To search, Click below search items. |
| All Published Papers Search Service |
|
|
|
Title |
A Random Forest-based Approach for Automated Heart Diagnosis from Cardiac Tomography Data |
|
Author |
Amerah Alabrah |
| Citation |
Vol. 22 No. 4 pp. 769-777 |
|
Abstract |
Myocardial perfusion imaging or scanning (MPI or MPS) system is a technology used for nuclear medicine to demonstrate heart muscle function. A computerized diagnosis of myocardial perfusion from cardiac single photon emission computed tomography (SPECT) images is a very important diagnostic tool for cardiologists. Machine learning methods are effective tools for analyzing pathological cases in many applications of medical images. Even though there are several methods of machine learning that can be used for myocardial perfusion diagnosis, selecting a suitable diagnosis method remains a major concern. In this paper, we propose to use a random forest (RF) for analyzing the heart abnormalities from the SPECT images. The RF classifier is proposed in this paper because of its ability against over-fitting and under-fitting problems. The approach is evaluated on a SPECT Heart dataset, consisting of 267 instances. This dataset was divided into 50% instances for training and 50% instances for testing. The RF classifier is also trained on the dataset using 10-fold cross-validation method and tested on the whole dataset. The experiment shows the ability the proposed RF classifier to classify the instances of patients in the testing dataset into normal and abnormal categories better than the other machine learning classifiers in the state-of-the arts. |
|
Keywords |
Myocardial perfusion imaging, SPECT tomography images, Random forest algorithm, 10-fold cross-validation technique, machine learning methods. |
|
URL |
