Database X-ray Coronary Angiograms (DCA1)
The present database consists of 130 X-ray coronary angiograms, and their corresponding ground-truth image
outlined by an expert cardiologist. Each angiogram is a 300 x 300 pixels, gray-scale image in pgm (Portable gray
map) format. The Cardiology Department of the Mexican Social Security Institute, UMAE T1-León has provided the
whole image database, and the ethics approval for its use in the present research, under reference R-2019-1001-078.
This database has been generated to assess the vessel detection and segmentation performance of our proposed method, and it
represents the first set of X-ray angiograms with ground-truth images made accessible for scientific
community for research and comparison purposes.
Please cite the following paper:
Cervantes-Sanchez, Fernando; Cruz-Aceves, Ivan; Hernandez-Aguirre, Arturo; Hernandez-Gonzalez, Martha A.; Solorio-Meza, Sergio E. 2019. "Automatic Segmentation of Coronary Arteries in X-ray Angiograms using Multiscale Analysis and Artificial Neural Networks." Appl. Sci. 9, no. 24: 5507.from the Special Issue: Signal Processing and Machine Learning for Biomedical Data (Applied Sciences,MDPI)
https://doi.org/10.3390/app9245507
Experimental results
Diagram of the proposed method, involving the steps of multiscale vessel detection, and coronary arteries segmentation by binary classification
Distribution of the generalization performance of each MLP configuration, measured in terms of the area under the ROC curve obtained using the testing set of 30 images.
Distribution of the number of architectures dominated by each configuration according to the ranking procedure using the testing set of images.
ROC curves obtained by the eleven vessel detection methods, and the proposed method, using the test set of 30 images.
