View in EDS

Segmenting Benign and Malignant Solid Masses in Ultrasound Images Using Deep Learning.

Saved in:
Bibliographic Details
Title: Segmenting Benign and Malignant Solid Masses in Ultrasound Images Using Deep Learning.
Authors: Vázquez-Ramirez, Alexis1 (AUTHOR) alexxis.97@outlook.com, Mújica-Vargas, Dante1 (AUTHOR) dante.mv@cenidet.tecnm.mx, Luna-Álvarez, Antonio1 (AUTHOR) jesus.luna18ce@cenidet.edu.mx, de Jesus Rubio, José2 (AUTHOR) rubio.josedejesus@gmail.com
Source: Programming & Computer Software. Dec2025, Vol. 51 Issue 8, p867-878. 12p.
Subjects: Image segmentation, Deep learning, Benign tumors, Object recognition (Computer vision), Convolutional neural networks
Abstract: This study presents a novel approach for the segmentation of solid masses in ultrasound images, integrating preprocessing with detection and segmentation. Two convolutional neural networks, namely DeepLabv3+ and Darknet-53, are employed to identify regions of interest and to segment solid masses in ultrasound images. Furthermore, a preprocessing step utilizing the SRAD filter is applied prior to the image, whereby the edges of the masses are enhanced and the image noise is reduced. The strength of the methodology lies in its use of the YOLOv3 algorithm to delineate the region of interest and enhance the precision of the segmentation process. The results demonstrate a notable level of accuracy in the segmentation of both benign and malignant masses, with a Jaccard index of 90.10% and a Dice index of 94.71%. These outcomes exceed those achieved by existing techniques, substantiating the superiority of the proposed method for the analysis of ultrasound images. [ABSTRACT FROM AUTHOR]
Copyright of Programming & Computer Software is the property of Springer Nature and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)
Database: Engineering Source
Description
Abstract:This study presents a novel approach for the segmentation of solid masses in ultrasound images, integrating preprocessing with detection and segmentation. Two convolutional neural networks, namely DeepLabv3+ and Darknet-53, are employed to identify regions of interest and to segment solid masses in ultrasound images. Furthermore, a preprocessing step utilizing the SRAD filter is applied prior to the image, whereby the edges of the masses are enhanced and the image noise is reduced. The strength of the methodology lies in its use of the YOLOv3 algorithm to delineate the region of interest and enhance the precision of the segmentation process. The results demonstrate a notable level of accuracy in the segmentation of both benign and malignant masses, with a Jaccard index of 90.10% and a Dice index of 94.71%. These outcomes exceed those achieved by existing techniques, substantiating the superiority of the proposed method for the analysis of ultrasound images. [ABSTRACT FROM AUTHOR]
ISSN:03617688
DOI:10.1134/S0361768825700616