U-Net Convolutional Neural Networks for breast IR imaging segmentation on frontal and lateral view

Carvalho, Elisson Carlos de; Coelho, Alessandra Martins; Conci, Aura; Baffa, Matheus de Freitas Oliveira

doi:10.1080/21681163.2022.2040053

Cited by 9 publications

(4 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…To enhance the dataset, we focused on segmenting the region of interest (ROI) to accurately extract key anatomical features, thereby improving the accuracy of our analyses (Figure 3). For segmentation, we adapted a U-Net based method originally developed by Carvalho et al 15 While the original model used a 64 × 64 input size for angular images, we found this insufficient for our needs. Consequently, we employed a modified U-Net model with an increased input size of 256 × 256, allowing for more detailed extraction of segmentation masks.…”

Section: Image Preprocessingmentioning

confidence: 99%

Breast cancer diagnostics using angular view infrared images

Samantaray,

Nair,

de Freitas Oliveira Baffa

et al. 2024

Medical Imaging 2024: Imaging Informatics for Healthcare, Research, and Applications

View full text Add to dashboard Cite

Worldwide, breast cancer presents a significant health challenge, necessitating innovative techniques for early detection and prognosis. Although mammography is the established screening method, it has drawbacks, including radiation exposure and high costs. Recent studies have explored the application of machine learning to frontal infrared images for breast cancer detection. However, the potential of infrared imaging from angular views has not been thoroughly explored. In this paper, we investigate, develop, and evaluate classification models for breast cancer diagnosis using lateral and oblique infrared images. Our approach incorporates radiomic features and Convolutional Neural Networks along with various feature fusion techniques to train deep neural networks. The primary objective is to determine the suitability of angular views for breast cancer detection, identify the most effective view, and assess its impact on classification accuracy. Utilizing the publicly available Database for Mastology Research with Infrared Images (DMR-IR), we apply an image processing pipeline for image improvement and segmentation. Additionally, we extract features using two strategies: radiomic features and convolutional neural network features. Subsequently, we conduct a series of k-fold cross-validation experiments to determine whether the features and feature fusion techniques are effective. Our findings indicate that oblique images, particularly when combined with DenseNet features, demonstrate superior performance. We achieved an average accuracy of 97.74%, specificity of 95.25%, and an F1 score of 98.24%. This study contributes to the advancement of machine learning in early breast cancer detection and underscores the significant potential of angular views in thermal infrared imaging, leading to improved diagnostic outcomes for patients worldwide.

show abstract

Section: Image Preprocessingmentioning

confidence: 99%

Breast cancer diagnostics using angular view infrared images

Samantaray,

Nair,

de Freitas Oliveira Baffa

et al. 2024

Medical Imaging 2024: Imaging Informatics for Healthcare, Research, and Applications

View full text Add to dashboard Cite

show abstract

“…Their IoU (intersection over union) value was 94.38% with average precision of 98.86%, and a recall of 98.46%. De Carvalho et al [15] applied U-Net to segment the breast region using thermal images taken from the front and lateral. Two different U-net architectures were trained and tested using 283 frontal view and 332 lateral view thermograms from the DMR-IR dataset.…”

Section: Segmentation Using Deep Learning Techniquesmentioning

confidence: 99%

“…(2) U-Net and other modified U-Net architectures demonstrate strong performance in thermal image segmentation. However, it was observed in [15] that the segmentation performance was affected negatively by lateral thermograms due to variations in orientation and limited information in lateral view, and demands separate analyses from frontal thermograms.…”

Section: Classificationmentioning

confidence: 99%

Dual-Tree Complex Wavelet Pooling and Attention based Modified U-Net Architecture for Automated Breast Thermogram Segmentation and Classification

Garia,

Muthusamy

2024

Preprint

View full text Add to dashboard Cite

Thermography is a non-invasive and non-contact method for detecting cancer in their initial stages by examining the temperature variation between both breasts. Resizing, ROI (region of interest) segmentation, and augmentation are frequently used methods for pre-processing breast thermograms. In this study, a modified U-Net architecture (DTCWAU-Net) that uses Dual-Tree Complex Wavelet Transform (DTCWT) and Attention gate for breast thermal image segmentation for frontal and lateral view thermograms, aiming to outline ROI for potential tumor detection was proposed. The proposed approach achieved an average Dice coefficient of 93.03% and a sensitivity of 94.82%, showcasing its potential for accurate breast thermogram segmentation. The automated segmentation of breast thermograms into categories as healthy or cancerous was achieved by texture and histogram-based feature and deep feature extraction from these segmented thermograms, feature selection using Neighborhood Component Analysis (NCA), and applying machine learning classifiers. When compared to other state-of the art approaches for detecting breast cancer using thermogram, the proposed methodology showed higher accuracy. Simulation results clearly expounds that the proposed method can be used in breast cancer screening, facilitating early detection, and enhancing treatment outcomes.

show abstract

“…We compare the performance of different networks using two indicators: Acc and IoU 17 . Acc, also known as accuracy, refers to the proportion of correctly predicted pixels in total pixels.…”

Section: Evaluation Metricsmentioning

confidence: 99%

Multi-stage context-guided semantic segmentation in construction sites

Hou,

Yang,

et al. 2023

Fifth International Conference on Artificial Intelligence and Computer Science (AICS 2023)

View full text Add to dashboard Cite

Semantic segmentation refers to the segmentation of different objects within images to obtain a semantic interpretation of each pixel position based on semantic information. With the development of artificial intelligence and computer vision, intelligent visual surveillance solutions for construction sites are playing an increasingly prominent role in site monitoring. For this purpose, we propose a semantic segmentation network for intelligent scene perception in construction sites. The features are extracted with the residual block and channel-wise feature aggregation module, thus obtaining multi-scale features with rich information. Then, high-level futures are further enhanced with long-range contextual information, which is adopted as guidance for the decoding process. Experimental results show that the proposed network can efficiently process construction site scenes in real-time and has important practical applications.

show abstract

U-Net Convolutional Neural Networks for breast IR imaging segmentation on frontal and lateral view

Cited by 9 publications

References 12 publications

Breast cancer diagnostics using angular view infrared images

Breast cancer diagnostics using angular view infrared images

Dual-Tree Complex Wavelet Pooling and Attention based Modified U-Net Architecture for Automated Breast Thermogram Segmentation and Classification

Multi-stage context-guided semantic segmentation in construction sites

Contact Info

Product

Resources

About