Detection and classification of breast lesions using multiple information on contrast-enhanced mammography by a multiprocess deep-learning system: A multicenter study

Objective: Accurate detection and classification of breast lesions in early stage is crucial to timely formulate effective treatments for patients. We aim to develop a fully automatic system to detect and classify breast lesions using multiple contrast-enhanced mammography(CEM) images.Methods: In this study, a total of 1,903 females who underwent CEM examination from three hospitals were enrolled as the training set, internal testing set, pooled external testing set and prospective testing set. Here we developed a CEM-based multiprocess detection and classification system(MDCS) to perform the task of detection and classification of breast lesions. In this system, we introduced an innovative auxiliary feature fusion(AFF)algorithm that could intelligently incorporates multiple types of information from CEM images. The average freeresponse receiver operating characteristic score(AFROC-Score) was presented to validate system’s detection performance, and the performance of classification was evaluated by area under the receiver operating characteristic curve(AUC). Furthermore, we assessed the diagnostic value of MDCS through visual analysis of disputed cases,comparing its performance and efficiency with that of radiologists and exploring whether it could augment radiologists’ performance.Results: On the pooled external and prospective testing sets, MDCS always maintained a high standalone performance, with AFROC-Scores of 0.953 and 0.963 for detection task, and AUCs for classification were 0.909[95% confidence interval(95% CI): 0.822-0.996] and 0.912(95% CI: 0.840-0.985), respectively. It also achieved higher sensitivity than all senior radiologists and higher specificity than all junior radiologists on pooled external and prospective testing sets. Moreover, MDCS performed superior diagnostic efficiency with an average reading time of 5 seconds, compared to the radiologists’ average reading time of 3.2 min. The average performance of all radiologists was also improved to varying degrees with MDCS assistance.Conclusions: MDCS demonstrated excellent performance in the detection and classification of breast lesions,and greatly enhanced the overall performance of radiologists.

Application Value of Contrast-Enhanced Ultrasound in the Diagnosis of Breast Lesions

Objective:To explore the value and effect of contrast-enhanced ultrasound in the diagnosis of breast lesions.Methods:Seventy-two patients with breast lesions in Shaanxi Provincial People’s Hospital from June 2020 to December 2021 were selected as the research subjects.All 72 patients met the diagnostic criteria of breast lesions.Two patients with incomplete clinical data were excluded;hence,there were 70 patients remaining.The diagnostic results of the two examination methods and the diagnostic value of the joint examination for breast lesions were analyzed and compared.Results:The results of benign,malignant,missed,and misdiagnosed breast lesions by contrast-enhanced ultrasound were 31,32,6,and 1 cases,respectively,accounting for 44.29%,45.71%,8.57%,and 1.43%,respectively.The results of benign,malignant,missed,and misdiagnosed breast lesions by ultrasound automatic volume imaging were 21,24,17,and 8 cases,respectively,accounting for 30.00%,34.28%,24.29%,and 11.43%,respectively.There were statistical differences between the two groups for missed diagnosis and misdiagnosis,but there was no significant difference between the two groups for benign and malignant lesions.The accuracy,sensitivity,and specificity of contrast-enhanced ultrasound were 87.69%,83.62%,and 83.45%,respectively;the accuracy,sensitivity,and specificity of ultrasound automatic volume imaging were 71.39%,68.99%,and 74.69%,respectively;the accuracy,sensitivity,and specificity of contrast-enhanced ultrasound combined with ultrasound automatic volume imaging were 96.29%,92.68%,and 91.78%,respectively.Conclusion:Contrast-enhanced ultrasonography has a high clinical application value and a low inspection error rate in the diagnosis of breast lesions.It merits clinical advancement since it helps doctors diagnose and treat breast lesions more effectively.

Differentiation between Benign and Malignant Breast Lesions Using ADC on Diffusion-Weighted Imaging at 3.0 T

Aim: To investigate the diagnostic efficiency of apparent diffusion coefficient value (ADC) in differentiating benign from malignant breast lesions at 3.0 T diffusion-weighted imaging with different pair of b value. Methods: Total 110 patients with 107 lesions (44 benign and 63 malignant) were selected for our study with five different b-values 0, 400, 800, 1200 and 1600 s/mm2. ADC values were calculated using different pairs of b values. The cut-off ADC values and diagnostic efficiency were evaluated by receiver operating characteristic analysis. Comparison of Mean ADC value for breast lesions was determined by using independent sample t test. ROC curves were used for diagnostic efficiency of ADC using different pairs of b values. Results: With increase of b value, mean ADC value decreases. The mean ADC values for benign were 1.73 × 10-3 mm2/s for b 0 and 400, 1.57 × 10-3 mm2/s for b 0 and 800, 1.43 × 10-3 mm2/s for b 0 and 1200 and 1.30 × 10-3 mm2/s for b 0 and 1600 s/mm2. The mean ADC values for the malignant breast lesion were 1.21 × 10-3 mm2/s for b 0 and 400, 1.06 × 10-3 mm2/s for b 0 and 800, 0.94 × 10-3 mm2/s for b 0 and 1200 and 0.86 × 10-3 mm2/s for b 0 and 1600 s/mm2. ADC diagnostic efficiency for benign and malignant lesion for all the pair of b value combination was significant (p > 0.05). The sensitivity, specificity, PPV, NPV and accuracy were 80.95%, 90.9%, 92.72%, 76.92%, 85.04% for b 0 and 400;84.12%, 90.9%, 92.98%, 80%, 86.91% for b 0 and 800;84.12%, 90.9%, 92.98%, 80%, 86.91% for b 0 and 1200;84.12%, 90.9%, 92.98%, 80%, 86.91% for b 0 and 1600 s/mm2 respectively. Conclusion: DWI is effective in differentiating benign and malignant breast lesion at 3.0 Tesla using ADC with higher b value combination.

Monoexponential, biexponential and stretched-exponential models based diffusion weighted imaging: a comparative study in the differential diagnosis of benign and malignant breast lesions

Objective To investgate the value of various parameters obtained from monoexponential,biexponential,and stretched exponential diffusion-weighted imaging models in the differential diagnosis of breast lesions.Methods A retrospective study was performed in 54 patients with pathologically confirmed malignant tumors(n=30),benign lesions(n=34)and normal fibroglandular

Diagnosis and excision of breast multi-focal lesions by ultrasound-guided vacuum-assisted biopsy:a comparative evaluation

Objective To compare and evaluate the efficacy of diagnosis and excision for appropriately selected breast multi-focal lesions and solitary lesion by ultrasound-guided vacuum-assisted biopsy(UGVAB).Methods Among 392 appropriately selected patients,187 patients with multi-focal lesions and 205 patients with solitary lesion were treated by the 8-gauge UGVAB from May 2007 to June 2009.All lesions were removed as completely as possible.The patients with benign pathology underwent physical and ultrasound examinations at one week and 6 months after procedure.Results During the procedure,only three patients had vasovagal syncope and twenty others complained of other intraoperative discomfort.An accurate pathological diagnosis was obtained in all lesions.There was no apparent false-negative result among the 696 lesions with benign pathology at a follow-up of 6 months after procedure.The rates of malignant or premalignant pathology,postoperative complications and residual lesions in patients with multi-focal lesions were higher than those in patients with solitary lesion.If each lesion was considered as a subject of study,there was no significant difference between the two groups.Conclusion UGVAB is an effective method for diagnosis and excision of appropriately selected breast multi-focal lesions and can be used routinely.

A STUDY OF LECTINS IN HUMAN BREAST CANCER AND PRECANCEROUS LESIONS

This paper report the binding pattern of 12 different kinds of lectin in human breast lesions. Of the 12 kinds of lectins, WGA showed the highest binding activity to the cells of the breast tissue derived; the binding of BSL, SBA and DBA were localized to membrane or cytoplasm of cancer cells and to the lumina membrane border of the normal and benign lesions; PNA receptor is related with the differentiation of the breast; cancer; we didn't find any relationship between the lectin receptor and the tendency of metastasis of breast cancer.

Diagnostic performance of contrast-enhanced ultrasound and enhanced magnetic resonance for breast nodules

In the current study, we sought to evaluate the diagnostic efficacies of conventional ultrasound（US）, contrastenhanced US（CEUS）, combined US and CEUS and magnetic resonance imaging（MRI） in detecting focal solid breast lesions. Totally 117 patients with 120 BI-RADS category 4A-5 breast lesions were evaluated by conventional US and CEUS, and MRI, respectively. SonoVue was used as contrast agent in CEUS and injected as an intravenous bolus; nodule scan was performed 4 minutes after bolus injection. A specific sonographic quantification software was used to obtain color-coded maps of perfusion parameters for the investigated lesion, namely the time-intensity curve.The pattern of contrast enhancement and related indexes regarding the time-intensity curve were used to describe the lesions, comparatively with pathological results. Histopathologic examination revealed 46 benign and 74 malignant lesions. Sensitivity, specificity, and accuracy of US in detecting malignant breast lesions were 90.14%, 95.92%, and 92.52%, respectively. Meanwhile, CE-MRI showed sensitivity, specificity, and accuracy of 88.73%, 95.92%, and91.67%, respectively. The area under the ROC curve for combined US and CEUS in discriminating benign from malignant breast lesions was 0.936, while that of MRI was 0.923, with no significant difference between them, as well as among groups. The time-intensity curve of malignant hypervascular fibroadenoma and papillary lesions mostly showed a fast-in/fast-out pattern, with no good correlation between them（kappa 〈0.20）. In conclusion, the combined use of conventional US and CEUS displays good agreement with MRI in differentiating benign from malignant breast lesions.

Breast Core-Needle Biopsy in a Large Tertiary Oncologic Centre—1-Year Experience after the Introduction of the Method

Ultrasound (US)-guided core-needle biopsy (CNB) is currently the procedure of choice for work-up of suspicious breast lesion. It is mainly used for evaluation of suspicious breast lesions categorized as BI-RADS 4 and 5 (Breast Imaging-Reporting and Data System). The conducted study included 56 female patients with detected suspicious breast leasions, and they underwent US-guided CNB during 1-year period with the aim to investigate the value of US-guided CNB of the breast in a tertiary-level large-volume oncological centre setting with respect of indications, technical adequacy and safety. 2 patients who entered the study were previously diagnosed as BIRADS 2, 3 patients as BIRADS 3, 18 patients as BIRADS 4 and 33 patients as BIRADS 5. In 14 patients with BC (breast cancer), both FNA (fine-needle aspiration) and CNB were performed, and the malignancy was accurately diagnosed by cytology in 9 patients, confirmed by subsequent CNB in all of them. ADH (atypical ductal hyperplasia) was initialy diagnosed by FNA in 5 patients, and in 2 of them, BC was initialy missed by FNA, but deteced by CNB. As it is known, the cytology has lower sensitivity for detection of BC than hystology, with false-negative rate ranging from 2.5% to 17.9%. In our material, 18.7% of carcinomas were initialy left undetected by FNAC, and subsequently confirmed by CNB. All confirmed carcinomas were correctly suspected on imaging, and categorized as BI-RADS 4 or 5, while all BI-RADS 2 and 3 findings were confirmed as benign on hystology. False-positive rate of imaging was 8%. An average number of 4 tissue cores (range: 2 - 7) was taken in our experience if good quality of the first 3 core was achieved, and there was no consistent reason to proceed with sampling.

Migration of the localization wire to the back in patient with nonpalpable breast carcinoma: A case report

BACKGROUND Due to the increasing number of diagnosed nonpalpable breast cancer cases,wire localization has been commonly performed for surgical guidance to remove nonpalpable breast lesions.This report presents a rare case of localized wire migration to a subcutaneous lesion of the upper back in a breast cancer patient undergoing breast-conserving surgery.CASE SUMMARY A 48-year-old female was scheduled for breast-conserving surgery for left breast cancer.Ultrasonography guided wire localization was performed intraoperatively by surgeon to localize the nonpalpable breast cancer.After axilla sentinel lymph node biopsy,we realized that the wire was not visualized.The wire was not found in the operation field,including the breast and axilla.Breast-conserving surgery was performed after wire re-localization.Intraoperative chest posteroanterior view revealed that the wire was located on the level of midaxillary line.Two days after the operation,a serial simple X-ray revealed that the wire was located on the subcutaneous lesion of the back.The wire tip was palpable under the skin of the upper back,and the wire was removed under local anesthesia.CONCLUSION Hooked wire misplacement can lead to fatal complications.Surgeons must consider the possibility of wire migration during breast cancer surgery.

An image segmentation framework for extracting tumors from breast magnetic resonance images

Magnetic resonance imaging(MRI)has been a prevalence technique for breast cancer diagnosis.Computer-aided detection and segmentation of lesions from MRIs plays a vital role for the MRI-based disease analysis.There are two main issues of the existing breast lesion segmentation techniques:requir ing manual delineation of Regions of Interests(ROIs)as a step of initialization;and requiring a large amount of labeled images for model construction or parameter lear ning,while in real clinical or experimental settings,it is highly challenging to get suficient labeled MRIs.To resolve these issues,this work proposes a semi-supervised method for breast tumor segmentation based on super voxel strategies.After image segmentation with advanced cluster techniques,we take a supervised learning step to classify the tumor and nontumor patches in order to automatically locate the tumor regions in an MRI To obtain the opt imal performance of tumor extraction,we take extensive experiments to learn par ameters for tumor segmentation and dassification,and design 225 classifiers corresponding to diferent parameter settings.We call the proposed method as Semi supervised Tumor Segmentation(SSTS),and apply it to both mass and nonmass lesions.Experimental results show better performance of SsTS compared with five state of-the art methods.