Trans pixelate substitution scheme for denoising computed tomography images towards high diagnosis accuracy
Research output: Journal contributions › Journal articles › Research › peer-review
Standard
In: Scientific Reports, Vol. 15, No. 1, 11525, 12.2025.
Research output: Journal contributions › Journal articles › Research › peer-review
Harvard
APA
Vancouver
Bibtex
}
RIS
TY - JOUR
T1 - Trans pixelate substitution scheme for denoising computed tomography images towards high diagnosis accuracy
AU - Hu, Fengjun
AU - Gu, Hanjie
AU - Wu, Fan
AU - Lhioui, Chahira
AU - Othmen, Salwa
AU - Alfahid, Ayman
AU - Yousef, Amr
AU - Mercorelli, Paolo
N1 - Publisher Copyright: © The Author(s) 2025.
PY - 2025/12
Y1 - 2025/12
N2 - Medical images are obtained from different optical scanners and devices to provide in-body diagnosis and detection. Such scanned/acquired images are tampered with/ distorted by the unnecessary noise present in the pixel levels. A Trans-Pixelate Denoising Scheme (TPDS) is implemented to denoise these pictures to enhance the diagnosis’s precision. This scheme is specific for CT images with high noise between pixelated and non-pixelated boundaries. Therefore, the boundary detected from an input CT image is suggested for a trans-pixel substitution using a two-layer neural network. The first layer is responsible for verifying the substitution-based diagnosis accuracy, and the second is identifying trans-pixels that improve accuracy. The outcome of the neural network is used to train the noisy inputs under either of the conditions to improve the diagnosis accuracy. The Proposed TPDS improves diagnosis accuracy, precision, and pixel detection by 7.3%, 8.14%, and 13.05% under different trans-pixel rates/boundaries. Under the same variant, this scheme reduces error and detection time by 11.15% and 9.03%, respectively.
AB - Medical images are obtained from different optical scanners and devices to provide in-body diagnosis and detection. Such scanned/acquired images are tampered with/ distorted by the unnecessary noise present in the pixel levels. A Trans-Pixelate Denoising Scheme (TPDS) is implemented to denoise these pictures to enhance the diagnosis’s precision. This scheme is specific for CT images with high noise between pixelated and non-pixelated boundaries. Therefore, the boundary detected from an input CT image is suggested for a trans-pixel substitution using a two-layer neural network. The first layer is responsible for verifying the substitution-based diagnosis accuracy, and the second is identifying trans-pixels that improve accuracy. The outcome of the neural network is used to train the noisy inputs under either of the conditions to improve the diagnosis accuracy. The Proposed TPDS improves diagnosis accuracy, precision, and pixel detection by 7.3%, 8.14%, and 13.05% under different trans-pixel rates/boundaries. Under the same variant, this scheme reduces error and detection time by 11.15% and 9.03%, respectively.
KW - CT Image
KW - Denoising
KW - Medical diagnosis
KW - Neural network
KW - Pixel substitution
KW - Engineering
UR - http://www.scopus.com/inward/record.url?scp=105002655098&partnerID=8YFLogxK
U2 - 10.1038/s41598-025-95866-2
DO - 10.1038/s41598-025-95866-2
M3 - Journal articles
C2 - 40181179
AN - SCOPUS:105002655098
VL - 15
JO - Scientific Reports
JF - Scientific Reports
SN - 2045-2322
IS - 1
M1 - 11525
ER -