Basara Akin I, Ozgul H, Simsek K, Altay C, Secil M, Balci P. Texture Analysis of Ultrasound Images to Differentiate Simple Fibroadenomas From Complex Fibroadenomas and Benign Phyllodes Tumors.
JOURNAL OF ULTRASOUND IN MEDICINE : OFFICIAL JOURNAL OF THE AMERICAN INSTITUTE OF ULTRASOUND IN MEDICINE 2020;
39:1993-2003. [PMID:
32329531 DOI:
10.1002/jum.15304]
[Citation(s) in RCA: 5] [Impact Index Per Article: 1.3] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Subscribe] [Scholar Register] [Received: 02/04/2020] [Revised: 03/15/2020] [Accepted: 03/26/2020] [Indexed: 06/11/2023]
Abstract
OBJECTIVES
American College of Radiology Breast Imaging Reporting and Data System (BI-RADS) category 4A lesions can be distinguished from BI-RADS 3 lesions with main ultrasound (US) findings such as a well-defined contour, round/oval shape, and parallel orientation with a homogeneous echo pattern. Breast Imaging Reporting and Data System 4A solid masses might be diagnosed as simple fibroadenomas (SFAs), complex fibroadenomas (CFAs), or benign phyllodes tumors (BPTs). Complex fibroadenomas have an increased risk of invasive cancer development than SFAs, and BPTs have a risk of borderline-malignant phyllodes tumor transformation; both of them are surgically treated, whereas follow-up procedures are applied in SFAs. It is essential to differentiate SFAs from CFAs and BPTs. Grayscale features of these lesions include a prominent overlap. Texture analyses in breast lesions have contributions in benign-malignant lesion differentiation. In this study, we aimed to use texture analysis of US images to differentiate these benign lesions.
METHODS
Grayscale US features of lesions (32 SFAs, 31 CFAs, and 32 BPTs) were classified according to the BI-RADS. Texture analysis of US images with LIFEx software (http://www.lifexsoft.org) was performed retrospectively. First- and second-order histogram parameters were evaluated.
RESULTS
In grayscale US, the shape, orientation, and posterior acoustic characteristics had statistical significance (P < .05). In the statistical analysis, skewness, kurtosis, excess kurtosis, gray-level co-occurrence matrix (GLCM)-energy, GLCM-entropy log 2, and GLCM-entropy log 10 revealed significant differences among all 3 groups (P < .05).
CONCLUSIONS
As grayscale US features show prominent intersections, and treatment options differ, correct diagnosis is essential in SFAs, CFAs, and BPTs. In this study, we concluded that texture analysis of US images can discriminate SFAs from CFAs and BPTs. Texture analyses of US images is a potential candidate diagnostic tool for these lesions, and accurate diagnoses will preclude patients from undergoing unnecessary biopsies.
Collapse