Proceedings Article | 17 February 2020
KEYWORDS: Signal to noise ratio, Ultrasonography, Transducers, Tissue optics, Photoacoustic imaging, Absorption, Ovarian cancer, Image quality, Ultrasonics, Medical image reconstruction, Photoacoustic tomography
A limited number of patients with ovarian cancer are diagnosed at its early stages. The reason is that current medical imaging techniques still present low sensitivity in the detection of this disease. Molecular imaging, such as photoacoustic (PA), can play an important role to improve ovarian cancer detection. For gynecologic applications, a transvaginal (endo-cavity) ultrasound probe is needed. In this study, the time reversal (TR) and delay and sum (DAS) image reconstruction methods were evaluated for PA signals acquired using a transvaginal transducer. PA signals were recorded from two different phantoms. Phantom A consisted of six hair strands immersed in a water tank positioned at different depths. Image reconstruction using TR after interpolating the acquired scan-lines from 128 to 256 (TR256) and to 512 lines (TR512) were also evaluated. Furthermore, simulations using the same configuration of phantom A were conducted and a similarity index (SSIM) was calculated for each reconstruction method. Phantom B was manufactured using the copolymer styrene-ethylene/butylene-styrene in mineral oil. The phantom had a cubic shape with a cavity to place the transducer and three irregular inclusions were added to the phantom. These inclusions were filled with black pigment to improve light absorption. The SSIM values obtained from simulations for DAS, TR, TR256 and TR512 were 0.52, 0.60, 0.68 and 0.72, respectively. Signal to noise ratio (SNR) and contrast to noise ratio (CNR), for phantom B images, were calculated to compare both methods. The SNR values for DAS, TR, TR256 and TR512 were 19.5 dB, 21.1 dB, 22.0 dB and 22.5 dB and the CNR values were 13.4 dB, 14.5 dB, 15.5 dB and 16.0 dB, respectively. Results showed better performance when the TR method was used, including SNR, CNR, and lateral and axial resolution. However, DAS was less time consuming compared to TR, maintaining a reasonable image quality.
