Abstract
ObjectiveThis study was conducted in order to investigate the feasibility of using radiomics analysis (RA) with machine learning algorithms based on breast magnetic resonance (MR) images for discriminating malignant from benign MR-detected additional lesions in patients with primary breast cancer.Materials and MethodsOne hundred seventy-four MR-detected additional lesions (benign, n = 86; malignancy, n = 88) from 158 patients with ipsilateral primary breast cancer from a tertiary medical center were included in this retrospective study. The entire data were randomly split to training (80%) and independent test sets (20%). In addition, 25 patients (benign, n = 21; malignancy, n = 15) from another tertiary medical center were included for the external test. Radiomics features that were extracted from three regions-of-interest (ROIs; intratumor, peritumor, combined) using fat-saturated T1-weighted images obtained by subtracting pre- from postcontrast images (SUB) and T2-weighted image (T2) were utilized to train the support vector machine for the binary classification. A decision tree method was utilized to build a classifier model using clinical imaging interpretation (CII) features assessed by radiologists. Area under the receiver operating characteristic curve (AUROC), accuracy, sensitivity, and specificity were used to compare the diagnostic performance.ResultsThe RA models trained using radiomics features from the intratumor-ROI showed comparable performance to the CII model (accuracy, AUROC: 73.3%, 69.6% for the SUB RA model; 70.0%, 75.1% for the T2 RA model; 73.3%, 72.0% for the CII model). The diagnostic performance increased when the radiomics and CII features were combined to build a fusion model. The fusion model that combines the CII features and radiomics features from multiparametric MRI data demonstrated the highest performance with an accuracy of 86.7% and an AUROC of 91.1%. The external test showed a similar pattern where the fusion models demonstrated higher levels of performance compared with the RA- or CII-only models. The accuracy and AUROC of the SUB+T2 RA+CII model in the external test were 80.6% and 91.4%, respectively.ConclusionOur study demonstrated the feasibility of using RA with machine learning approach based on multiparametric MRI for quantitatively characterizing MR-detected additional lesions. The fusion model demonstrated an improved diagnostic performance over the models trained with either RA or CII alone.
Highlights
Breast magnetic resonance imaging (MRI) is widely used for the preoperative evaluation of the extent of malignancy and the detection of any ipsilateral or contralateral additional lesions, especially for candidates for breast-conserving therapy [1]
Among a total of 6,558 breast MRI examinations that were performed at Chonnam National University Hwasun Hospital between January 2012 and July 2020, the MRI exams that met the following criteria were included: 1) initial breast Magnetic resonance (MR) images for preoperative evaluation of pathologically proven primary breast cancer; 2) interpretation reports containing the following keywords applied to ipsilateral breast lesion, “BI-RADS (Breast Imaging Reporting and Data System) 0,” “BI-RADS 4”, “BI-RADS 5”, “targeted”, “second-look”, or “US”; and 3) no history of neoadjuvant chemotherapy (NAC), excision, or vacuum-assisted biopsy
The following exclusion criteria were used: 1) additional lesions that were described as “daughter” or “satellite” lesions; 2) lesions that were interpreted as “non-mass enhancement” or “focus” according to the BI-RADS MR lexicon from the American College of Radiology [19]; 3) lesions with size less than 7 mm, which were too small to extract sufficient radiomics features; 4) patients without a second-look US; 5) lesions that were sonographically occult or lacking a MR– US correlation; 6) lesions whose pathologic confirmation did not come from a separate excision; 7) lesions that were confirmed as a “borderline-risk” lesion, such as atypical ductal hyperplasia or lobular carcinoma in situ to avoid a doubtful ground truth of diagnoses; and 8) patients who were lost to follow-up
Summary
Breast magnetic resonance imaging (MRI) is widely used for the preoperative evaluation of the extent of malignancy and the detection of any ipsilateral or contralateral additional lesions, especially for candidates for breast-conserving therapy [1]. Magnetic resonance (MR)-detected additional lesions are the lesions found on preoperative MRI, which have not been identified on prior mammogram or ultrasound. A second-look ultrasound (US) or a targeted US is generally performed to further evaluate the additional lesion [2, 3]; US is usually unspecific for malignancy detection and the correlation rates between MR and US have been reported to be variable, ranging from 23% to 89%, depending on several factors, such as the performance of an individual radiologist or patient-specific differences [4]. No defined protocol exists for the further workup of MR-detected additional lesion, whose clinical protocol usually relies upon the discretion of radiologists [2, 5]
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
