and a Tikhonov L-2 norm regularized matrix inversion method. Our experiments were carried out on subjects of various appearances, namely a pencil lead, two man-made phantoms, an in vivo subcutaneous mouse tumor model, and a perfused and excised mouse brain. All subjects were scanned using an imaging system with a rotatable hemispherical bowl, into which 128 ultrasound transducer elements were embedded in a spiral pattern. We characterized the frequency response of each deconvolution method, compared the final image quality achieved by each deconvolution technique, and evaluated each method's robustness to noise. The frequency response was quantified by measuring the accuracy with which each filter recovered the ideal flat frequency spectrum of an experimentally measured impulse response. Image quality under the various scenarios was quantified by computing noise versus resolution curves for a point source phantom, as well as the full width at half maximum (FWHM) and contrast-to-noise ratio (CNR) of selected image features such as dots and linear structures in additional imaging subjects. It was found that the Tikhonov filter yielded the most accurate balance of lower and higher frequency content (as measured by comparing the spectra of deconvolved impulse response signals to the ideal flat frequency spectrum), achieved a competitive image resolution and contrast-to-noise ratio, and yielded the greatest robustness to noise. While the Wiener filter achieved a similar image resolution, it tended to underrepresent the lower frequency content of the deconvolved signals, and hence of the reconstructed images after backprojection. In addition, its robustness to noise was poorer than that of the Tikhonov filter. The performance of the Fourier filter was found to be the poorest of all three methods, based on the reconstructed images' lowest resolution (blurriest appearance), generally lowest contrast-to-noise ratio, and lowest robustness to noise. Overall, the Tikhonov filter was deemed to produce the most desirable image reconstructions.

Show MeSH

Related in: MedlinePlus

Nexus 128 scanner.(a-c) are reproduced with permission from http://www.endrainc.com. The green box in (d) delineates the spatial support of a representative reconstructed field of view. The dimensions of this reconstructed volume are typically set to 2 × 2 × 2 cm, and the isotropic reconstruction resolution typically to 0.2 mm. The blue dots show the locations of the 128 transducers in the bowl.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC4816281&req=5

pone.0152597.g002: Nexus 128 scanner.(a-c) are reproduced with permission from http://www.endrainc.com. The green box in (d) delineates the spatial support of a representative reconstructed field of view. The dimensions of this reconstructed volume are typically set to 2 × 2 × 2 cm, and the isotropic reconstruction resolution typically to 0.2 mm. The blue dots show the locations of the 128 transducers in the bowl.

Mentions: The photoacoustic scans for this study were acquired using the Nexus 128 scanner, manufactured by Endra Life Sciences (Ann Arbor, MI, USA). Fig 2 illustrates the system. The detection geometry consists of 128 unfocused transducers arranged on a hemispherical bowl. The radius of the bowl is 101.3 mm. A plastic tray is placed on top of the bowl for animal positioning. The tray is covered by a lid that must be closed while scanning (an interlock prevents scanning while the lid is open) as a protective measure to rule out the possibility of eye damage from the laser to the user. A webcam inside the lid allows users to monitor the mouse during a scan. The bowl is filled with water to provide acoustic coupling. A water heating system maintains the water temperature at 38°C between scans to enable small animal imaging. The readings of the water heating system’s thermometer can be accessed in a log file. The water heating pump is switched off during scans to eliminate the risk of bubble formation. The blue dots in Fig 2(d) indicate the transducer positions, and the green box indicates the spatial support of a standard reconstructed volume (200 × 200 × 200 with an isotropic resolution of 0.1 mm centered on the focal center of the hemispherical bowl). The maximum field of view supported by the system is currently a sphere of radius 3.8 cm, again centered on the focal center of the hemispherical bowl. Each transducer has a circular detection surface with a diameter of 3 mm and records at a sampling rate of 20 MHz. The center frequency of the transducers is 5 MHz with a bandwidth of approximately 70%. Note that if the bandwidth were lower, the transducers would detect fewer sound frequencies in the lower and higher ranges. In the image space, this would translate into fewer low and high spatial frequencies, i.e. lower contrast for larger masses and blurrier edges for small details, respectively. The scanner impulse response pd0(t) was measured by the scanner’s manufacturer by imaging a hair-like mylar thread, approximately 75 μm in diameter, which was placed directly in the water at the isocenter of the detector bowl. The signal pd0(t) was measured without the standard diverging lens of the scanner to obtain a higher than normal laser power (43 mJ per pulse) and hence to ensure a high signal-to-noise measurement. The scanner impulse response pd0(t) was assumed equal for all transducers, and is shown in Fig 3, along with its frequency content. To increase geometric sampling of the detection surface, the bowl is rotated in steps through multiple positions during a single scan. Usually, approximately 120 to 180 bowl positions, distributed uniformly over 360 degrees, suffice to obtain a high quality reconstructed image. Each such bowl position will be referred to as a ‘view’ in the remainder of this paper. Due to the small angular steps and light weight of the bowl, no significant shaking of the transducers nor bubble formation was observed as the stepper motor rotated the bowl.


Comparison of Deconvolution Filters for Photoacoustic Tomography.

Van de Sompel D, Sasportas LS, Jokerst JV, Gambhir SS - PLoS ONE (2016)

Nexus 128 scanner.(a-c) are reproduced with permission from http://www.endrainc.com. The green box in (d) delineates the spatial support of a representative reconstructed field of view. The dimensions of this reconstructed volume are typically set to 2 × 2 × 2 cm, and the isotropic reconstruction resolution typically to 0.2 mm. The blue dots show the locations of the 128 transducers in the bowl.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC4816281&req=5

pone.0152597.g002: Nexus 128 scanner.(a-c) are reproduced with permission from http://www.endrainc.com. The green box in (d) delineates the spatial support of a representative reconstructed field of view. The dimensions of this reconstructed volume are typically set to 2 × 2 × 2 cm, and the isotropic reconstruction resolution typically to 0.2 mm. The blue dots show the locations of the 128 transducers in the bowl.
Mentions: The photoacoustic scans for this study were acquired using the Nexus 128 scanner, manufactured by Endra Life Sciences (Ann Arbor, MI, USA). Fig 2 illustrates the system. The detection geometry consists of 128 unfocused transducers arranged on a hemispherical bowl. The radius of the bowl is 101.3 mm. A plastic tray is placed on top of the bowl for animal positioning. The tray is covered by a lid that must be closed while scanning (an interlock prevents scanning while the lid is open) as a protective measure to rule out the possibility of eye damage from the laser to the user. A webcam inside the lid allows users to monitor the mouse during a scan. The bowl is filled with water to provide acoustic coupling. A water heating system maintains the water temperature at 38°C between scans to enable small animal imaging. The readings of the water heating system’s thermometer can be accessed in a log file. The water heating pump is switched off during scans to eliminate the risk of bubble formation. The blue dots in Fig 2(d) indicate the transducer positions, and the green box indicates the spatial support of a standard reconstructed volume (200 × 200 × 200 with an isotropic resolution of 0.1 mm centered on the focal center of the hemispherical bowl). The maximum field of view supported by the system is currently a sphere of radius 3.8 cm, again centered on the focal center of the hemispherical bowl. Each transducer has a circular detection surface with a diameter of 3 mm and records at a sampling rate of 20 MHz. The center frequency of the transducers is 5 MHz with a bandwidth of approximately 70%. Note that if the bandwidth were lower, the transducers would detect fewer sound frequencies in the lower and higher ranges. In the image space, this would translate into fewer low and high spatial frequencies, i.e. lower contrast for larger masses and blurrier edges for small details, respectively. The scanner impulse response pd0(t) was measured by the scanner’s manufacturer by imaging a hair-like mylar thread, approximately 75 μm in diameter, which was placed directly in the water at the isocenter of the detector bowl. The signal pd0(t) was measured without the standard diverging lens of the scanner to obtain a higher than normal laser power (43 mJ per pulse) and hence to ensure a high signal-to-noise measurement. The scanner impulse response pd0(t) was assumed equal for all transducers, and is shown in Fig 3, along with its frequency content. To increase geometric sampling of the detection surface, the bowl is rotated in steps through multiple positions during a single scan. Usually, approximately 120 to 180 bowl positions, distributed uniformly over 360 degrees, suffice to obtain a high quality reconstructed image. Each such bowl position will be referred to as a ‘view’ in the remainder of this paper. Due to the small angular steps and light weight of the bowl, no significant shaking of the transducers nor bubble formation was observed as the stepper motor rotated the bowl.

Bottom Line: It was found that the Tikhonov filter yielded the most accurate balance of lower and higher frequency content (as measured by comparing the spectra of deconvolved impulse response signals to the ideal flat frequency spectrum), achieved a competitive image resolution and contrast-to-noise ratio, and yielded the greatest robustness to noise.In addition, its robustness to noise was poorer than that of the Tikhonov filter.Overall, the Tikhonov filter was deemed to produce the most desirable image reconstructions.

View Article: PubMed Central - PubMed

Affiliation: Molecular Imaging Program at Stanford (MIPS), Stanford University School of Medicine, Stanford University, Stanford, CA 94305, United States of America.

ABSTRACT
In this work, we compare the merits of three temporal data deconvolution methods for use in the filtered backprojection algorithm for photoacoustic tomography (PAT). We evaluate the standard Fourier division technique, the Wiener deconvolution filter, and a Tikhonov L-2 norm regularized matrix inversion method. Our experiments were carried out on subjects of various appearances, namely a pencil lead, two man-made phantoms, an in vivo subcutaneous mouse tumor model, and a perfused and excised mouse brain. All subjects were scanned using an imaging system with a rotatable hemispherical bowl, into which 128 ultrasound transducer elements were embedded in a spiral pattern. We characterized the frequency response of each deconvolution method, compared the final image quality achieved by each deconvolution technique, and evaluated each method's robustness to noise. The frequency response was quantified by measuring the accuracy with which each filter recovered the ideal flat frequency spectrum of an experimentally measured impulse response. Image quality under the various scenarios was quantified by computing noise versus resolution curves for a point source phantom, as well as the full width at half maximum (FWHM) and contrast-to-noise ratio (CNR) of selected image features such as dots and linear structures in additional imaging subjects. It was found that the Tikhonov filter yielded the most accurate balance of lower and higher frequency content (as measured by comparing the spectra of deconvolved impulse response signals to the ideal flat frequency spectrum), achieved a competitive image resolution and contrast-to-noise ratio, and yielded the greatest robustness to noise. While the Wiener filter achieved a similar image resolution, it tended to underrepresent the lower frequency content of the deconvolved signals, and hence of the reconstructed images after backprojection. In addition, its robustness to noise was poorer than that of the Tikhonov filter. The performance of the Fourier filter was found to be the poorest of all three methods, based on the reconstructed images' lowest resolution (blurriest appearance), generally lowest contrast-to-noise ratio, and lowest robustness to noise. Overall, the Tikhonov filter was deemed to produce the most desirable image reconstructions.

Show MeSH
Related in: MedlinePlus