An Enhanced Convolutional Neural Network Based on L1 Regularization for Segmentation of Coronary Arteries in X Ray Angiograms

Wahid, Fazli, Ma, YingLiang ORCID:, Sheikh, Ramla and Aamir, Muhammad (2024) An Enhanced Convolutional Neural Network Based on L1 Regularization for Segmentation of Coronary Arteries in X Ray Angiograms. In: 5th International Conference on Advancements in Computational Sciences, 2024-02-19.

Full text not available from this repository. (Request a copy)


The automatic segmentation of coronary arteries in X-Ray angiograms has a significant role in identification and detection of various abnormalities in vessels which can be adequately helpful in computer aided diagnosis (CAD) for supporting in analyzing the coronary heart diseases. Previously many machine learning (ML) algorithms and deep learning (DL) models have been applied for this purpose but all of them are suffered from some serious drawbacks when applied in their standardized working formats which leads to low performance of these models. Therefore, in this work an enhanced convolutional neural network named eCNN which is based on L1 regularization has been applied for automatic segmentation of coronary arteries. It has been observed that incorporating L1 regularization with different coefficient values has distinct effects on the working mechanism of CNN architecture resulting in improving its performance. In the proposed method, firstly, Gaussian Matched Filters (GMF) have been applied to enhance the quality of X-Ray angiograms. In segmentation stages, the eCNN has been applied for the segmentation. In this stage, dual L1 regularization (coefficient: 0.001 for convolutional and 0.01 for dense layers) improves model interpretability and generalization, facilitating effective segmentation of arteries. Finally, standard performance evaluation parameters have been used to evaluate the performance of the proposed model and compared with state-of-the-art techniques.

Item Type: Conference or Workshop Item (Paper)
Additional Information: Funding information: This work was supported by EPSRC UK (EP/X023826/1).
Faculty \ School: Faculty of Science > School of Computing Sciences
UEA Research Groups: Faculty of Science > Research Groups > Norwich Epidemiology Centre
Faculty of Medicine and Health Sciences > Research Groups > Norwich Epidemiology Centre
Depositing User: LivePure Connector
Date Deposited: 01 Feb 2024 03:21
Last Modified: 22 May 2024 13:30
DOI: 10.1109/ICACS60934.2024.10473261

Actions (login required)

View Item View Item