Ultrasonic Diagnosis of Breast Nodules Using Modified Faster R-CNN

2019 ◽  
Vol 41 (6) ◽  
pp. 353-367 ◽  
Author(s):  
Zihao Zhang ◽  
Xuesheng Zhang ◽  
Xiaona Lin ◽  
Licong Dong ◽  
Sure Zhang ◽  
...  
Keyword(s):  
Breast Cancer ◽  
Breast Imaging ◽  
Detection Method ◽  
Malignant Tumors ◽  
Ultrasonic Diagnosis ◽  
Test Results ◽  
Average Precision ◽  
Multi Scale ◽  
The Mean

Breast cancer has become the biggest threat to female health. Ultrasonic diagnosis of breast cancer based on artificial intelligence is basically a classification of benign and malignant tumors, which does not meet clinical demand. Besides, the current target detection method performs poorly in detecting small lesions, while it is clinically required to detect nodules below 2 mm. The objective of this study is to (a) propose a diagnostic method based on Breast Imaging Reporting and Data System (BI-RADS) and (b) increase its detectability of small lesions. We modified the framework of Faster R-CNN (Faster Region-based Convolutional Neural Network) by introducing multi-scale feature extraction and multi-resolution candidate bound extraction into the network. Then, it was trained using 852 images of BI-RADS C2, 739 images of C3, and 1662 images of malignancy (BI-RADS 4a/4b/4c/5/6). We compared our model with unmodified Faster R-CNN and YOLO v3 (You Only Look Once v3). The mean average precision (mAP) is significantly increased to 0.913, while its average detection speed is slightly declined to 4.11 FPS (frames per second). Meanwhile, its detectivity of small lesions is effectively improved. Moreover, we also tentatively applied our model on video sequences and got satisfactory results. We modified Faster R-CNN and trained it partly based on BI-RADS. Its detectability of lesions, as well as small nodules, was significantly improved. In view of wide coverage of dataset and satisfactory test results, our method can basically meet clinical needs.

scholarly journals Examination of Abnormal Behavior Detection Based on Improved YOLOv3

Electronics ◽  
2021 ◽  
Vol 10 (2) ◽  
pp. 197
Author(s):  
Meng-ting Fang ◽  
Zhong-ju Chen ◽  
Krzysztof Przystupa ◽  
Tao Li ◽  
Michal Majka ◽  
...  
Keyword(s):  
Detection Method ◽  
Abnormal Behavior ◽  
Detection Accuracy ◽  
Average Precision ◽  
Research Results ◽  
Behavior Detection ◽  
Abnormal Behavior Detection ◽  
The Mean ◽  
Examination Room ◽  
Detection Speed

Examination is a way to select talents, and a perfect invigilation strategy can improve the fairness of the examination. To realize the automatic detection of abnormal behavior in the examination room, the method based on the improved YOLOv3 (The third version of the You Only Look Once algorithm) algorithm is proposed. The YOLOv3 algorithm is improved by using the K-Means algorithm, GIoUloss, focal loss, and Darknet32. In addition, the frame-alternate dual-thread method is used to optimize the detection process. The research results show that the improved YOLOv3 algorithm can improve both the detection accuracy and detection speed. The frame-alternate dual-thread method can greatly increase the detection speed. The mean Average Precision (mAP) of the improved YOLOv3 algorithm on the test set reached 88.53%, and the detection speed reached 42 Frames Per Second (FPS) in the frame-alternate dual-thread detection method. The research results provide a certain reference for automated invigilation.

scholarly journals Systemic immune-inflammation index and ultrasonographic classification of breast imaging-reporting and data system predict outcomes of triple-negative breast cancer

2019 ◽  
Vol Volume 11 ◽  
pp. 813-819 ◽  
Author(s):  
Ping Wang ◽  
Wensheng Yue ◽  
Wenyan Li ◽  
Yuqun Luo ◽  
Zukun Li ◽  
...  
Keyword(s):  
Breast Cancer ◽  
Triple Negative Breast Cancer ◽  
Breast Imaging ◽  
Triple Negative ◽  
Data System ◽  
Immune Inflammation

Interinstitutional multidisciplinary virtual tumor board implementation for the management of breast cancer in a public health setting in Mexico.

2021 ◽  
Vol 39 (28_suppl) ◽  
pp. 107-107
Author(s):  
Omar Peña-Curiel ◽  
Orestes Valles-Guerra ◽  
Karen M Velazquez-Ayala ◽  
Griselda Peña-Iturbide ◽  
Sonia Maria Flores Moreno ◽  
...  
Keyword(s):  
Public Health ◽  
Breast Cancer ◽  
Breast Imaging ◽  
Tertiary Care ◽  
Locally Advanced ◽  
Tumor Board ◽  
Pathology Report ◽  
Referral Center ◽  
The Mean ◽  
Mean Time

107 Background: UNEME-DEDICAM (UD) clinics are part of a national public health initiative to provide women prompt access to cervical and breast cancer (BC) screening and diagnosis. Furthermore, UD clinics play a central role in the coordination and prioritization of patient transfer to treatment-specialized institutions. To facilitate this process, we planned and implemented an interinstitutional virtual multidisciplinary tumor board (VMDT). Herein, we present our current experience. Methods: We planned and implemented our VMDT in September 2020. Weekly sessions were established for the multidisciplinary discussion of every newly diagnosed patient at UD with a complete radiology and pathology report. Communication was accomplished through an encrypted and secure internet connection using Microsoft Teams software. VMDT members included breast pathologist, breast imaging, radio oncologist, medical oncologist, and surgical oncologist. Treatment consensus were registered in a Microsoft Word template and integrated into the medical record for each patient. Importantly, the report also included date and time for the consultation at the referral institution. Results: Between September 2020 through May 2021, 74 BC patients were diagnosed at UD. Mean age at diagnosis was 52 years. Sixty-eight patients had invasive BC, of whom early stage (I and II) accounted for 67% of patients; locally advanced (III) for 29%, and advanced (IV) for 4%. Luminal A and B type accounted for 68%; HER2+ve for 25%; and triple negative for 7%. Mean time from biopsy to complete histopathology report (biopsy-report interval) was 2.5 weeks. The mean time from VMDT consensus to patient´s first consultation at referral center (VMDT-referral interval) was 2 weeks. The mean time from biopsy to patient´s first consultation at referral center (total interval) was 5.5 weeks. Conclusions: The VMDT is a plausible strategy to streamline the inter-institutional organization for the timely care of BC patients. UD clinics play a central role in the coordination of transfer of BC patients to tertiary care centers.

scholarly journals Lobular Breast Cancer Conspicuity on Digital Breast Tomosynthesis Compared to Synthesized 2D Mammography: A Multireader Study

2021 ◽  
Vol 7 (9) ◽  
pp. 185
Author(s):  
Giovanna Romanucci ◽  
Lisa Zantedeschi ◽  
Anna Ventriglia ◽  
Sara Mercogliano ◽  
Maria Vittoria Bisighin ◽  
...  
Keyword(s):  
Breast Cancer ◽  
Breast Imaging ◽  
Digital Breast Tomosynthesis ◽  
Matched Pairs ◽  
Breast Cancers ◽  
Test Results ◽  
Lobular Breast Cancer ◽  
Point Scale ◽  
Breast Tomosynthesis ◽  
Invasive Lobular Breast Cancer

Objectives: To compare the conspicuity of lobular breast cancers at digital breast tomosynthesis (DBT) versus synthesized 2D mammography (synt2D). Materials and methods: Seventy-six women (mean age 61.2 years, range 50–74 years) submitted to biopsy in our institution, from 2019 to 2021, with proven invasive lobular breast cancer (ILC) were enrolled in this retrospective study. The participants underwent DBT and synt2D. Five breast radiologists, with different years of experience in breast imaging, independently assigned a conspicuity score (ordinal 6-point scale) to DBT and synt2D. Lesion conspicuity was compared, for each reader, between the synt2D overall conspicuity interpretation and DBT overall conspicuity interpretation using a Wilcoxon matched pairs test. Results: A total of 50/78 (64%) cancers were detected on both synt2D and DBT by all the readers, while 28/78 (26%) cancers where not recognized by at least one reader on synt2D. For each reader, in comparison with synt2D, DBT increased significantly the conspicuity of ILC (p < 0.0001). The raw proportion of high versus low conspicuity by modality confirmed that cancers were more likely to have high conspicuity at DBT than synt2D. Conclusions: ILCs were more likely to have high conspicuity at DBT than at synt2D, increasing the chances of the detection of ILC breast cancer.

Does insurance mix impact utilization of published care practices and guidelines among breast imaging sites?

2014 ◽  
Vol 32 (30_suppl) ◽  
pp. 150-150
Author(s):  
Anne Marie Murphy ◽  
Christine B. Weldon ◽  
Julia Rachel Trosman ◽  
Julian C. Schink ◽  
David Ansell ◽  
...  
Keyword(s):  
Breast Cancer ◽  
Cancer Detection ◽  
Breast Imaging ◽  
Response Rate ◽  
Private Insurance ◽  
Breast Mri ◽  
Test Results ◽  
Exact Test ◽  
Diagnostic Practices ◽  
Care Practices

150 Background: Significant variation exists across Metropolitan Chicago in the quality and timeliness measures for breast cancer detection (Rauscher GH 2014). We examined utilization of published care practices and guidelines at breast imaging sites in Chicago and compared them based on insurance mix. Methods: We conducted an IRB approved web survey of all 58 breast imaging sites in Chicago. Using guidelines (NCCN, NAPBC, ACR) and peer-reviewed literature (38 studies) we developed a survey of breast diagnostic practices. Results analyzed using simple frequencies and Fisher's exact test. Results: We achieved a response rate of 91% (53/58 sites): 27 sites with over 40% privately insured (PI) patients and 26 sites with over 60% Medicare, Medicaid, charity and self pay (MM) patients (IDPH 2012). Utilization of practices vary (Table). The use of breast MRI for diagnostic patients is 81% (17/21) of PI vs. 47% (8/17) of MM sites, p=0.04. Image guided biopsy is used by 81% (17/21) of PI vs. 25% (3/12) of MM sites, p=0.03. Clip placement at biopsy is done at 86% (18/21) of PI vs. 50% (6/12) of MM sites, p=0.04. Conclusions: Sites with higher rates of private insurance show better utilization of three published breast cancer diagnostic care practices. However, improvement is needed across sites, regardless of insurance mix, to provide care to all patients that is up-to-date on published breast cancer screening and diagnostic practices. [Table: see text]

The quantitation of thioredoxin 1 from serum as a novel means to detect breast cancer (BC).

2020 ◽  
Vol 38 (15_suppl) ◽  
pp. 557-557
Author(s):  
Kyoung Hoon Suh
Keyword(s):  
Breast Cancer ◽  
Histological Grade ◽  
Accurate Method ◽  
Mean Value ◽  
Test Results ◽  
Thioredoxin 1 ◽  
The Mean ◽  
Normal Women ◽  
Detect Breast Cancer ◽  
One Way Anova

557 Background: Breast cancer (BC) is the most common cancer and the second leading cause of cancer death in women. Enormous effort has been conducted without success to develop a means to detect BC using the blood. We have reported that the level of thioredoxin 1 (Trx1) in serum could be a novel standard to evaluate the risk of BC. Therefore, we have investigated the clinical utility of Trx1 as a biomarker to detect BC by testing sera from normal women, women with BC, women with five other types of cancer. Methods: We have developed an ELISA kit that quantitates Trx1 in sera. The level of Trx1 was determined in each serum from normal healthy women (n = 114), as well as patients with BC (n = 106), cervical cancer (n = 17), lung cancer (n = 14), stomach cancer (n = 9), and thyroid cancer (n = 4). BC patients were recruited according to their age and cancer stage. Each test was duplicated more than three times, and test results were analyzed by ROC analysis, one-way ANOVA tests, and unpaired t-tests. Results: The mean value of Trx1 from normal women was 5.60±4.39(±SD) and that from BC was 22.25±7.07. The Trx1 level was effective to distinguish BC serum from healthy serum with a sensitivity of 94.3% and specificity of 93.9% (AUC 0.985, p< 0.001). The levels of Trx1 from BC patients were higher than the cut-off value of 14.13 U/ml regardless of age, stage, histological grade, type, ER/PR/HER2 expression profile, and proliferation activity of BC cells. The levels of Trx1 from the other five types of cancers (2.34±1.82 - 3.64±2.99) were low enough to be distinguishable from BC. Especially, Trx1 levels could rescue patients whose mammography resulted in a false judgement. Conclusions: These results indicated that the blood level of Trx1 is an effective and accurate method to detect breast cancer, and particularly as a complement to mammography.

scholarly journals Confidence-Aware Object Detection Based on MobileNetv2 for Autonomous Driving

Sensors ◽  
2021 ◽  
Vol 21 (7) ◽  
pp. 2380
Author(s):  
Wei Li ◽  
Kai Liu
Keyword(s):  
Object Detection ◽  
Loss Function ◽  
Autonomous Vehicles ◽  
Autonomous Driving ◽  
Average Precision ◽  
Detection Model ◽  
Multi Scale ◽  
Proposed Model ◽  
The Mean ◽  
High Level

Object detection is an indispensable part of autonomous driving. It is the basis of other high-level applications. For example, autonomous vehicles need to use the object detection results to navigate and avoid obstacles. In this paper, we propose a multi-scale MobileNeck module and an algorithm to improve the performance of an object detection model by outputting a series of Gaussian parameters. These Gaussian parameters can be used to predict both the locations of detected objects and the localization confidences. Based on the above two methods, a new confidence-aware Mobile Detection (MobileDet) model is proposed. The MobileNeck module and loss function are easy to conduct and integrate with Generalized-IoU (GIoU) metrics with slight changes in the code. We test the proposed model on the KITTI and VOC datasets. The mean Average Precision (mAP) is improved by 3.8 on the KITTI dataset and 2.9 on the VOC dataset with less resource consumption.

scholarly journals Klasifikasi Citra Histopatologi Kanker Payudara menggunakan Data Resampling Random dan Residual Network

2021 ◽  
Vol 11 (1) ◽  
pp. 70-77
Author(s):  
Wahyudi Setiawan
Keyword(s):  
Breast Cancer ◽  
Image Classification ◽  
Training Data ◽  
Unbalanced Data ◽  
Residual Network ◽  
Average Precision ◽  
Average Recall ◽  
Average Accuracy ◽  
Optimal Classification

Data imbalance between classes is one of the problems on image classification. The data in each class is not equal and has a relatively large difference generated in less than optimal classification results. Ideally, the data in each class is equal or have a slight difference. This article discusses the classification of the histopathology breast cancer image. The data consist of  8 classes with unbalanced data. The method for balancing the data in each class uses random resampling which is applied to training data only. The data used from BreakHist with magnifications of 40x, 100x, 200x, and 400x . The classification uses Residual Network (ResNet) 18 and 50. The best results are obtained on images with a magnification of 400x. Classification results using ResNet18 has an average accuracy of 79.82%, an average precision of 71.39%, and an average recall of 82.37%. Meanwhile using ResNet50 showed an average accuracy of 81.67%, average precision of 78.41%, and an average recall of 82.91%.

scholarly journals Classification of Breast Ultrasound Tomography by Using Textural Analysis

2020 ◽  
Vol 17 (2) ◽  
Author(s):  
Chih-Yu Liang ◽  
Tai-Been Chen ◽  
Nan-Han Lu ◽  
Yi-Chen Shen ◽  
Kuo-Ying Liu ◽  
...  
Keyword(s):  
Roc Analysis ◽  
Malignant Tumors ◽  
Rapid Development ◽  
Ultrasound Image ◽  
Breast Ultrasound ◽  
Image Features ◽  
Benign Tumors ◽  
Ultrasound Images ◽  
The Mean

Background: Ultrasound imaging has become one of the most widely utilized adjunct tools in breast cancer screening due to its advantages. The computer-aided detection of breast ultrasound is rapid development via significant features extracted from images. Objectives: The main aim was to identify features of breast ultrasound image that can facilitate reasonable classification of ultrasound images between malignant and benign lesions. Patients and Methods: This research was a retrospective study in which 85 cases (35 malignant [positive group] and 50 benign [negative group] with diagnostic reports) with ultrasound images were collected. The B-mode ultrasound images have manually selected regions of interest (ROI) for estimated features of an image. Then, a fractal dimensional (FD) image was generated from the original ROI by using the box-counting method. Both FD and ROI images were extracted features, including mean, standard deviation, skewness, and kurtosis. These extracted features were tested as significant by t-test, receiver operating characteristic (ROC) analysis and Kappa coefficient. Results: The statistical analysis revealed that the mean texture of images performed the best in differentiating benign versus malignant tumors. As determined by the ROC analysis, the appropriate qualitative values for the mean and the LR model were 0.85 and 0.5, respectively. The sensitivity, specificity, accuracy, positive predicted value (PPV), negative predicted value (NPV), and Kappa for the mean was 0.77, 0.84, 0.81, 0.77, 0.84, and 0.61, respectively. Conclusion: The presented method was efficient in classifying malignant and benign tumors using image textures. Future studies on breast ultrasound texture analysis could focus on investigations of edge detection, texture estimation, classification models, and image features.

scholarly journals Breast Cancer Detection, Diagnosis, and Prediction

2020 ◽  
Vol 9 (6) ◽  
pp. 38-42
Keyword(s):  
Breast Cancer ◽  
Cancer Detection ◽  
Breast Imaging ◽  
Malignant Tumors ◽  
Region Of Interest ◽  
Breast Cancer Detection ◽  
Support Vector ◽  
Svm Classifier ◽  
Cad System ◽  
Mammogram Images

The early detection, diagnosis, prediction, and treatment of breast cancer are challenginghealthcare problems. This study focuses on outlining the traditional and trending techniques used for breast cancer detection, diagnosis, and prediction, including trending noninvasive, nonionizing, and biomarker genetic techniques.In addition, a Computer Aided Detection (CAD) is introduced to classify benign and malignant tumors in mammograms. This CAD system involves three steps. First, the Region of Interest (ROI) that includesthe tumor is identified using a threshold-based method. Second, a deep learning Convolutional Neural Network (CNN) processes the ROI to extract relevant mammogram features. Finally, a Support Vector Machine (SVM) classifier is used to decode two classes of mammogram structures (i.e., Benign (B), and Malignant (M) nodules). The training processes and implementations were carried out using 2800 mammogram images taken from the Curated Breast Imaging Subset of DDSM (CBIS-DDSM). Results have shown that the accuracy of CNN-SVM system achieves 85.1% using AlexNet CNN. Comparison with related work shows the promise of the proposed CAD system

Sign in / Sign up

Export Citation Format

Share Document