Automatic And Semi-Automatic Bladder Volume Detection In Ultrasound Images


Kavuncubaşı U., Tek G., Acar K., Ertosun B., Akşahin M. F.

ICAIAME 21, Antalya, Turkey, 1 - 03 October 2021, pp.70

  • Publication Type: Conference Paper / Summary Text
  • City: Antalya
  • Country: Turkey
  • Page Numbers: pp.70
  • Gazi University Affiliated: Yes

Abstract

The bladder is the organ that stores urine after the blood is filtered in the kidneys, in the anatomy of mammals. The bladder is usually visualized by ultrasonography. The bladder volume should be determined for diagnostic and therapeutic purposes with the use of these ultrasonographic images. For this purpose, two different algorithms are proposed in this study for the detection of bladder volume using ultrasound images. Hence, two algorithms, semi-automatic and fully automatic, were developed with the integration method as an alternative to the traditional ellipsoid method. In the semi-automatic algorithm, the bladder volume was calculated by the integration method after the user determined the bladder region visually. In addition, in order to eliminate user subjectivity in the study, the bladder area was automatically determined regardless of user. Therefore, the bladder region was determined fully automatically from ultrasound images using 2D Discrete Wavelet Transform, Canny Edge Detection Method, and morphological operations. Bladder volume was calculated with the same integration method as the semi-automatic algorithm using the determined region. The Prolate Ellipsoid Method was accepted as a reference. Then, according to these results, semi-automatic and fully automatic algorithm Mean Absolute Percentage Error results were obtained and compared. It has been observed that different results are achieved for different users in the semi-automatic algorithm. This result showed the fully automatic algorithm is more objective and effective. Consequently, a fully automatic algorithm that is easy to implement, gives fast results, and has high accuracy has been composed.