Related Experiment Video
Updated: Nov 7, 2025
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
3.1K
Introducing frequency representation into convolution neural networks for medical image segmentation via twin-Kernel
Xianlun Tang1, Jiangping Peng1, Bing Zhong1
1Chongqing University of Posts and Telecommunications, Chongqing 400065, China.
Computer Methods and Programs in Biomedicine
|April 28, 2021
Summary
This study introduces frequency representation into deep learning for medical image segmentation, significantly improving accuracy. The novel tKFC-Net model enhances segmentation performance across diverse medical imaging datasets.
Area of Science:
- Medical Image Analysis
- Deep Learning
- Computer Vision
Background:
- Deep learning excels in medical image segmentation.
- Spectral representation is underutilized in current deep learning models for this task.
Purpose of the Study:
- To integrate frequency (spectral) representation into convolutional neural networks (CNNs).
- To develop a novel model, tKFC-Net, for enhanced medical image segmentation by combining spatial and frequency domain features.
Main Methods:
- Introduced frequency representation via Fast Fourier Transform (FFT) into CNN operations (pooling, upsampling, convolution).
- Designed a new layer, twin-Kernel Fourier Convolution (t-KFC), to extract features from different frequency components.
- Integrated spectral and spatial feature extraction without altering network architecture.
Main Results:
- Achieved superior performance in medical image segmentation tasks compared to other models.
- Demonstrated high F1-Scores on diverse datasets: ISIC 2018 (0.878), DRIVE (0.8185), COVID-19-CT-Seg (0.9830), and BraTS 2019 (0.8457).
Conclusions:
- Incorporating spectral representation improves segmentation accuracy by preserving spectral features.
- The proposed tKFC-Net method is compatible with existing topology improvement techniques.
Related Concept Videos
01:11Continuous -time Fourier Transform
570
The Fourier series is instrumental in representing periodic functions, offering a powerful method to decompose such functions into a sum of sinusoids. This technique, however, necessitates modification when applied to nonperiodic functions. Consider a pulse-train waveform consisting of a series of rectangular pulses. When these pulses have a finite period, they can be accurately represented by a Fourier series. Yet, as the period approaches infinity, resulting in a single, isolated pulse, the...
570
01:15Discrete Fourier Transform
554
The Discrete Fourier Transform (DFT) is a fundamental tool in signal processing, extending the discrete-time Fourier transform by evaluating discrete signals at uniformly spaced frequency intervals. This transformation converts a finite sequence of time-domain samples into frequency components, each representing complex sinusoids ordered by frequency. The DFT translates these sequences into the frequency domain, effectively indicating the magnitude and phase of each frequency component present...
554
01:21Convergence of Fourier Series
255
The Fourier series is a powerful mathematical tool for representing periodic signals as an infinite sum of complex exponentials. In practice, this infinite series is truncated to a finite number of terms, yielding a partial sum. This truncation makes the approximation of the signal feasible but introduces certain challenges, particularly near discontinuities, known as the Gibbs phenomenon.
The Gibbs phenomenon refers to the persistent oscillations and overshoots that occur near discontinuities...
The Gibbs phenomenon refers to the persistent oscillations and overshoots that occur near discontinuities...
255
01:10Fast Fourier Transform
602
The Fast Fourier Transform (FFT) is a computational algorithm designed to compute the Discrete Fourier Transform (DFT) efficiently. By breaking down the calculations into smaller, manageable sections, the FFT significantly reduces the computational complexity involved. Direct computation of an N-point DFT requires N2 complex multiplications, whereas the FFT algorithm needs only (N/2)log2N multiplications, offering a much faster performance.
The computational efficiency of the FFT becomes...
The computational efficiency of the FFT becomes...
602
01:26Linear Approximation in Frequency Domain
224
Linear systems are characterized by two main properties: superposition and homogeneity. Superposition allows the response to multiple inputs to be the sum of the responses to each individual input. Homogeneity ensures that scaling an input by a scalar results in the response being scaled by the same scalar.
In contrast, nonlinear systems do not inherently possess these properties. However, for small deviations around an operating point, a nonlinear system can often be approximated as linear....
In contrast, nonlinear systems do not inherently possess these properties. However, for small deviations around an operating point, a nonlinear system can often be approximated as linear....
224