Diagnostic Accuracy Of Breast Cancer Research Articles

Multi-task learning for segmentation and classification of breast tumors from ultrasound images

Segmentation and classification of breast tumors are critical components of breast ultrasound (BUS) computer-aided diagnosis (CAD), which significantly improves the diagnostic accuracy of breast cancer. However, the characteristics of tumor regions in BUS images, such as non-uniform intensity distributions, ambiguous or missing boundaries, and varying tumor shapes and sizes, pose significant challenges to automated segmentation and classification solutions. Many previous studies have proposed multi-task learning methods to jointly tackle tumor segmentation and classification by sharing the features extracted by the encoder. Unfortunately, this often introduces redundant or misleading information, which hinders effective feature exploitation and adversely affects performance. To address this issue, we present ACSNet, a novel multi-task learning network designed to optimize tumor segmentation and classification in BUS images. The segmentation network incorporates a novel gate unit to allow optimal transfer of valuable contextual information from the encoder to the decoder. In addition, we develop the Deformable Spatial Attention Module (DSAModule) to improve segmentation accuracy by overcoming the limitations of conventional convolution in dealing with morphological variations of tumors. In the classification branch, multi-scale feature extraction and channel attention mechanisms are integrated to discriminate between benign and malignant breast tumors. Experiments on two publicly available BUS datasets demonstrate that ACSNet not only outperforms mainstream multi-task learning methods for both breast tumor segmentation and classification tasks, but also achieves state-of-the-art results for BUS tumor segmentation. Code and models are available at https://github.com/qqhe-frank/BUS-segmentation-and-classification.git.

5. DIAGNOSTIC ACCURACY OF EX-VIVO FUSION CONFOCAL MICROSCOPY IN BREAST CANCER

Abstract Introduction Ex-Vivo fusion confocal microscopy (EVFCM) provides a fast examination of breast tissues with the potential to reduce the rate of positive surgical margins and the need for additional interventions. This is an observational study aimed at verifying the diagnostic accuracy of EVFCM in diagnosing neoplasia and histologic subtypes. Methods Thirty-six breast tissue samples, including non-neoplastic (n=20) and neoplastic (n=16), underwent scanning using EVFCM. These samples were also processed through routine histopathological examination with hematoxylin and eosin, and the two techniques were subsequently compared. The assessment was performed by a Mohs surgeon dermatologist familiar with EVFCM (evaluator A) and two breast pathology experts (evaluators B and C) without experience in EVFCM, in a blinded diagnosis analysis. Results The ability to recognize neoplasia in the EVFCM images obtained a high level of concordance, as indicated by all kappa values (κ) presenting a p-value <0.001. Investigator A demonstrated a sensitivity and specificity of 100% in diagnosing neoplasia. Evaluators B and C exhibited a sensitivity >87% and a specificity >94% in diagnosing neoplasia, with positive predictive values (PPV)>95% and the negative predictive values (NPV) of 81% and 94%, respectively. The accuracy in recognizing the histological subtype was >86,1%, with a sensitivity >82%, specificity >88%, PPV>75% and NPV>92%. Conclusion EVFCM demonstrates excellent diagnostic accuracy for breast cancer, with high levels of concordance, sensitivity and specificity. Based on our data, we suggest and propose that EVFCM will be an excellent tool for intra-operative and definitive margin assessment, guiding breast conserving surgeries.

Added value of the EUSOBI diffusion levels in breast MRI

Abstract Objectives To investigate whether using the diffusion levels (DLs) proposed by the European Society of Breast Imaging (EUSOBI) improves the diagnostic accuracy of breast MRI. Materials and methods This retrospective study included 145 women who, between September 2019 and June 2020, underwent breast 1.5-T MRI with DWI. Reader 1 and reader 2 (R1-R2) independently assessed breast lesions using the BI-RADS on dynamic contrast-enhanced imaging and T2-weighted imaging. DWI was subsequently disclosed, allowing readers able to measure lesions ADC and subjectively express the overall risk of malignancy on a 1–5 Likert scale. ADCs were interpreted as a range of values corresponding to the EUSOBI DLs. The analysis evaluated the inter-reader agreement in measuring ADC and DLs, the per-DL malignancy rate, and accuracy for malignancy using ROC analysis against histological examination or a 3-year follow-up. Results Lesions were malignant and showed non-mass enhancement in 67.7% and 76.1% of cases, respectively. ADC was measurable in 63.2%/66.7% of lesions (R1/R2), with a minimal discrepancy on Bland–Altman analysis and 0.948 (95%CI 0.925–0.965)/0.989 (95%CI 0.988–0.991) intraclass correlation coefficient in measuring ADC/DLs. The malignancy rate (R1/R2) increased from 0.5/0.5% (“very high” DL) to 96.0/96.8% (“very low” DL), as expected. Likert categorization showed larger areas under the curve than the BI-RADS for both R1 (0.91 versus 0.87; p = 0.0208) and R2 (0.91 versus 0.89; p = 0.1171), with improved specificity (81.5% versus 78.5% for R1 and 84.4% versus 81.2% for R2). Conclusion Though ADC was not measurable in about one-third of lesions, DLs were categorized with excellent inter-reader agreement, improving the specificity for malignancy. Clinical relevance statement DLs proposed by the EUSOBI are a reproducible tool to interpret the ADC of breast lesions and, in turn, to improve the specificity of breast MRI and reduce unnecessary breast biopsies. Key Points • The European Society of Breast Imaging proposed diffusion levels for the interpretation of the apparent diffusion coefficient in diffusion-weighted imaging of the breast. • Adding diffusion levels to the interpretation of magnetic resonance imaging improved the diagnostic accuracy for breast cancer, especially in terms of specificity. • Diffusion levels can favor a more widespread and standardized use of diffusion-weighted imaging of the breast.

Deep learning applications to breast cancer detection by magnetic resonance imaging: a literature review.

Deep learning analysis of radiological images has the potential to improve diagnostic accuracy of breast cancer, ultimately leading to better patient outcomes. This paper systematically reviewed the current literature on deep learning detection of breast cancer based on magnetic resonance imaging (MRI). The literature search was performed from 2015 to Dec 31, 2022, using Pubmed. Other database included Semantic Scholar, ACM Digital Library, Google search, Google Scholar, and pre-print depositories (such as Research Square). Articles that were not deep learning (such as texture analysis) were excluded. PRISMA guidelines for reporting were used. We analyzed different deep learning algorithms, methods of analysis, experimental design, MRI image types, types of ground truths, sample sizes, numbers of benign and malignant lesions, and performance in the literature. We discussed lessons learned, challenges to broad deployment in clinical practice and suggested future research directions.

Spatiotemporal Learning of Dynamic Positron Emission Tomography Data Improves Diagnostic Accuracy in Breast Cancer

Positron emission tomography (PET) is a noninvasive imaging technology able to assess the metabolic or functional state of healthy and/or pathological tissues. In clinical practice, PET data are usually acquired statically and normalized for the evaluation of the standardized uptake value (SUV). In contrast, dynamic PET acquisitions provide information about radiotracer delivery to tissue, its interaction with the target and its physiological washout. The shape of the time activity curves (TACs) embeds tissue-specific biochemical properties. Conventionally, TACs are employed along with information about blood plasma activity concentration, i.e., the arterial input function, and tracer-specific compartmental models to obtain a full quantitative analysis of PET data. This method’s primary disadvantage is the requirement for invasive arterial blood sample collection throughout the whole PET scan. In this study, we employ a variety of deep learning models to illustrate the diagnostic potential of dynamic PET acquisitions of varying lengths for discriminating breast cancer lesions in the absence of arterial blood sampling compared to static PET only. Our findings demonstrate that the use of TACs, even in the absence of arterial blood sampling and even when using only a share of all timeframes available, outperforms the discriminative ability of conventional SUV analysis.

HoVer-Trans: Anatomy-aware HoVer-Transformer for ROI-free Breast Cancer Diagnosis in Ultrasound Images.

Ultrasonography is an important routine examination for breast cancer diagnosis, due to its non-invasive, radiation-free and low-cost properties. However, the diagnostic accuracy of breast cancer is still limited due to its inherent limitations. Then, a precise diagnose using breast ultrasound (BUS) image would be significant useful. Many learning-based computer-aided diagnostic methods have been proposed to achieve breast cancer diagnosis/lesion classification. However, most of them require a pre-define region of interest (ROI) and then classify the lesion inside the ROI. Conventional classification backbones, such as VGG16 and ResNet50, can achieve promising classification results with no ROI requirement. But these models lack interpretability, thus restricting their use in clinical practice. In this study, we propose a novel ROI-free model for breast cancer diagnosis in ultrasound images with interpretable feature representations.We leverage the anatomical prior knowledge that malignant and benign tumors have different spatial relationships between different tissue layers, and propose a HoVer-Transformer to formulate this prior knowledge. The proposed HoVer-Trans block extracts the inter- and intra-layer spatial information horizontally and vertically. We conduct and release an open dataset GDPH&SYSUCC for breast cancer diagnosis in BUS. The proposed model is evaluated in three datasets by comparing with four CNN-based models and three vision transformer models via five-fold cross validation. It achieves state-of-the-art classification performance (GDPH&SYSUCC AUC: 0.924, ACC: 0.893, Spec: 0.836, Sens: 0.926) with the best model interpretability. In the meanwhile, our proposed model outperforms two senior sonographers on the breast cancer diagnosis when only one BUS image is given (GDPH&SYSUCC-AUC ours: 0.924 vs. reader1: 0.825 vs. reader2: 0.820).

MicroRNA Dysregulation in Early Breast Cancer Diagnosis: A Systematic Review and Meta-Analysis.

Breast cancer continues to be the leading cause of death in women worldwide. Mammography, which is the current gold standard technique used to diagnose it, presents strong limitations in early ages where breast cancer is much more aggressive and fatal. MiRNAs present in numerous body fluids might represent a new line of research in breast cancer biomarkers, especially oncomiRNAs, known to play an important role in the suppression and development of neoplasms. The aim of this systematic review and meta-analysis was to evaluate dysregulated miRNA biomarkers and their diagnostic accuracy in breast cancer. Two independent researchers reviewed the included studies according to the preferred reporting items for systematic reviews and meta-analyses (PRISMA) guidelines. A protocol for this review was registered in PROSPERO with the registration number "CRD42021256338". Observational case-control-based studies analyzing concentrations of microRNAs which have been published within the last 10 years were selected, and the concentrations of miRNAs in women with breast cancer and healthy controls were analyzed. Random-effects meta-analyses of miR-155 were performed on the studies which provided enough data to calculate diagnostic odds ratios. We determined that 34 microRNAs were substantially dysregulated and could be considered biomarkers of breast cancer. Individually, miR-155 provided better diagnostic results than mammography on average. However, when several miRNAs are used to screen, forming a panel, sensitivity and specificity rates improve, and they can be associated with classic biomarkers such us CA-125 or CEA. Based on the results of our meta-analysis, miR-155 might be a promising diagnostic biomarker for this patient population.

32P Improvement of diagnostic accuracy of breast cancer using multi-protein signature markers through machine learning

We have developed a 3-protein signature blood marker (Mastocheck®) for early diagnosis of breast cancer. The purpose of this study is to improve the performance of the previously developed blood markers. Blood from 196 breast cancer patients and 196 healthy control groups were prospectively collected.Through the development of a biomarker detectable library, PepQuant, peptides that are optimal for MS/MS detection were selected. After chemically synthesizing these selected proteins, these were quantified by multiple reaction monitoring (MRM) methods. Seven final proteins were derived by applying the PepQuant library for breast cancer biomarker discovery and verification. Machine learning algorithms was trained as protein candidates identified between breast cancer patients and healthy controls. As in previous studies, performance evaluation was conducted based on sensitivity, specificity, accuracy, false positive rate, false negative rate, positive predictive value, and negative predictive value for breast cancer diagnosis. The sensitivity, specificity, and accuracy of Mastocheck®, were 69.4%, 83.7%, and 76.5%, respectively, which is relatively similar to that of previous studies’ results. The false positive rate(FPR) and the false negative rate(FNR) were 16.3% and 30.7%, and the positive predictive value(PPV) and negative predictive value(NPV) were 81.0% and 73.2%, respectively. During the study when 7-protein signature was combined with an artificial intelligence(AI) technique for the analysis, the sensitivity, specificity, and accuracy of diagnosis were 88.3%, 83.2%, and 85.7%, respectively, showing superior performance compared to Mastocheck®. The FPR and FNR were 16.8% and 11.7%, indicating that the FNR was improved by 20% compared to Mastecheck®. In addition, it was recognized that the PPV and NPV were also improved to 84.0% and 87.6%. Through the collection of new prospective samples, the study confirmed that the diagnostic performance of Mastocheck® was repeatedly maintained. In addition, breast cancer diagnosis using 7-protein signatures with AI model showed that breast cancer diagnosis can be remarkably improved.

Deep learning model for breast cancer diagnosis based on bilateral asymmetrical detection (BilAD) in digital breast tomosynthesis images.

The purpose of this study was to develop a deep learning model to diagnose breast cancer by embedding a diagnostic algorithm that examines the asymmetry of bilateral breast tissue. This retrospective study was approved by the institutional review board. A total of 115 patients who underwent breast surgery and had pathologically confirmed breast cancer were enrolled in this study. Two image pairs [230 pairs of bilateral breast digital breast tomosynthesis (DBT) images with 115 malignant tumors and contralateral tissue (M/N), and 115 bilateral normal areas (N/N)] were generated from each patient enrolled in this study. The proposed deep learning model is called bilateral asymmetrical detection (BilAD), which is a modified convolutional neural network (CNN) model of Xception with two-dimensional tensors for bilateral breast images. BilAD was trained to classify the differences between pairs of M/N and N/N datasets. The results of the BilAD model were compared to those of the unilateral control CNN model (uCNN). The results of BilAD and the uCNN were as follows: accuracy, 0.84 and 0.75; sensitivity, 0.73 and 0.58; and specificity, 0.93 and 0.92, respectively. The mean area under the receiver operating characteristic curve of BilAD was significantly higher than that of the uCNN (p = 0.02): 0.90 and 0.84, respectively. The proposed deep learning model trained by embedding a diagnostic algorithm to examine the asymmetry of bilateral breast tissue improves the diagnostic accuracy for breast cancer.

Spatiotemporal learning of dynamic positron emission tomography data improves diagnostic accuracy in breast cancer.

Positron emission tomography (PET) can reveal metabolic activity in a voxelwise manner. PET analysis is commonly performed in a static manner by analyzing the standardized uptake value (SUV) obtained from the plateau region of PET acquisitions. A dynamic PET acquisition can provide a map of the spatiotemporal concentration of the tracer in vivo, hence conveying information about radiotracer delivery to tissue, its interaction with the target and washout. Therefore, tissue-specific biochemical properties are embedded in the shape of time activity curves (TACs), which are generally used for kinetic analysis. Conventionally, TACs are employed along with information about blood plasma activity concentration, i.e., the arterial input function (AIF), and specific compartmental models to obtain a full quantitative analysis of PET data. The main drawback of this approach is the need for invasive procedures requiring arterial blood sample collection during the whole PET scan. In this paper, we address the challenge of improving PET diagnostic accuracy through an alternative approach based on the analysis of time signal intensity patterns. Specifically, we demonstrate the diagnostic potential of tissue TACs provided by dynamic PET acquisition using various deep learning models. Our framework is shown to outperform the discriminative potential of classical SUV analysis, hence paving the way for more accurate PET-based lesion discrimination without additional acquisition time or invasive procedures. Clinical Relevance- The diagnostic accuracy of dynamic PET data exploited by deep-learning based time signal intensity pattern analysis is superior to that of static SUV imaging.

Abstract P5-12-04: Osteogenic cocktail induces calcifications in human breast cancer cell line via placental alkaline phosphatase expression

Abstract Breast cancer is frequently characterized by calcifications in mammography. Although breast calcifications were previously thought to be produced passively as the final stage of cell regression, recent studies have shown active production. In another context, vascular smooth muscle cells have been found to produce vascular calcifications by acquiring osteoblast-like characteristics. The same mechanism has been suggested to account for breast calcifications. A report showed that breast calcifications are clearly generated by the mouse breast cancer cell line 4T1 through osteoblast-like characteristics. However, the details underlying such a mechanism of mammary calcification is unknown. Understanding this mechanism will improve diagnostic accuracy for breast cancer. We induced calcifications using osteogenic cocktail on the human breast cancer cell line MDA-MB-231. Our hypothesis was that breast cancer calcifications are generated in a process similar to that of osteoblast differentiation. By a comprehensive analysis of transcription compared with osteoblastic cell line MC3T3-E1 in a microarray, we showed that placental alkaline phosphatase (PLAP) is involved in producing calcifications through PI3K-Akt signaling pathways. Herein, we demonstrated that calcifications occur and that alkaline phosphatase enzyme activity increases in MDA-MB-231 cells cultured using an osteogenic cocktail-containing medium. Microarray transcript analysis showed that the PI3K-Akt signaling pathway was significantly involved, with recruitment of placental alkaline phosphatase. Calcifications and alkaline phosphatase enzyme activity were suppressed by silencing placental alkaline phosphatase using a small interfering RNA. Inhibition of the PI3K-Akt signaling pathway suppressed phospho-c-Jun and placental alkaline phosphatase and resulted in absence of calcifications. These findings reveal that breast cancer cells acquire alkaline phosphatase enzyme activity via placental alkaline phosphatase expression and suggest that breast calcification formation is closely associated with the PI3K-Akt signaling pathway. Our results suggest that PLAP produces ALP enzyme activity required for calcification in MDA-MB-231 cells. PLAP is clinically expressed in some breast cancers. A report showed that dexamethasone induces PLAP enzyme activity and mRNA in the human breast cancer BC-M1 cell line. Therefore, dexamethasone in osteogenic cocktail may be associated with PLAP expression in the MDA-MB-231 cells. On the other hand, c-Jun binds to the promoter region of ALPP in MDA-MB-231, and we showed that c-Jun phosphorylation appears in MDA-MB-231 cells cultured using osteogenic cocktail. We suggest that PLAP transcript would be activated when phospho-c-Jun enters the nucleus and binds to the promoter region of PLAP. Our proposed mechanism in breast calcifications is that osteogenic cocktail induces c-Jun phosphorylation via PI3K-Akt signaling pathway and that PLAP transcription would be activated when phospho-c-Jun enters the nucleus and binds to the promoter region of PLAP. Future research might assess whether increased PLAP levels are the main driver of intracellular calcifications in breast cancer, as well as whether the PI3K-Akt signaling pathway has a similar effect in vivo that we demonstrated to cause calcifications in vitro. Citation Format: Atsushi Fushimi, Hiroshi Takeyama, Toshiaki Tachibana, Yoshinobu Manome. Osteogenic cocktail induces calcifications in human breast cancer cell line via placental alkaline phosphatase expression [abstract]. In: Proceedings of the 2021 San Antonio Breast Cancer Symposium; 2021 Dec 7-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2022;82(4 Suppl):Abstract nr P5-12-04.

Deep-Learning-Empowered Breast Cancer Auxiliary Diagnosis for 5GB Remote E-Health

Breast cancer, the most common cancer in women, is receiving increasing attention. The lack of high-quality medical resources, especially highly skilled doctors, in remote areas makes the diagnosis of breast cancer inefficient and causes great harm to women. The emergence of remote e-health has improved the situation to a certain extent, but its capabilities are still hampered by technical limitations, which manifest in two main aspects. First, due to network bandwidth limitations, it is difficult to guarantee the real-time transmission of breast cancer pathology images between remote areas and cities. Second, the highly skilled breast cancer doctors at large city hospitals are not guaranteed to be available for online diagnosis at all times. To overcome these limitations, this article proposes a deep-learning-empowered breast cancer auxiliary diagnosis scheme for remote e-health supported by 5G technology and beyond (5GB remote e-health). In this scheme, breast pathology images are first received from major hospitals via 5G, and a deep learning model based on the Inception-v3 network is subjected to transfer learning to obtain a diagnostic model. This diagnostic model is then employed on edge servers for auxiliary diagnosis at remote area hospitals. A theoretical analysis and experimental results show that this solution not only overcomes the two problems mentioned above but also improves the diagnostic accuracy for breast cancer in remote areas to 98.19 percent.

Additive value of exposure parameters for breast cancer diagnosis in digital mammography.

To investigate the effect of different breast lesions on exposure parameters in digital mammography and to determine whether the exposure parameters can additively improve diagnostic efficiency. Craniocaudal view and mediolateral view full-field digital mammography images from 982 women with unilateral lesions (341 with malignant lesions, 189 with benign lesions, and 452 healthy women) obtained at Nanfang Hospital were reviewed. Differences in exposure parameters (tube voltage and load, breast thickness (BT), and average glandular dose (AGD)) between breasts were calculated. The relationships between parameter differences and lesion size were explored. A logistic regression model was used based on the AGD and BT differences, and the area under the receiver operating characteristic curve (AUC) was used to assess the performance of these parameters in differentiating malignant from benign and healthy subjects. Independently, data from 129 women (82 with malignant and 47 with benign lesions) treated at Sun Yat-sen Memorial Hospital were collected to validate the model. Differences in tube voltage and load, BT, and AGD between breasts were significantly greater in the malignant subjects than benign (p < 0.05) and healthy subjects (p < 0.05). The AUCs for the comparisons of malignant vs. healthy subjects, malignant vs. benign subjects, and benign vs. healthy subjects were 0.77 ± 0.02, 0.72 ± 0.02, and 0.57 ± 0.02, respectively. The model combining the exposure parameters with the BI-RADS category resulted in a higher AUC (0.910 ± 0.03) compared with physician diagnosis alone (0.820 ± 0.04) for differentiating between malignant and benign lesions. Exposure parameters additively improved diagnostic accuracy for breast cancer and yielded more reliable results. • Differences in kVp, mAs, BT, and AGD between breasts were significantly greater in the malignant subjects than benign and healthy subjects. • The model combining exposure parameters with the BI-RADS category resulted in a higher AUC compared with the physician's diagnosis for differentiating between malignant and benign lesions. • Exposure parameters additively improved diagnostic accuracy for breast cancer.

Abstract PO-003: Early detection of breast cancer during the Covid-19 pandemic using artificial intelligence neural networks

Abstract Currently, as governments across the world attempt to halt the spread of COVID-19, many of their efforts are being hampered by shortages of crucial testing kits. Even in many developed countries, the health system has come to the point of standstill due to the increasing demand for intensive care units. One study reveals that 40,000 women die in a year from breast cancer, which is one woman every 13 minutes dying from this disease every day. Breast cancer remains the number one form of cancer that women are diagnosed with around the world. Any hospital in the world will have a multidisciplinary team of experts including radiologists, oncologists, surgeons, and other respiratory physicians and consultants. This multidisciplinary team uses a wide variety of data taken from CT scans, x-rays, PET scans, biopsies, clinical examination, imaging mammography, ultrasound (sonography), pathology biopsy, and fine-needle aspiration to identify whether a patient has cancer. Individuals with cancer, particularly those who are receiving systemic anticancer treatments, have been postulated to be at increased risk of mortality from COVID-19. Therefore, it is the need of an hour to implement an early detection of cancer that will prevent patients from COVID-19 and due treatment will be provided. Breast cancer that is detected early is far easier to cure, but at the same time it is difficult to detect. An accurate breast cancer staging is an essential task performed by pathologists (experts who examine the laboratory samples) to inform clinical management for the appropriate treatment. For instance, evaluation of the extent of cancer spread by histopathologic analysis or microscopic examination is an important part of breast cancer staging, and the process involves the decision-making of human behavior with time-consuming and costly sample examination and a period of more than one month to have breast cancer type, stage, etc., to decide the appropriate therapy to patients by the time it can affect the results of therapeutic treatments and survival of patients. The application of artificial intelligence (AI) machine learning technology with deep learning algorithms to whole-slide pathology images can potentially improve the diagnostic accuracy of breast cancer at a very early stage. I have developed a simple machine-learning model using the Python programming language and neural network, inspired by the structure of the human brain, with 12 attributes, out of which 10 real-valued features from each cell nucleus obtained from the Wisconsin diagnostic breast cancer dataset available in public domain for the education capacity that explains about the stage of breast cancer M (Malignant) and B (Benign). This model predicts the presence of breast cancer with an accuracy of 98.8% at a very early stage. The continuous development of technology in the medical field will save countless lives and the overall quality of human life will continue to improve over time. Citation Format: Subash Kumar. Early detection of breast cancer during the Covid-19 pandemic using artificial intelligence neural networks [abstract]. In: Proceedings of the AACR Virtual Meeting: COVID-19 and Cancer; 2020 Jul 20-22. Philadelphia (PA): AACR; Clin Cancer Res 2020;26(18_Suppl):Abstract nr PO-003.

Interpretative characteristics and case features associated with the performances of radiologists in reading mammograms: A study from a non-screening population in Asia.

To explore radiologist characteristics and case features associated with diagnostic performances in cancer detection on mammograms in a South East Asian population. Fifty-three radiologists reported 60 mammographic examinations which consisted of 40 normal and 20 cancer-containing cases at the BREAST workshops. Radiologists were asked to examine each mammogram using the BIRADS on diagnostic monitors. Differences in reader characteristics and case features between correct and incorrect decisions were assessed separately for cancer and normal cases. Univariate and multivariate logistic regressions were applied to generate odds ratios (OR) for significant factors related to correct decisions. Radiologists who spent ≥10hours/week reporting mammograms had a higher possibility of detecting cancer lesions (OR=1.6; P=0.01). A higher rate of accuracy in reporting negative cases was associated with female radiologists (OR=1.4; P=0.002), radiologists who read ≤20 mammograms per week (OR=1.5; P<0.0001), had completed training course (OR=1.7; P<0.0001) or wore eyeglasses (OR=1.4; P=0.01). Cancer cases with breast density>50% (OR=2.1; P<0.0001), having abnormal lesions ≥9mm (OR=1.8; P<0.0001), or displaying calcifications, a discrete mass or nonspecific density (OR=1.6; P<0.0001) were recorded with a higher detection rate by radiologists than other cases. Lesions located on the right breasts (OR=1.8; P<0.0001) or found in the lower inner, upper outer or mixed locations (OR=2.7; P<0.0001) were also recorded with a better diagnostic possibility compared with other lesions. This work identified key features related to diagnostic accuracy of breast cancer on mammograms in a nonscreening population, which is helpful for developing appropriate strategies to improve breast cancer detectability of radiologists.

Breast Cancer Image Classification via Multi-Network Features and Dual-Network Orthogonal Low-Rank Learning

Histopathological image analysis is an important technique for early diagnosis and detection of breast cancer in clinical practice. However, it has limited efficiency and thus the detection of breast cancer is still an open issue in medical image analysis. To improve the early diagnostic accuracy of breast cancer and reduce the workload of doctors, we devise a classification framework based on histology images by combining deep learning with machine learning methodologies in this paper. Specifically, we devise a multi-network feature extraction model by using pre-trained deep convolution neural networks (DCNNs), develop an effective feature dimension reduction method and train an ensemble support vector machine (E-SVM). First, we preprocess the histological images via scale transformation and color enhancement methods. Second, the multi-network features are extracted by using four pre-trained DCNNs (e.g., DenseNet-121, ResNet-50, multi-level InceptionV3, and multi-level VGG-16). Third, a feature selection method via dual-network orthogonal low-rank learning (DOLL) is further developed for performance boosting and overfitting alleviation. Finally, an E-SVM is trained via fused features and voting strategy to perform the classification task, which classifies the images into four classes (i.e., benign, in situ carcinomas, invasive carcinomas, and normal). We evaluate the proposed method on the public ICIAR 2018 Challenge dataset of histology images of breast cancer and achieve a high classification accuracy of 97.70%. Experimental results show that our method can achieve quite promising performance and outperform state-of-the-art methods.

Circulating miRNAs in Untreated Breast Cancer: An Exploratory Multimodality Morpho-Functional Study.

The aim of this study was to identify new disease-related circulating miRNAs with high diagnostic accuracy for breast cancer (BC) and to correlate their deregulation with the morpho-functional characteristics of the tumour, as assessed in vivo by positron emission tomography/magnetic resonance (PET/MR) imaging. A total of 77 untreated female BC patients underwent same-day PET/MR and blood collection, and 78 healthy donors were recruited as negative controls. The expression profile of 84 human miRNAs was screened by using miRNA PCR arrays and validated by real-time PCR. The validated miRNAs were correlated with the quantitative imaging parameters extracted from the primary BC samples. Circulating miR-125b-5p and miR-143-3p were upregulated in BC plasma and able to discriminate BC patients from healthy subjects (miR-125-5p area under the receiver operating characteristic ROC curve (AUC) = 0.85 and miR-143-3p AUC = 0.80). Circulating CA15-3, a soluble form of the transmembrane glycoprotein Mucin 1 (MUC-1) that is upregulated in epithelial cancer cells of different origins, was combined with miR-125b-5p and improved the diagnostic accuracy from 70% (CA15-3 alone) to 89% (CA15-3 plus miR-125b-5p). MiR-143-3p showed a strong and significant correlation with the stage of the disease, apparent diffusion coefficient (ADCmean), reverse efflux volume transfer constant (Kepmean) and maximum standardized uptake value (SUVmax), and it might represent a biomarker of tumour aggressiveness. Similarly, miR-125b-5p was correlated with stage and grade 2 but inversely correlated with the forward volume transfer constant (Ktransmean) and proliferation index (Ki67), suggesting a potential role as a biomarker of a relatively more favourable prognosis. In situ hybridization (ISH) experiments revealed that miR-143-3p was expressed in endothelial tumour cells, miR-125-5p in cancer-associated fibroblasts, and neither in epithelial tumour cells. Our results suggested that miR-125-5p and miR-143-3p are potential biomarkers for the risk stratification of BC, and Kaplan-Maier plots confirmed this hypothesis. In addition, the combined use of miR-125-b-5p and CA15-3 enhanced the diagnostic accuracy up to 89%. This is the first study that correlates circulating miRNAs with in vivo quantified tumour biology through PET/MR biomarkers. This integration elucidates the link between the plasmatic increase in these two potential circulating biomarkers and the biology of untreated BC. In conclusion, while miR-143-3b and miR-125b-5p provide valuable information for prognosis, a combination of miR-125b-5p with the tumour marker CA15-3 improves sensitivity for BC detection, which warrants consideration by further validation studies.

Proton MR spectroscopy in the breast: Technical innovations and clinical applications.

Proton magnetic resonance spectroscopy (MRS) is a promising noninvasive diagnostic technique for investigation of breast cancer metabolism. Spectroscopic imaging data may be obtained following contrast-enhanced MRI by applying the point-resolved spectroscopy sequence (PRESS) or the stimulated echo acquisition mode (STEAM) sequence from the MR voxel encompassing the breast lesion. Total choline signal (tCho) measured in vivo using either a qualitative or quantitative approach has been used as a diagnostic test in the workup of malignant breast lesions. In addition to tCho metabolites, other relevant metabolites, including multiple lipids, can be detected and monitored. MRS has been heavily investigated as an adjunct to morphologic and dynamic MRI to improve diagnostic accuracy in breast cancer, obviating unnecessary benign biopsies. Besides its use in the staging of breast cancer, other promising applications have been recently investigated, including the assessment of treatment response and therapy monitoring. This review provides guidance on spectroscopic acquisition and quantification methods and highlights current and evolving clinical applications of proton MRS. Level of Evidence 5 Technical Efficacy: Stage 5 J. Magn. Reson. Imaging 2019.

The association of five preoperative serum tumor markers and pathological features in patients with breast cancer.

BackgroundThe utility of frequently used serum tumor markers in breast cancer remains controversial. The study aimed to investigate the role of preoperative carcinoembryonic antigen (CEA), cancer antigen 125 (CA125), cancer antigen 153 (CA153), cancer antigen 724 (CA724), and ferritin (FER) in the management of breast cancer and their relationships with pathological features.MethodsA total of 804 patients with breast mass who underwent breast surgery and 305 healthy volunteers were enrolled. Preoperative serum levels of CEA, CA125, CA153, CA724, and FER were measured. And the pathological features of all the patients were recorded. The association of preoperative serum tumor markers and pathological features was analyzed.ResultsAmong the 804 patients, 355 were identified as malignant cases and 449 as benign cases. CEA, CA153, and FER of patients with breast cancer were higher than those of healthy volunteer group and patients with benign breast diseases. The area under curve (AUC) of CEA, CA153, and FER for distinguishing patients with breast cancer and subjects with non‐breast cancer was 0.688 (95% CI: 0.656‐0.721), 0.609 (95% CI: 0.574‐0.645), and 0.623 (95% CI: 0.586‐0.660), respectively. CA153 correlated with tumor size, node status, and TNM stage, whereas CA125 with node status. No statistic differences of the five markers were observed among the four molecular subtypes.ConclusionPreoperative levels of CEA, CA153, and FER exhibit low diagnostic accuracy for breast cancer (stage I‐III). CA153 correlates with tumor burden, suggesting its prognostic value. The five serum markers do not correlate with molecular subtypes.

Performance evaluation of breast cancer diagnosis with mammography, ultrasonography and magnetic resonance imaging.

Various imaging modalities have been used to diagnose suspicious breast lesions. Purpose of this study is to compare the diagnostic accuracy for breast cancer using mammography, ultrasonography and magnetic resonance imaging (MRI). Total 107 patients aged from 19 to 62 years are included in this retrospective study. Mammography, ultrasonography and MRI scans were performed for each patient detected with suspected breast tumor within a month. In addition, the tumor diversity (10 types of benign and 5 types of malignant) was confirmed by pathological findings of tumor biopsy. To compare the diagnosis performance of the three imaging modalities, the overall fraction correct (accuracy), positive predict value (PPV), negative predict value (NPV), sensitivity and specificity were calculated. Meanwhile, the receiver operating characteristic (ROC) analysis was also performed. The diagnostic accuracy ranged from 78.5% to 86.9% among three imaging modalities. All modalities yielded a PPV lower than 77.8% and a NPV higher than 90.0% in identifying the presence of malignant tumors. MRI presented a diagnostic accuracy of 86.9%, as well as a sensitivity of 95.5% and an area under curve (AUC) of 0.948, which are higher than mammography and ultrasonography. By using a diverse dataset and comparing the diagnostic accuracy of three imaging modalities commonly used in breast cancer detection and diagnosis, this study also demonstrated that mammography, ultrasonography and MRI had different diagnostic performance in breast tumor identification. Among them, MRI yielded the highest performance even though the unexpected specificity may lead to over-diagnosis, and ultrosonography is slightly better than mammography.