Mobile-Aided Breast Cancer Diagnosis by Deep Convolutional Neural Networks

2020 ◽  
pp. 145-162
Keyword(s):  
Breast Cancer ◽  
Image Recognition ◽  
Cancer Diagnosis ◽  
Medical Images ◽  
Human Cancer ◽  
Lobular Carcinoma ◽  
Breast Cancer Diagnosis ◽  
Breast Cancers ◽  
Deep Convolutional Neural Networks ◽  
Important Health

After verifying the capability of deep learning for basic image recognition, this chapter further extends image recognition to App-aided breast cancer diagnosis. Human cancer has been considered as the most important health problem. For medical image recognition of breast cancer, the presented approach is no longer the same as the traditional. It needs no axioms for distinguishing malignant and benign tumors, and no hand-crafted textural descriptors for feature extraction. The goal is to develop a mobile-aided diagnosis system of directly processing raw medical images. It automatically extracts features and learn filters of a deep CNN subject to labelled medical images in advance. This chapter presents a CNN architecture for diagnosing breast cancer images, illustrating effectiveness of problem solving by designing classifiers, respectively diagnosing lobular carcinoma breast cancer against phyllodes tumor and papillary carcinoma against adenosis. The performances of two classifiers for breast cancers diagnosis are separately summarized by the testing accuracy rates of 94.9% and 87.3%.

Breast cancer ER, PR, and HER2 expression variance by germline cancer predisposition genes.

2021 ◽  
Vol 39 (15_suppl) ◽  
pp. 10526-10526
Author(s):  
Grace Wei ◽  
Marilin Rosa ◽  
Maxine Chang ◽  
Brian J. Czerniecki ◽  
Xia Wang
Keyword(s):  
Breast Cancer ◽  
Cancer Diagnosis ◽  
Breast Cancer Diagnosis ◽  
Cancer Predisposition ◽  
Tumor Evolution ◽  
Breast Cancers ◽  
Her2 Expression ◽  
Expression Levels ◽  
Genetic Test Results ◽  
Predisposition Genes

10526 Background: The association between breast cancer characteristics and survival with estrogen receptor (ER) and progesterone receptor (PR) expression has been primarily studied via binomial categories, ER-positive and ER-negative. In order to better characterize germline genetic influences on these markers, we investigated their IHC expression semi-quantitatively in cancer predisposition germline pathogenic variant (PV) carriers of the following genes: BRCA1, BRCA2, PALB2, TP53, PTEN, CDH1, ATM, CHEK2, and Lynch syndrome genes. The HER2 expression was also analyzed. Methods: We conducted a retrospective chart review of patients with germline panel genetic testing for cancer predisposition genes at Moffitt Cancer Center’s GeneHome clinic. Inclusion criteria included 1) women ≥18 years old, 2) breast cancer diagnosis, 3) cancer predisposition germline panel genetic test results, 4) available ER and PR expression levels, and 5) available HER expression and/or amplification status. ER, PR, and HER2 status were compared between PV carriers and non-PV carriers via Mann-Whitney U at p>0.05. Results: A total of 847 cases were reviewed for the study. Among 658 patients with a breast cancer diagnosis and complete ER PR data, 365 cases (55.5%) were non-PV carriers and 293 cases (44.5%) carried a PV in at least one of the genes listed above. Among 635 cases with available HER2 expression/amplification status, 355 (55.9%) cases were non-PV carriers and 288 (45.4%) cases were PV-carriers. When compared with non-PV carrier controls, BRCA1 PV carriers’ breast tumors had significantly lower ER and/or PR expression. Further, BRCA2 and TP53 PV tumors also displayed moderately lower ER expression. Contrarily, CHEK2 tumors displayed higher ER and PR expression compared to controls. Further, BRCA1 and BRCA2 PV carriers were more likely to have HER2- breast cancers. Conclusions: Differences in ER, PR, HER2 expression levels were observed in germline PV carrier breast cancers, signaling differential impacts by germline PVs on the tumor evolution process. It is likely that tumor differences in PV carriers influence responses to therapies, including hormone therapy, anti-HER2 therapy, and subsequent survival.[Table: see text]

Trends in synchronous invasive ductal carcinomas of the breast: A SEER database analysis.

2017 ◽  
Vol 35 (15_suppl) ◽  
pp. e13081-e13081
Author(s):  
Rutika Jitesh Mehta ◽  
Adrienne Groman ◽  
Rohit K. Jain ◽  
Ellis Glenn Levine
Keyword(s):  
Breast Cancer ◽  
Older Women ◽  
Cancer Diagnosis ◽  
Breast Cancer Diagnosis ◽  
Incidence Rates ◽  
Age Group ◽  
Breast Cancers ◽  
Seer Database ◽  
Younger Women ◽  
Synchronous Breast Cancer

e13081 Background: Synchronous breast cancers are uncommon and account for around 2% of all breast cancer diagnosis. Lobular histology is considered a risk factor for synchronous breast cancers. We sought to study the trends in synchronous breast cancer of ductal histology and influence of age by interrogating the SEER database. Methods: The SEER Research data 1973-2013 was interrogated for synchronous infiltrating ductal carcinoma diagnosis (2 diagnosis within 6 months of each other). Overall survival (OS), the primary endpoint, was defined as the time (in months) from diagnosis to death from any cause. Univariate proportional hazards modeling results were used to assess the effect of age, race and stage on overall survival. All associations were considered statistically significant at an alpha error < 0.01. All analyses were performed using SAS version 9.4. Results: 1469 cases were identified. Data was categorized by age group: ≤ 65 years or > 65 years. 60% were 65 years or younger. 85% were Caucasians, 9.6% African Americans and 5.2% others. Younger women (≤ 65 years) had a statistically higher proportion of Stage III/IV breast cancer diagnosis as compared to older women (33.4% vs 25.2%; p = 0.002). The incidence rate of synchronous breast cancers has been rising since 1973, more pronounced in the older age group. Incidence rates overall have risen from 0.09/100,000 persons in 1973-1980 to 0.29/100,000 persons in 2001-2013 (p < 0.001). Incidence rates for synchronous breast cancer in women > 65 years has increased from 0.30/100,000 persons in 1973-1980 to 1.03/100,000 persons in 2001-2013. The adjusted OS among older women is significantly worse than that of younger women (HR 1.05; 95% CI 1.04-1.05; p < 0.001). Conclusions: Better imaging techniques and breast cancer screening guidelines have likely improved breast cancer detection rates thus leading to a rise in the incidence of synchronous breast cancers. It can be speculated that underlying medical problems and advanced age result in more morbidity and subsequent mortality in older women with standard treatment. The finding of more advanced disease among younger women deserves scrutiny as to cause.

The clinical impact of ASCO “choosing wisely” recommendations on staging imaging for early stage breast cancers: An interrupted time-series analysis utilizing SEER-Medicare data.

2020 ◽  
Vol 38 (15_suppl) ◽  
pp. 2078-2078
Author(s):  
Alan Baltz ◽  
Issam Makhoul ◽  
Eric R Siegel
Keyword(s):  
Breast Cancer ◽  
Cancer Diagnosis ◽  
Negative Binomial ◽  
Early Stage ◽  
Breast Cancer Diagnosis ◽  
Choosing Wisely ◽  
Breast Cancers ◽  
Interrupted Time Series Analysis ◽  
Medicare Data ◽  
The Impact

2078 Background: The “Choosing Wisely” (CW) list, released by the American Society for Clinical Oncology (ASCO), highlights low-value procedures. In 2012, the CW recommendations advised against the use of staging imaging, including Positron Emission Tomography (PET), Computerized Tomography (CT) and radionuclide bone scans, for the staging of early breast cancer at low risk for metastasis. The objective of this study was therefore to assess the impact of the ASCO CW recommendations on staging imaging among early stage breast cancers. Methods: Women above the age of 66 with an early stage incident breast cancer diagnoses between 2010 and 2015 were identified within the linked SEER-Medicare data. The primary outcome of interest was the proportion of patients with a claim for staging imaging in the six months following the breast cancer diagnosis. Negative binomial regression, adjusting for pre-recommendation trends, was performed to estimate the changes in the rate of imaging staging within each year following the release of the recommendation. Results: A total of 50,004 women were identified during the study period. Prior to the release of the recommendations in 2012, the staging imaging rates among women newly diagnosed with early stage breast cancers were 5% greater in 2010 (p<.01) and 4% greater in 2011 (p<.01). Following the release of the recommendations, staging imaging rates did not decrease significantly in 2013 (2%;p=0.18). Imaging rates did, however, significantly decrease by 13% in 2014 (p<0.01) and by 16% in 2015 (p<0.01). Conclusions: The CW recommendation was associated with a significant decrease in unadvised staging imaging among incident early stage breast cancer diagnosis in the second and third year following its release. These findings demonstrate an improvement in the proportion of potentially inappropriate staging imaging in early stage breast cancers. The creation and dissemination of resources, such as the CW recommendations, serves as a powerful tool to improve clinical practice, quality of care, and patient safety from secondary malignancies, anxiety, and overdiagnosis.

A fuzzy based high-resolution multi-view deep CNN for breast cancer diagnosis through SVM classifier on visual analysis

2020 ◽  
Vol 39 (6) ◽  
pp. 8573-8586
Author(s):  
Sudhakar Sengan ◽  
V. Priya ◽  
A. Syed Musthafa ◽  
Logesh Ravi ◽  
Saravanan Palani ◽  
...  
Keyword(s):  
Breast Cancer ◽  
Deep Learning ◽  
High Resolution ◽  
Cancer Diagnosis ◽  
Medical Images ◽  
Breast Cancer Diagnosis ◽  
Treatment Decisions ◽  
Diagnosis And Treatment ◽  
Automated System ◽  
Svm Classifier

Breast cancer should be diagnosed as early as possible. A new approach of the diagnosis using deep learning for breast cancer and the particular process using segmentation strategies presented in this article. Medical imagery is an essential tool used for both diagnosis and treatment in many fields of medical applications. But, it takes specially trained medical specialists to read medical images and make diagnoses or treatment decisions. New practices of interpreting medical images are labour exhaustive, time-wasting, expensive, and prone to error. Using a computer-aided program which can render diagnosis and treatment decisions automatically would be more beneficial. A new computer-based detection method for the classification between compassionate and malignant mass tumours in mammography images of the breast proposed. (a) We planned to determine how to use the challenging definition, which produces severe examples that boost the segmentation of mammograms. (b) Employing well designing multi-instance learning through deep learning, we validated employing inadequately labelled data of breast cancer diagnosis using a mammogram. (c) The study is going through the Deep Lung method incorporating deep multi-dimensional automated identification and classification of the lung nodule. (d) By combining a probabilistic graphic model in deep learning, it authorizes how weakly labelled data can be used to improve the existing breast cancer identification method. This automated system involves manually defining the Region Of Interest (ROI), with the region and threshold values based on the next region. The High-Resolution Multi-View Deep Convolutional Neural Network (HRMP-DCNN) mainly developed for the extraction of function. The findings collected through the subsequent in available public databases like mammography screening information database and DDSM Curated Breast Imaging Subset. Ultimately, we’ll show the VGG that’s thousands of times quicker, and it is more reliable than earlier programmed anatomy segmentation.

Elasto-Mammography: Elastic Property Reconstruction in Breast Tissues

2005 ◽  
Vol 874 ◽  
Author(s):  
Z. Wang ◽  
Y. Liu ◽  
L.Z. Sun ◽  
G. Wang
Keyword(s):  
Breast Cancer ◽  
Cancer Diagnosis ◽  
Imaging Modality ◽  
Breast Cancer Diagnosis ◽  
Reconstruction Method ◽  
Breast Cancers ◽  
Primary Method ◽  
Conventional Mammography ◽  
Breast Tissues ◽  
Higher Sensitivity

AbstractMammography is the primary method for screening and detecting breast cancers. However, it frequently fails to detect small tumors and is not quite specific in terms of tumor benignity and malignancy. The objective of this paper is to develop a new imaging modality called elastomammography that generates the modulus elastograms based conventional mammographs. A new elastic reconstruction method is described based on elastography and mammography for breast tissues. Elastic distribution can be reconstructed through the measurement of displacement provided by mammographic projection. It is shown that the proposed elasto-mammography provides higher sensitivity and specificity than the conventional mammography on its own for breast cancer diagnosis.

Racial disparities among women who develop invasive secondary breast cancer.

2021 ◽  
Vol 39 (15_suppl) ◽  
pp. e18523-e18523
Author(s):  
Kekoa Taparra ◽  
Jami Aya Fukui ◽  
Jeffrey Killeen ◽  
Kenneth N. M. Sumida ◽  
Lenora Loo ◽  
...  
Keyword(s):  
Breast Cancer ◽  
Racial Disparities ◽  
White Women ◽  
Native Hawaiian ◽  
Lobular Carcinoma ◽  
Breast Cancer Diagnosis ◽  
Breast Cancers ◽  
Younger Women ◽  
Treatment Information ◽  
Treatment Type

e18523 Background: Noninvasive breast cancers ( e.g. ductal and lobular carcinoma in situ) are highly treatable nonobligate precursors to invasive breast cancers. However, even after treatment, some women develop second breast cancers (SBCs), increasing their mortality risk. Prognosticators may inform treatment recommendations for women at higher risk of SBC. In this study, risk factors for SBC were evaluated using deidentified data from the Hawaiʻi Tumor Registry (HTR), an NCI SEER registry. The HTR covers a uniquely multiethnic, statewide population allowing for elucidation of disparities in understudied U.S. populations. Methods: Women initially diagnosed between 1973-2017 with noninvasive (ductal and lobular) breast cancer were identified. Patient demographics, cancer characteristics, and treatment information were collected. Univariate (UVA) and multivariate (MVA) logistic regression analyses were used to identify factors associated with SBC, defined as a breast cancer diagnosis > 6 months after their prior cancer. Results: Of 7,057 women diagnosed with a first noninvasive breast cancer, 696 (10%) developed SBC. Invasive ipsilateral (iiSBC) and invasive contralateral (icSBC) disease represented 9% and 20% of patients who developed SBC, respectively. The five most prevalent ethnic groups were Chinese, Filipino, Japanese, Native Hawaiian, and White. When adjusting for confounders, women who developed iiSBC were more likely to be Native Hawaiian (odds ratio [OR] = 3.20, 95% CI = 2.07-4.94) or Filipino (OR = 1.72, 95%CI = 1.02-2.91) when compared to Whites; diagnosed between 1990-1999 (OR = 2.06, 95%CI = 1.27-3.34); and not have undergone surgical treatment (OR = 2.93, 95%CI = 1.42-6.04). Women who developed iiSBC were less likely to be > 50 years old (OR = 0.67, 95%CI = 0.49-0.90); diagnosed between 2010-2017 (OR = 0.18, 95%CI = 0.09-0.35); received lumpectomy with radiation therapy (OR = 0.54, 95%CI = 0.35-0.72); and undergone mastectomy (OR = 0.48, 95%CI = 0.32-0.72). Women who developed an icSBC were more likely to be Native Hawaiian (OR = 1.58, 95%CI = 1.06-2.35) or Filipino (OR = 1.60, 95%CI = 1.06-2.42). These women were also less likely to have been diagnosed between 2010-2017 (OR = 0.30, 95%CI = 0.17-0.53). On a subset analysis separating all patients with SBC by first course treatment type, there were no statistically significant differences for treatment type based on race/ethnicity. Conclusions: Overall, in this observational study, Native Hawaiian women, Filipino women, and younger women had increased odds of developing invasive SBC. This study highlights racial disparities in SBC development risk that was not previously appreciated among disaggregated groups of Pacific Islanders and Asian women when compared to White women. This may help oncologists understand the risk of developing SBC in these understudied populations.

scholarly journals Differences in plasma microRNA content impair microRNA-based signature for breast cancer diagnosis in cohorts recruited from heterogeneous environmental sites

2021 ◽  
Vol 11 (1) ◽  
Author(s):  
Jeanne P. Uyisenga ◽  
Ahmed Debit ◽  
Christophe Poulet ◽  
Pierre Frères ◽  
Aurélie Poncin ◽  
...  
Keyword(s):  
Breast Cancer ◽  
Cancer Diagnosis ◽  
Breast Cancer Diagnosis ◽  
Circulating Microrna ◽  
Cancer Tissue ◽  
Healthy Controls ◽  
Breast Cancers ◽  
Breast Cancer Patients ◽  
Cancer Subtypes ◽  
Plasma Microrna

AbstractCirculating microRNAs are non-invasive biomarkers that can be used for breast cancer diagnosis. However, differences in cancer tissue microRNA expression are observed in populations with different genetic/environmental backgrounds. This work aims at checking if a previously identified diagnostic circulating microRNA signature is efficient in other genetic and environmental contexts, and if a universal circulating signature might be possible. Two populations are used: women recruited in Belgium and Rwanda. Breast cancer patients and healthy controls were recruited in both populations (Belgium: 143 primary breast cancers and 136 healthy controls; Rwanda: 82 primary breast cancers and 73 healthy controls; Ntot = 434), and cohorts with matched age and cancer subtypes were compared. Plasmatic microRNA profiling was performed by RT-qPCR. Random Forest was used to (1) evaluate the performances of the previously described breast cancer diagnostic tool identified in Belgian-recruited cohorts on Rwandan-recruited cohorts and vice versa; (2) define new diagnostic signatures common to both recruitment sites; (3) define new diagnostic signatures efficient in the Rwandan population. None of the circulating microRNA signatures identified is accurate enough to be used as a diagnostic test in both populations. However, accurate circulating microRNA signatures can be found for each specific population, when taken separately.

Optimizing breast cancer diagnosis in Kenya: Importance of standardization of technical methodologies for comparative breast cancer data.

2012 ◽  
Vol 30 (27_suppl) ◽  
pp. 139-139
Author(s):  
Shahin Sayed ◽  
Zahir Moloo ◽  
Ronald Wasike ◽  
Rajendra R. Chauhan ◽  
Sudhir Vinayak ◽  
...  
Keyword(s):  
Breast Cancer ◽  
Cancer Diagnosis ◽  
Breast Cancer Diagnosis ◽  
Breast Cancers ◽  
Cancer Data ◽  
Tissue Handling ◽  
Study Results ◽  
Er Positive ◽  
The Difference ◽  
Developed World

139 Background: An analysis of 322 cases referred to Aga Khan University, Nairobi, revealed 56% estrogen receptor (ER) positive tumors and 35% prevalence of triple-negative breast cancer (TNBC). Findings were retrospective and limited by inability to control pre-analytical variables that could potentially impact results. Methods: As part of an ongoing prospective study assessing prevalence of TNBC in the three major ethnic groups in Kenya, we gathered a multidisciplinary team from 10 collaborating health facilities around Kenya for an educational workshop. The objectives were to assess baseline capabilities and pre-analytic variables at each center, identify gaps and provide hands-on training in order to ensure accuracy and validity of ER/PR/HER2 prevalence data gathered as part of the study. Results: See table. Breast cancer biopsies ranged from one to 20 per month per center. Diagnosis was predominantly by FNA and ER/PR/HER2 was not routinely performed. Buffered formalin fixative and standardized CAP reporting format was employed only at one center. A survey 3 months following the workshop demonstrated increase in diagnostic core biopsiesby 90%, and uniform use of buffered formalin fixative, and adoption of synoptic reporting. 66 prospective cases of breast cancer from the 10 institutions with patients from different ethnic backgrounds have been subsequently collected and IHC data will be presented. Conclusions: Much has been made of the difference in prevalence of TNBC in Africa as compared to North America, yet little attention has been paid to differences in diagnostic methodologies and basic tissue handling techniques that can potentially alter results. Despite limitations of resources, educational workshops make it possible to improve the practice of breast cancer diagnosis, and thereby enable accurate comparative analysis between breast cancers in the developing and the developed world. [Table: see text]

scholarly journals The COVID-19 pandemic and impact on breast cancer diagnoses: what happened in England in the first half of 2020

2020 ◽  
Author(s):  
Toral Gathani ◽  
Gill Clayton ◽  
Emma MacInnes ◽  
Kieran Horgan
Keyword(s):  
Breast Cancer ◽  
Cancer Diagnosis ◽  
Waiting Time ◽  
Breast Cancer Diagnosis ◽  
Breast Cancers ◽  
Time Data ◽  
Breast Cancer Outcomes ◽  
Time Period ◽  
Number Of Patients ◽  
The Impact

AbstractDelays in cancer diagnosis and treatment due to the COVID-19 pandemic is a widespread source of concern, but the scale of the challenge for different tumour sites is not known. Routinely collected NHS England Cancer Waiting Time data were analysed to compare activity for breast cancer in the first 6 months of 2020 compared to the same time period in 2019. The number of referrals for suspected breast cancer was 28% lower (N = 231,765 versus N = 322,994), and the number of patients who received their first treatment for a breast cancer diagnosis was 16% lower (N = 19,965 versus N = 23,881). These data suggest that the number of breast cancers diagnosed during the first half of 2020 is not as low as initially feared, and a substantial proportion of the shortfall can be explained by the suspension of routine screening in March 2020. Further work is needed to examine in detail the impact of measures to manage the COVID-19 pandemic on breast cancer outcomes.

Sign in / Sign up

Export Citation Format

Share Document