Automated detection of parenchymal changes of ischemic stroke in non-contrast computer tomography: A fuzzy approach. (August 2018)

• Record Type: Journal Article
• Authors: Davis, A.; Gordillo, N.; Montseny, E.; Aymerich, F.X.; López-Córdova, M.; Mej'ia, J.; Ortega, L.; Mederos, B.
• Is Part Of: Biomedical signal processing and control. Volume 45(2018)
• Page Start: 117
• Page End: 127
• Publication Date: 2018-08
• Subjects: Ischemic stroke -- Brain tissue segmentation -- Fuzzy logic; Signal processing -- Periodicals; Biomedical engineering -- Periodicals; Signal Processing, Computer-Assisted -- Periodicals; Image Processing, Computer-Assisted -- Periodicals; Biomedical Engineering -- Periodicals; 610.28
• DOI: 10.1016/j.bspc.2018.05.037
• Languages: English
• ISSNs: 1746-8094

Abstract

Highlights: The method identifies changes in parenchymal density in Non-Contrast Computer Tomography for the detection of early stages of ischemic stroke. Based on the analysis of cerebral symmetry, the method identifies non-pathological and cases with abnormalities. The identification of normal tissues (CSF, WM and GM), using fuzzy logic techniques, simplifies the detection of parenchymal changes of ischemic stroke. Abstract: The detection of ischemic changes is a primary task in the interpretation of brain Computer Tomography (CT) of patients suffering from neurological disorders. Although CT can easily show these lesions, their interpretation may be difficult when the lesion is not easily recognizable. The gold standard for the detection of acute stroke is highly variable and depends on the experience of physicians. This research proposes a new method of automatic detection of parenchymal changes of ischemic stroke in Non-Contrast CT. The method identifies non-pathological cases (94 cases, 40 training, 54 test) based on the analysis of cerebral symmetry. Parenchymal changes in cases with abnormalities (20 cases) are detected by means of a contralateral analysis of brain regions. In order to facilitate the evaluation of abnormal regions, non-pathological tissues in Hounsfield Units were characterized using fuzzy logic techniques. Cases of non-pathological and stroke patients were used to discard/confirm abnormality with a sensitivity (TPR) of 91% and specificity (SPC) of 100%. Abnormal regions were evaluated and the presence of parenchymal changes was detected with a TPR of 96% and SPC of 100%. The presence of parenchymal changes of ischemic stroke was detected by the identification of tissues using fuzzy logic techniques. Because of abnormal regions are identified, the expert can prioritize the examination to a previously delimited region, decreasing the diagnostic time. The identification of tissues allows a better visualization of the region to be evaluated, helping to discard or confirm a stroke.