Related Experiment Video
Updated: Jan 25, 2026
07:50
Author Spotlight: Integrating 2D-HPLC-MS and Molecular Networking in Natural Medicine Analysis
Published on: December 8, 2023
1.2K
Identification and Localization of Breast Tumor Components via a Convolutional Neural Network Based on High-Frequency
Jia-Qian Yao1, Wen-Wen Zhou1,2, Zhi-Fei Chai3
1Department of Medical Ultrasonics, The First Affiliated Hospital, Sun Yat-sen University, 58 Zhongshan 2nd Road, Guangzhou, 510080, China, +86-020-8776 518.
JMIR Medical Informatics
|January 23, 2026
Summary
This study developed a convolutional neural network to identify breast cancer regions in ultrasound images, achieving high accuracy by registering them with whole slide images (WSIs). The FCN-101 model showed superior performance in pinpointing cancerous areas for improved breast cancer diagnosis.
Area of Science:
- Utilizes advanced artificial intelligence (AI) and deep learning techniques for medical image analysis.
- Focuses on computational pathology and digital imaging in oncology.
- Integrates radiology (ultrasound) with pathology (histopathology) for enhanced diagnostics.
Background:
- Breast cancer biology is highly diverse, necessitating improved noninvasive diagnostic tools.
- Current methods struggle to accurately capture microscopic histopathology patterns noninvasively.
- There is an urgent need for advanced imaging techniques to complement existing diagnostic procedures.
Purpose of the Study:
- To identify cancerous regions within breast ultrasound images using convolutional neural networks (CNNs).
- To achieve accurate spatial registration between grayscale ultrasound images and whole slide images (WSIs) of biopsy specimens.
- To develop and evaluate CNN models for pixel-level classification of breast cancer in ultrasound data.
Main Methods:
- Prospective enrollment of 105 participants with Breast Imaging Reporting and Data System category 4 or 5 breast lesions.
- Collection of ultrasound images, biopsy tissue specimens, and corresponding WSIs.
- Development of CNN models (FCN-101, DeepLabV3) for identifying cancer cells, using registered ultrasound and WSI data.
- Quantitative evaluation using pixel accuracy, Dice similarity coefficient, and recall; qualitative assessment of clinical applications.
Main Results:
- The FCN-101 model demonstrated superior pixel accuracy (86.91%) and Dice similarity coefficient (77.47%) compared to DeepLabV3.
- Both models showed good performance in predicting cancerous regions, with FCN-101 excelling in cancer area prediction.
- Visualizations confirmed high consistency between identified cancerous regions in ultrasound images and WSIs.
Conclusions:
- A novel technique for spatial registration of breast WSIs and ultrasound images was successfully established.
- Advanced CNNs accurately identified and localized breast cancer regions at the pixel level in ultrasound images.
- Histopathologic WSI serves as a reliable reference standard for validating AI-driven analysis of ultrasound images.
More Related Videos
10:04A Neural Network-Based Identification of Developmentally Competent or Incompetent Mouse Fully-Grown Oocytes
Published on: March 3, 2018
7.1K
03:07Single-Port Robotic-assisted Transaxillary Breast-conserving Surgery: A Prospective, Single-arm, Non-randomized Phase IIa Clinical Trial
Published on: August 19, 2025
857
Related Concept Videos
01:17Convolution Properties II
583
The important convolution properties include width, area, differentiation, and integration properties.
The width property indicates that if the durations of input signals are T1 and T2, then the width of the output response equals the sum of both durations, irrespective of the shapes of the two functions. For instance, convolving two rectangular pulses with durations of 2 seconds and 1 second results in a function with a width of 3 seconds.
The area property asserts that the area under the...
The width property indicates that if the durations of input signals are T1 and T2, then the width of the output response equals the sum of both durations, irrespective of the shapes of the two functions. For instance, convolving two rectangular pulses with durations of 2 seconds and 1 second results in a function with a width of 3 seconds.
The area property asserts that the area under the...
583
01:20Convolution Properties I
568
Convolution computations can be simplified by utilizing their inherent properties.
The commutative property reveals that the input and the impulse response of an LTI (Linear Time-Invariant) system can be interchanged without affecting the output:
The commutative property reveals that the input and the impulse response of an LTI (Linear Time-Invariant) system can be interchanged without affecting the output:
568
02:26Protein Networks
4.5K
An organism can have thousands of different proteins, and these proteins must cooperate to ensure the health of an organism. Proteins bind to other proteins and form complexes to carry out their functions. Many proteins interact with multiple other proteins creating a complex network of protein interactions.
These interactions can be represented through maps depicting protein-protein interaction networks, represented as nodes and edges. Nodes are circles that are representative of a protein,...
These interactions can be represented through maps depicting protein-protein interaction networks, represented as nodes and edges. Nodes are circles that are representative of a protein,...
4.5K
01:21Frequency-dependent Selection
23.3K
When the fitness of a trait is influenced by how common it is (i.e., its frequency) relative to different traits within a population, this is referred to as frequency-dependent selection. Frequency-dependent selection may occur between species or within a single species. This type of selection can either be positive—with more common phenotypes having higher fitness—or negative, with rarer phenotypes conferring increased fitness.
23.3K
02:18Network Covalent Solids
16.1K
Network covalent solids contain a three-dimensional network of covalently bonded atoms as found in the crystal structures of nonmetals like diamond, graphite, silicon, and some covalent compounds, such as silicon dioxide (sand) and silicon carbide (carborundum, the abrasive on sandpaper). Many minerals have networks of covalent bonds.
To break or to melt a covalent network solid, covalent bonds must be broken. Because covalent bonds are relatively strong, covalent network solids are typically...
To break or to melt a covalent network solid, covalent bonds must be broken. Because covalent bonds are relatively strong, covalent network solids are typically...
16.1K
01:23Components of Stress
512
Stress analysis under multiple loading conditions is intricate, necessitating a comprehensive grasp of normal and shearing stresses. Consider a small cube at point O, subjected to stress on all six faces, visible or not. Normal stress components σx, σy, σz act perpendicularly to the x, y, and z axes. Shearing stress components τxy and τxz are exerted on faces perpendicular to these axes.
Interestingly, the hidden cube faces also experience these stresses, equal and...
Interestingly, the hidden cube faces also experience these stresses, equal and...
512