Images Of Breast Tissue Research Articles

MR_NET: A Method for Breast Cancer Detection and Localization from Histological Images Through Explainable Convolutional Neural Networks.

Breast cancer is the most prevalent cancer among women globally, making early and accurate detection essential for effective treatment and improved survival rates. This paper presents a method designed to detect and localize breast cancer using deep learning, specifically convolutional neural networks. The approach classifies histological images of breast tissue as either tumor-positive or tumor-negative. We utilize several deep learning models, including a custom-built CNN, EfficientNet, ResNet50, VGG-16, VGG-19, and MobileNet. Fine-tuning was also applied to VGG-16, VGG-19, and MobileNet to enhance performance. Additionally, we introduce a novel deep learning model called MR_Net, aimed at providing a more accurate network for breast cancer detection and localization, potentially assisting clinicians in making informed decisions. This model could also accelerate the diagnostic process, enabling early detection of the disease. Furthermore, we propose a method for explainable predictions by generating heatmaps that highlight the regions within tissue images that the model focuses on when predicting a label, revealing the detection of benign, atypical, and malignant tumors. We evaluate both the quantitative and qualitative performance of MR_Net and the other models, also presenting explainable results that allow visualization of the tissue areas identified by the model as relevant to the presence of breast cancer.

White Light Diffraction Phase Microscopy in Imaging of Breast and Colon Tissues.

This paper reports results obtained using white light diffraction phase microscopy (wDPM) on captured images of breast and colon tissue samples, marking a contribution to the advancement in biomedical imaging. Unlike conventional brightfield microscopy, wDPM offers the capability to capture intricate details of biological specimens with enhanced clarity and precision. It combines high resolution, enhanced contrast, and quantitative capabilities with non-invasive, label-free imaging. These features make it a useful tool for tissue imaging, providing detailed and accurate insights into tissue structure and dynamics without compromising the integrity of the samples. Our findings underscore the potential of quantitative phase imaging in histopathology, in the context of automating the process of tissue analysis and diagnosis. Of particular note are the insights gained from the reconstructed phase images, which provide physical data regarding peripheral glandular cell membranes. These observations serve to focus attention on pathologies involving the basal membrane, such as early invasive carcinoma. Through our analysis, we aim to contribute to catalyzing further advancements in tissue (breast and colon) imaging.

Deep learning-enabled high-speed, multi-parameter diffuse optical tomography.

Frequency-domain diffuse optical tomography (FD-DOT) could enhance clinical breast tumor characterization. However, conventional diffuse optical tomography (DOT) image reconstruction algorithms require case-by-case expert tuning and are too computationally intensive to provide feedback during a scan. Deep learning (DL) algorithms front-load computational and tuning costs, enabling high-speed, high-fidelity FD-DOT. We aim to demonstrate a simultaneous reconstruction of three-dimensional absorption and reduced scattering coefficients using DL-FD-DOT, with a view toward real-time imaging with a handheld probe. A DL model was trained to solve the DOT inverse problem using a realistically simulated FD-DOT dataset emulating a handheld probe for human breast imaging and tested using both synthetic and experimental data. Over a test set of 300 simulated tissue phantoms for absorption and scattering reconstructions, the DL-DOT model reduced the root mean square error by and , increased the spatial similarity by and , increased the anomaly contrast accuracy by ( ), and reduced the crosstalk by and , respectively, compared with model-based tomography. The average reconstruction time was reduced from 3.8min to 0.02s for a single reconstruction. The model was successfully verified using two tumor-emulating optical phantoms. There is clinical potential for real-time functional imaging of human breast tissue using DL and FD-DOT.

Abstract 7391: Beyond traditional staining: A virtual-IHC digital assay for HER2 score prediction from routine H&E images

Abstract Background: In breast cancer diagnostics, the accurate assessment of HER2 status is essential for treatment stratification. Conventionally, HER2 immunohistochemistry (IHC) scoring necessitates specialized staining protocols, which are resource- and time-intensive. This study presents a virtual IHC (vIHC) digital assay to predict HER2 scores directly from hematoxylin and eosin (H&E) stained images, which are routinely acquired for all specimens. This methodology aims to accelerate standard practices by reducing the need for additional IHC staining, thereby decreasing laboratory workload, costs, and turnaround time. It also presents an opportunity to screen a wider array of samples promptly, potentially identifying candidates for HER2-targeted therapy that might otherwise be delayed or overlooked. Methods: Serially sectioned breast core needle biopsy samples were stained with H&E and HER2-IHC following Leica-Bond staining protocols. Stained slides were imaged using the Hologic GeniusTM Digital Diagnostics System with Research Use Only software for Whole Slide Imaging. From this initial image dataset, 240 regions of interest (ROIs) were manually selected in order to create a dataset that uniformly covers: (i) all HER2-IHC scores (0-3), (ii) a wide range of HER2 stain intensities, and (iii) expression patterns. The IHC scores were assigned to each selected ROI by a pathology expert, following the scoring recommendations of ASCO/CAP American Society of Clinical Oncology and College of American Pathologists, with the caveat that ductal carcinoma in situ was also given a score for the purpose of training the model. These ROIs were then extracted from the images of the serially sectioned H&E-stained slides using Reveal Bio’s proprietary co-registration digital assay. A predictive digital assay was developed at Reveal Biosciences to classify the 240 H&E-stained breast tissue ROIs into four HER2 IHC score categories: 0, 1, 2, and 3. For model evaluation, cross-validation was performed 10 times by randomly splitting the dataset of H&E ROIs into 80% for training MIL and 20% for testing the model, followed by averaging the performance metrics across the 10 dataset splits. The performance metrics included classification accuracy, precision, recall, and F1 scores for each HER2 score category. Results: The vIHC digital assay for predicting the HER2 scores directly from H&E stained breast tissue images achieved the average classification accuracy of 93% across the four HER2 score categories on the test set of ROIs. This performance demonstrates that the model is capable of successfully distinguishing between the HER2 score categories directly from H&E images without the need for IHC staining. The results suggest that the vIHC digital assay could be a reliable and cost-effective tool for preliminary breast cancer screening, providing a valuable adjunct to traditional histopathological analysis. Citation Format: Misagh Naderi, Nikola Mitic, Nikola Spasic, Djordje Cikic, Kristen Ruf, Nikolaus Wagner, Igor Mihajlovic, Sinisa Todorovic, Ray Jenoski, Sid Mayer, Michael Quick, Claire Weston. Beyond traditional staining: A virtual-IHC digital assay for HER2 score prediction from routine H&E images [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 7391.

Breast tissue imaging atlas using ultra-fast confocal microscopy to identify cancer lesions.

New generation ultra-fast fluorescence confocal microscopy (UFCM) allows to image histological architecture of fresh breast tissue and may be used for ex vivo intraoperative analysis for margin status. The criteria to identify breast tumoral and non-tumoral tissues in UFCM images are still objects of investigation. The objective of the study was to create an atlas of ex vivo UFCM images of breast tissues and breast carcinomas based on the first extensive collection of large field-of-view UFCM breast images. One hundred sixty patients who underwent conserving surgery for breast cancer were included. Their fresh surgical specimens were sliced, stained with acridine orange, and imaged at high resolution with large-field-of-view UFCM. The resulting images were digitally false colored to resemble frozen sections. Each UFCM image was correlated with the corresponding definitive histology. Representative images of normal tissue, inflammation, benign lesions, invasive carcinoma (IC), and ductal carcinoma in situ (DCIS) were collected. A total of 320 large-field images were recorded from 58 IC of no special type, 44 invasive lobular carcinomas, 1 invasive mucinous carcinoma, 47 DCIS, 2 lobular carcinomas in situ, and 8 specimens without cancer. Representative images of the main components of the normal breast and the main types of ICs and DCIS were annotated to establish an UFCM atlas. UFCM enables the imaging of the fresh breast tissue sections. Main morphological criteria defined in traditional histopathology such as tissue architecture and cell features can be applied to describe UFCM images content. The generated atlas of the main normal or tumoral tissue features will support the adoption of this optical technology for the intraoperative examination of breast specimens in clinical practice as it can be used to train physicians on UFCM images and develop artificial intelligence algorithms. Further studies are needed to document rare breast lesions.

Improving lesion detection in mammograms by leveraging a Cycle-GAN-based lesion remover

BackgroundThe wide heterogeneity in the appearance of breast lesions and normal breast structures can confuse computerized detection algorithms. Our purpose was therefore to develop a Lesion Highlighter (LH) that can improve the performance of computer-aided detection algorithms for detecting breast cancer on screening mammograms.MethodsWe hypothesized that a Cycle-GAN based Lesion Remover (LR) could act as an LH, which can improve the performance of lesion detection algorithms. We used 10,310 screening mammograms from 4,832 women that included 4,942 recalled lesions (BI-RADS 0) and 5,368 normal results (BI-RADS 1). We divided the dataset into Train:Validate:Test folds with the ratios of 0.64:0.16:0.2. We segmented image patches (400 × 400 pixels) from either lesions marked by MQSA radiologists or normal tissue in mammograms. We trained a Cycle-GAN to develop two GANs, where each GAN transferred the style of one image to another. We refer to the GAN transferring the style of a lesion to normal breast tissue as the LR. We then highlighted the lesion by color-fusing the mammogram after applying the LR to its original. Using ResNet18, DenseNet201, EfficientNetV2, and Vision Transformer as backbone architectures, we trained three deep networks for each architecture, one trained on lesion highlighted mammograms (Highlighted), another trained on the original mammograms (Baseline), and Highlighted and Baseline combined (Combined). We conducted ROC analysis for the three versions of each deep network on the test set.ResultsThe Combined version of all networks achieved AUCs ranging from 0.963 to 0.974 for identifying the image with a recalled lesion from a normal breast tissue image, which was statistically improved (p-value < 0.001) over their Baseline versions with AUCs that ranged from 0.914 to 0.967.ConclusionsOur results showed that a Cycle-GAN based LR is effective for enhancing lesion conspicuity and this can improve the performance of a detection algorithm.

Usporedba između magnetne rezonancije dojki i kontrastne mamografije

Introduction: Imaging modalities such as contrast-enhanced mammography (CEM) and magnetic resonance imaging (MRI) are used as valuable tools for understanding breast pathology. They enable early detection of breast cancer, provide a precise insi-ght into theslesions, and facilitate the monitoring of responses to treatment. Modern CEM improves the detection of breast abnor-malitiesfusing contrast agents and the combination of higher energy levels in a single examination. MRI uses magnetic fields and coils to produce detailed images of breast tissue in high-risk cases, and to evaluate suspicious findings with other imaging methods using different sequences.Aim of the paper: The aim of the paper is to introduce CEM and MRI separately, while also identifying the strengths and weaknesses of each technique. This will facilitate their direct comparison.Discussion: CEM excels in detecting multifocal or contralateral lesions, assessing response to treatment, and identifying microcalcifications. Combining the power of mammography and contrast, CEM has the disadvantages of higher radiation dose, artifacts, and possible reactions to the use of iodinated contrast agent. In contrast, MRI provides detailed images, better lesion identification and treatment assessment and possesses exceptional contrast for breast soft tissue evaluation. It has proven invalua-ble for assessing implant condition and detecting lesions not seen on mammography images. This is crucial for high-risk cases and hiddenstumours. However, MRI has limitations such as artifacts and cost. CEM is a viable alternative to MRI, with high sensitivity.Conclusion: The combined use of CEM and MRI has the potential to transform breast cancer treatment by improving diagnostic accuracy and enabling more personalized treatment approaches, especially in challenging cases, and among high-risk populations with dense breast tissue. With continued technological advances and standardization efforts, CEM and MRI will continue to play a critical role in early cancer detection, lesion characterization, and treatment monitoring, ultimately improving patient outcomes in the field of breast cancer.

In vivo optical coherence tomography attenuation imaging of the breast surgical cavity using a handheld probe

Approximately 20–30% of breast-conserving surgery (BCS) patients require follow-up re-excision surgery due to inadequate cancer resection. To aid cancer detection and resection, we present a handheld optical coherence tomography (OCT) probe and attenuation imaging method to enable in vivo imaging of the surgical cavity during BCS. The attenuation imaging method incorporates three-dimensional correction of system factors in the probe, namely, confocal function and sensitivity roll-off function, demonstrating accurate and repeatable attenuation imaging of phantoms, validated using the Mie scattering model. To assess the imaging performance of the probe prior to in vivo imaging, ex vivo attenuation imaging of breast specimens from eight mastectomy patients was performed and the results compared to those acquired using a benchtop imaging system, showing consistent tissue contrast in attenuation images. In vivo imaging of breast tissue in the surgical cavity of seven patients was subsequently performed during BCS. The results demonstrate the potential of attenuation imaging to delineate cancer in vivo in the surgical cavity. Attenuation imaging provides an additional tissue contrast mechanism to complement OCT imaging, potentially contributing to more complete identification and resection of cancer during BCS.

Detection of Breast Cancer from Mammogram using Convolution Neural Network

Abstract: Breast cancer is the leading cause of cancer among women. In the last few years, the number of cases of breast cancer has skyrocketed among younger women. Detection of breast cancer during the initial stages can greatly reduce the risk of fatality. Mammograms which are X-ray images of breast tissues are used extensively by doctors to determine the early onset of breast cancer. However, due to human error, a lack of resources and knowledge can result in inaccurate predictions which can prove fatal. The power of Artificial Intelligence to process and predict images has greatly increased in the past few years. Many modern medical devices utilize the power of sophisticated AI algorithms to aid in detecting and predicting the early onset of breast cancer. In this paper, we demonstrate how we can utilize a convolution neural network to predict the early onset of breast cancer and help solve the issue of inaccurate detection of cancer cells from mammogram images.

From bedside to portable and wearable: development of a conformable ultrasound patch for deep breast tissue imaging.

A breakthrough study from Du etal. has developed a wearable, ultrasound imaging patch for standardized and reproducible breast tissue imaging. The technology utilizes a honeycomb patch design to facilitate guided movement of the ultrasound array, enabling comprehensive, multiangle breast imaging. The system was validated in vitro and in vivo with a single human subject and has the potential for early-stage breast cancer detection. This study addressed the current limitations of wearable ultrasound technologies, including imaging over large, curvilinear organs and integration of superior piezoelectric materials for high-performance ultrasound arrays. The transition of ultrasound from the bedside to portable and wearable devices will pave the way for integration with big data collection, such as artificial intelligence (AI)-based diagnosis and personalized ultrasonographic profile generation, for rapid and objective measurements. This advancement is especially important in the context of breast cancer, where early diagnosis and assessment of medical therapy responses are paramount to patient care.

Minimum resolution requirements of digital pathology images for accurate classification.

Digitization of pathology has been proposed as an essential mitigation strategy for the severe staffing crisis facing most pathology departments. Despite its benefits, several barriers have prevented widespread adoption of digital workflows, including cost and pathologist reluctance due to subjective image quality concerns. In this work, we quantitatively determine the minimum image quality requirements for binary classification of histopathology images of breast tissue in terms of spatial and sampling resolution. We train an ensemble of deep learning classifier models on publicly available datasets to obtain a baseline accuracy and computationally degrade these images according to our derived theoretical model to identify the minimum resolution necessary for acceptable diagnostic accuracy. Our results show that images can be degraded significantly below the resolution of most commercial whole-slide imaging systems while maintaining reasonable accuracy, demonstrating that macroscopic features are sufficient for binary classification of stained breast tissue. A rapid low-cost imaging system capable of identifying healthy tissue not requiring human assessment could serve as a triage system for reducing caseloads and alleviating the significant strain on the current workforce.

A cadaveric breast cancer tissue phantom for phase-contrast X-ray imaging applications.

As mammography X-ray imaging technologies advance and provide elevated contrast in soft tissues, a need has developed for reliable imaging phantoms for use in system design and component calibration. In advanced imaging modalities such as refraction-based methods, it is critical that developed phantoms capture the biological details seen in clinical precancerous and cancerous cases while minimizing artifacts that may be caused due to phantom production. This work presents the fabrication of a breast tissue imaging phantom from cadaveric breast tissue suitable for use in both transmission and refraction-enhanced imaging systems. Human cancer cell tumors were grown orthotopically in nude athymic mice and implanted into the fixed tissue while maintaining the native tumor/adipose tissue interface. The resulting human-murine tissue hybrid phantom was mounted on a clear acrylic housing for absorption and refraction X-ray imaging. Digital breast tomosynthesis was also performed. Both attenuation-based imaging and refraction-based imaging of the phantom are presented to confirm the suitability of this phantom's use in both imaging modalities.

Computer-aided diagnosis using white shark optimizer with attention-based deep learning for breast cancer classification

Breast cancer (BC) is categorized as the most widespread cancer among women throughout the world. The earlier analysis of BC assists to increase the survival rate of the disease. BC diagnosis on histopathology images (HIS) is a tedious process that includes recognizing cancerous regions within the microscopic image of breast tissue. There are various methods to discovering BC on HSI, namely deep learning (DL) based methods, classical image processing techniques, and machine learning (ML) based methods. The major problems in BC diagnosis on HSI are the larger size of images and the high degree of variability in the appearance of tumorous regions. With this motivation, this study develops a computer-aided diagnosis using a white shark optimizer with attention-based deep learning for the breast cancer classification (WSO-ABDLBCC) model. The presented WSO-ABDLBCC technique performs accurate classification the breast cancer using DL techniques. In the WSO-ABDLBCC technique, the Guided filtering (GF) based noise removal is applied to improve the image quality. Next, the Faster SqueezeNet model with WSO-based hyperparameter tuning performs the feature vector generation process. Finally, the classification of histopathological images takes place using attention-based bidirectional long short-term memory (ABiLSTM). A detailed experimental validation of the WSO-ABDLBCC occurs utilizing the benchmark Breakhis database. The proposed model achieved an accuracy of 95.2%. The experimental outcomes portrayed that the WSO-ABDLBCC technique accomplishes improved performance compared to other existing models.

Conformable ultrasound breast patch for deep tissue scanning and imaging.

Ultrasound is widely used for tissue imaging such as breast cancer diagnosis; however, fundamental challenges limit its integration with wearable technologies, namely, imaging over large-area curvilinear organs. We introduced a wearable, conformable ultrasound breast patch (cUSBr-Patch) that enables standardized and reproducible image acquisition over the entire breast with less reliance on operator training and applied transducer compression. A nature-inspired honeycomb-shaped patch combined with a phased array is guided by an easy-to-operate tracker that provides for large-area, deep scanning, and multiangle breast imaging capability. The in vitro studies and clinical trials reveal that the array using a piezoelectric crystal [Yb/Bi-Pb(In1/2Nb1/2)O3-Pb(Mg1/3Nb2/3)O3-PbTiO3] (Yb/Bi-PIN-PMN-PT) exhibits a sufficient contrast resolution (~3 dB) and axial/lateral resolutions of 0.25/1.0 mm at 30 mm depth, allowing the observation of small cysts (~0.3 cm) in the breast. This research develops a first-of-its-kind ultrasound technology for breast tissue scanning and imaging that offers a noninvasive method for tracking real-time dynamic changes of soft tissue.

Performance Analysis of an Aperture-Coupled THz Antenna for Diagnosing Breast Cancer.

The most important technique for exposing early-stage breast cancer is terahertz imaging. It aids in lowering the number of breast cancer-related fatalities and enhancing the quality of life. An essential component of developing the THz imaging system for high-quality photos is choosing the right sensor. In this article, a wideband antenna for microwave imaging of breast tissue with an operating frequency of 30 GHz (107 GHz to 137 GHz) is constructed and analyzed. An aperture-coupled antenna with an optimized ground aperture is proposed and analyzed, which made it possible to obtain better and consistent impedance matching in the wideband spectrum. The variation of backscattered signal energy in body tissue is assessed with healthy breast tissue and in the presence of malignant cells. A significant difference in energy scattering is observed for both situations. The suggested antenna's linear and stable time domain characteristics make it an appropriate component for THz imaging technology.

Design of a Low-Cost Diffuse Optical Mammography System for Biomedical Image Processing in Breast Cancer Diagnosis.

Worldwide, breast cancer is the most common type of cancer that mainly affects women. Several diagnosis techniques based on optical instrumentation and image analysis have been developed, and these are commonly used in conjunction with conventional diagnostic devices such as mammographs, ultrasound, and magnetic resonance imaging of the breast. The cost of using these instruments is increasing, and developing countries, whose deaths indices due to breast cancer are high, cannot access conventional diagnostic methods and have even less access to newer techniques. Other studies, based on the analysis of images acquired by traditional methods, require high resolutions and knowledge of the origin of the captures in order to avoid errors. For this reason, the design of a low-cost diffuse optical mammography system for biomedical image processing in breast cancer diagnosis is presented. The system combines the acquisition of breast tissue photographs, diffuse optical reflectance (as a biophotonics technique), and the processing of digital images for the study and diagnosis of breast cancer. The system was developed in the form of a medical examination table with a 638 nm red-light source, using light-emitted diode technology (LED) and a low-cost web camera for the acquisition of breast tissue images. The system is automatic, and its control, through a graphical user interface (GUI), saves costs and allows for the subsequent analysis of images using a digital image-processing algorithm. The results obtained allow for the possibility of planning in vivo measurements. In addition, the acquisition of images every 30° around the breast tissue could be used in future research in order to perform a three-dimensional (3D) reconstruction and an analysis of the captures through deep learning techniques. These could be combined with virtual, augmented, or mixed reality environments to predict the position of tumors, increase the likelihood of a correct medical diagnosis, and develop a training system for specialists. Furthermore, the system allows for the possibility to develop analysis of optical characterization for new phantom studies in breast cancer diagnosis through bioimaging techniques.

407 Moving from observational studies to a clinical trial: the impact of obesity and surgical weight loss on breast imaging, tissue, and cancer screening experience: the B-BRITE study

OBJECTIVES/GOALS: Obesity is associated with increased incidence of breast cancer (BC), yet is not included in many lifetime-risk calculators. Obesity may impact breast cancer screening sensitivity. Retrospective studies show that bariatric surgery is associated with a lower risk of BC, but the effects of surgical weight loss on breast tissue are poorly understood. METHODS/STUDY POPULATION: We proposed a mixed-methods before and after study design to investigate the effects of surgical weight loss on breast tissue via pre- and post-weight loss breast tissue biopsies and imaging. In addition, we aimed to better understand barriers to BC screening for patients with obesity by conducting qualitative interviews. With institutional review board approval, we have begun recruiting 14 cisgender women who plan to undergo Roux-en-Y gastric bypass or sleeve gastrectomy. Participants must be at least 40 years old, with no prior history of breast biopsies or breast cancer and will undergo comprehensive breast cancer screening including mammography with quantitative density assessment, breast MRI, as well as breast core biopsies. RESULTS/ANTICIPATED RESULTS: We hypothesize that obesity and its associated metabolic changes lead to altered breast stroma, including increased inflammation, and tissue stiffness, with subsequent risk of carcinogenesis. If true, we expect to find obese women will have measurably increased inflammatory markers in their breast tissue, which are reduced after bariatric surgery. We expect that change in mammographic density may correlate with fibroglandular volume change on MRI; there are little data on change in background parenchymal enhancement in the setting of obesity and weight change and quantifying this will provide preliminary data for future work. Last, we expect that undergoing BC screening will be easier for patients after weight loss due to constraints of imaging equipment and potential bias in the screening process. DISCUSSION/SIGNIFICANCE: Screening for BC is paramount to improving outcomes yet people with obesity are screened less with worse outcomes. Studying the effects of weight loss on the breast may improve interpretation of breast imaging in the setting of obesity and identify markers of risk. Understanding barriers to screening may help us develop strategies to improve screening.

Image Translation of Breast Ultrasound to Pseudo Anatomical Display by CycleGAN.

Ultrasound imaging is cost effective, radiation-free, portable, and implemented routinely in clinical procedures. Nonetheless, image quality is characterized by a granulated appearance, a poor SNR, and speckle noise. Specific for breast tumors, the margins are commonly blurred and indistinct. Thus, there is a need for improving ultrasound image quality. We hypothesize that this can be achieved by translation into a more realistic display which mimics a pseudo anatomical cut through the tissue, using a cycle generative adversarial network (CycleGAN). In order to train CycleGAN for this translation, two datasets were used, "Breast Ultrasound Images" (BUSI) and a set of optical images of poultry breast tissues. The generated pseudo anatomical images provide improved visual discrimination of the lesions through clearer border definition and pronounced contrast. In order to evaluate the preservation of the anatomical features, the lesions in both datasets were segmented and compared. This comparison yielded median dice scores of 0.91 and 0.70; median center errors of 0.58% and 3.27%; and median area errors of 0.40% and 4.34% for the benign and malignancies, respectively. In conclusion, generated pseudo anatomical images provide a more intuitive display, enhance tissue anatomy, and preserve tumor geometry; and can potentially improve diagnoses and clinical outcomes.

CNN-Wavelet scattering textural feature fusion for classifying breast tissue in mammograms

Visual interpretation from radiologists employs computer-aided diagnosis (CAD) to make clinical diagnoses by analyzing breast tissue images and assessing their texture. Aside from needing more training images for Convolutional Neural Network (CNN), selecting textural feature input to the deep learning-based CAD is challenging to control since the entire feature map is extracted automatically and only considers spatial domain analysis. Hence, integrating spatial and frequency information from the images requires complex feature representation in a classification problem. Wavelet representation could improve the feature descriptors in acquiring information from the spatial-frequency analysis. However, wavelet transforms’ shift insensitivity may affect class feature representation by suppressing high-frequency information. This study uses CNN, and wavelet scattering (WS) features to classify fatty and fibroglandular tissue with benign and malignant masses from digital mammograms to overcome CNN overfitting based on a limited dataset. Scattered WS coefficients formed from layers of wavelet dilations and averaging are utilized to retain the loss of high-frequency signal from the images, while simultaneously, the added CNN features improve sparse image representation resulting from convoluted spatial information of the images. Finally, the model is cascaded with an ensemble classifier for classifying fatty and fibroglandular tissue and mass images, n = 112, from the INbreast mammogram dataset. The best performance of mass and breast tissue classification models reach 98.0 % and 99.3 % on 10-fold cross-validation accuracy. By combining textural features from WS and deep learning model architecture, the model may adopt a training system with fewer data without augmentation to achieve good performance and low complexity costs.

Hybrid Models for Breast Cancer Detection via Transfer Learning Technique

Currently, breast cancer has been a major cause of deaths in women worldwide and the World Health Organization (WHO) has confirmed this. The severity of this disease can be minimized to the large extend, if it is diagnosed properly at an early stage of the disease. Therefore, the proper treatment of a patient having cancer can be processed in better way, if it can be diagnosed properly as early as possible using the better algorithms. Moreover, it has been currently observed that the deep neural networks have delivered remarkable performance for detecting cancer in histopathological images of breast tissues. To address the above said issues, this paper presents a hybrid model using the transfer learning to study the histopathological images, which help in detection and rectification of the disease at a low cost. Extensive dataset experiments were carried out to validate the suggested hybrid model in this paper. The experimental results show that the proposed model outperformed the baseline methods, with F-scores of 0.81 for DenseNet + Logistic Regression hybrid model, (F-score: 0.73) for Visual Geometry Group (VGG) + Logistic Regression hybrid model, (F-score: 0.74) for VGG + Random Forest, (F-score: 0.79) for DenseNet + Random Forest, and (F-score: 0.79) for VGG + Densenet + Logistic Regression hybrid model on the dataset of histopathological images.