Yoshikazu Uchiyama, Presenter: Nothing to Disclose

Chisako Muramatsu PhD, Abstract Co-Author: Nothing to Disclose

Takeshi Hara PhD, Abstract Co-Author: Nothing to Disclose

Junji Shiraishi, Abstract Co-Author: Research Grant, Konica Minolta Group

Hiroshi Fujita PhD, Abstract Co-Author: Nothing to Disclose

The detection of asymptomatic lacunar infarcts is important because their presence indicate an increased risk of severe cerebral infarction. However accurate identification of lacunar infarcts on MR images is often hard for radiologists because of the difficulty in distinguishing lacunar infarcts and enlarged perivascular spaces. Therefore, we have been developing CAD schemes for the detection of lacunar infarcts. Our previous method was applied to 132 patient cases and the sensitivity of 96.8% with 0.71 FP per slice was obtained. However, further reduction of FPs was remained as an issue to be solved for the clinical application. The purpose of this study is to improve our CAD scheme for detection of lacunar infarcts by use of a template matching technique on eigen space.

Our database consisted of T1- and T2-weighted images obtained from 132 patients, which included 93 lacunar infarcts. The images were acquired on a 1.5 T MR scanner. Using our previous method, almost all lacunar infarcts were detected accurately. However, 753 FPs were obtained. For reduction of these FPs, we employed template matching technique. Although the conventional template matching is useful for the reduction of FPs, it has 2 pitfalls. (1) It needs to keep a large number of templates for improving the detection performance. (2) The calculation of the cross-correlation coefficient with these templates takes much time. To solve these problems, we used template matching in the lower dimension space made by a principal component analysis.

Our computerized method was evaluated by using 2-hold cross validation. By using the proposed method, 34.1 % of FPs were eliminated. The final performance indicated that the sensitivity of the detection of lacunar infarcts was 96.8% (90/93) with 0.47 FP per slice.

By using the proposed method, the quantity of the template became 1.9 % of conventional template matching. The computation time could be shortened largely. Moreover, the number of FPs was decreased from 0.71 to 0.47 per slice while keeping the same sensitivity of 96.8 %. Therefore, the proposed method is useful technique in the CAD scheme for the detection of lacunar infarcts on MR images.

Our CAD scheme can output locations of lacunar infarcts with a small number of FPs, and would be useful in assisting radiologists for identifying lacunar infarcts on MR images.

Uchiyama, Y, Muramatsu, C, Hara, T, Shiraishi, J, Fujita, H, Improvement of CAD Scheme for Detection of Silent Lacunar Infarcts on MR Images by Using Template Matching on Eigen Space  .  Radiological Society of North America 2013 Scientific Assembly and Annual Meeting, December 1 - December 6, 2013 ,Chicago IL. http://archive.rsna.org/2013/13021584.html