Report of RSNA2006

Abstract

In the diagnosis of hyperacute stroke, an early CT sign such as the loss of gray-white matter interface may be difficult to detect within the first hours of the onset of symptoms because of the presence of quantum noise on CT images. We present a method for improving the detection of early CT sign, and to evaluate whether our method can help radiologists detect the early CT signs at emergency CT scan. Our method provides an adaptive partial median filter (APMF), which can reduce local noise without blurring of anatomical structure using variable filter shape and size according to the pixel value distribution of object around a center pixel. The APMF can enhance the loss of gray-white matter interface due to hyperacute stroke. Twenty-six patients with early CT signs at the lentiform nucleus and/or the cortical ribbon and 49 control subjects were included in an observer study. The CT images of 26 patients with acute (<5 hours) middle cerebral artery territory infarction were proved with follow-up CT. The APMF was applied to all the CT images. Four radiologists, without and with applying the APMF, indicated their confidence level regarding the presence (or absence) of the early CT signs on each CT images. Receiver operating characteristic (ROC) curves were calculated to evaluate the observers' performance. A 78% noise reduction with the APMF was obtained from simulation. The average area under the ROC curve (Az) was improved from 0.868 to 0.924 for all radiologists by applying the APMF to the original images. The difference in Az values with and without the APMF was statistically significant with a P value of .002 for all radiologists. Our proposed APMF can improve the visibility of gray-white matter interface. As a result, the APMF can help radiologists detect the early CT signs at emergency CT scan.