Forgery detection in medical images using Complex Valued Neural Network (CVNN)
With the advent of telemedicine and telediagnosis over the internet, medical images are watermarked to ensure it integrity and authenticity. The current problem with the watermarking system used for medical images is distortion introduced during the patient data/information embedding. This factor ha...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
INSI Publications
2011
|
| Subjects: | |
| Online Access: | http://irep.iium.edu.my/6983/ http://irep.iium.edu.my/6983/ http://irep.iium.edu.my/6983/1/Forgery_Detection_in_Medical_Images_Using_Complex_Valued_Neural_Network1251-1264.pdf |
| Summary: | With the advent of telemedicine and telediagnosis over the internet, medical images are watermarked to ensure it integrity and authenticity. The current problem with the watermarking system used for medical images is distortion introduced during the patient data/information embedding. This factor has hindered proper detection and treatment. A new technique for detecting forgery in medical watermarked image using CVNN is proposed in this paper. Capabilities of Neural Networks features have been exploited using the Complex version of ANN, trained by Complex backpropagation (CBP) algorithm. This technique was used to embed and detect forge watermark in Fast Fourier Transform FFT domain. The performance of the algorithm has been evaluated using mammogram images. The imperceptibility and detection accuracy was appraised with objective performance measure; Detector response, PSNR, BER, IFM SSIM and Normalize Correlation. Results indicate that watermarked mammogram were perceptually indistinguishable from the host mammogram, hence the application of the developed CVNN-based watermarking technique in medical images can improve correct diagnoses. Ability of the algorithm to localize modification undergone makes it a unique and efficient algorithm for authentication and tamper detection as well as blind detection applications. |
|---|