Title: Ultrasound object detection using morphological region-based active contour: an application system
Authors: Anan Nugroho; Risanuri Hidayat; Hanung Adi Nugroho; Johan Debayle
Addresses: Department of Electrical and Information Engineering, Universitas Gadjah Mada, Indonesia; Department of Electrical Engineering, Universitas Negeri Semarang, Indonesia ' Department of Electrical and Information Engineering, Universitas Gadjah Mada, Indonesia ' Department of Electrical and Information Engineering, Universitas Gadjah Mada, Indonesia ' Mines Saint-Etienne, CNRS, UMR 5307 LGF, Centre SPIN, F-42023 Saint-Etienne, France
Abstract: Ultrasound (US) is intensively employed as a screening tool for suspicious objects such as breast lesions and thyroid nodules. Avoiding the subjectivity radiologists and to overcome high variability of US interpretations among them, technological innovations in computer-aided diagnosis or CAD are massively developed. Automation and accuracy in object detection and segmentation techniques as the core of CAD are becoming prestigious knowledge creations in the current Industrial Revolution 4.0. In this paper, an application system of morphological region-based active contour called MoRbAC is presented to automatically detect the suspicious US objects. The proposed MoRbAC application was validated by applying it to detect breast lesions and thyroid nodules on 20 real US images. Quantitative measurements based on overlapping area compared to referred ground truth achieve an average accuracy of up to 98.58 ± 1.15% with a quite short mean execution time, i.e., 2.38 ± 0.89 seconds. This promising performance concludes the effectiveness and efficiency of MoRbAC as an empowered method for CAD.
Keywords: ultrasound; computer-aided diagnosis; CAD; morphology; active contour; MoRbAC.
International Journal of Innovation and Learning, 2021 Vol.29 No.4, pp.412 - 430
Received: 08 Feb 2020
Accepted: 04 Jun 2020
Published online: 07 Jun 2021 *