2008 | OriginalPaper | Chapter
CAD on Brain, Fundus, and Breast Images
Authors : Hiroshi Fujita, Yoshikazu Uchiyama, Toshiaki Nakagawa, Daisuke Fukuoka, Yuji Hatanaka, Takeshi Hara, Yoshinori Hayashi, Yuji Ikedo, Gobert N. Lee, Xin Gao, Xiangrong Zhou
Published in: Medical Imaging and Informatics
Publisher: Springer Berlin Heidelberg
Activate our intelligent search to find suitable subject content or patents.
Select sections of text to find matching patents with Artificial Intelligence. powered by
Select sections of text to find additional relevant content using AI-assisted search. powered by
Three computer-aided detection (CAD) projects are hosted at the Gifu University, Japan as part of the “Knowledge Cluster Initiative” of the Japanese Government. These projects are regarding the development of CAD systems for the early detection of (1) cerebrovascular diseases using brain MRI and MRA images by detecting lacunar infarcts, unruptured aneurysms, and arterial occlusions; (2) ocular diseases such as glaucoma, diabetic retinopathy, and hypertensive retinopathy using retinal fundus images; and (3) breast cancers using ultrasound 3-D volumetric whole breast data by detecting the breast masses. The brain CAD system achieves a sensitivity of 96.8% at 0.71 false positive (FP) per image for the lacunar-infarct detection, and 93.8% at 1.2 FPs per patient for the small unruptured aneurysm detection. The sensitivity and specificity for the detection of abnormal cases with arterial occlusions in MRA images are 80.0% and 95.3%, respectively. For the glaucoma detection using the retinal fundus CAD system, a sensitivity and specificity of 77.8% and 74.5% are obtained in the analysis of the optic nerve head and a sensitivity of 61.5% at 1.3 FPs per image is achieved in the detection of the retinal nerve fiber layer defects. Hemorrhages and exudates in diabetic retinopathy diagnosis are detected at a sensitivity and specificity of 84.6% and 20.6%, respectively, for the former and 76.9% and 83.3%, respectively, for the latter. For hypertensive retinopathy, the arteriolar-narrowing scheme can identify 76.2% of true positives at 1.4 FPs per image. For the breast CAD system, the image viewer that constructs the breast volume image data is developed, which also includes the CAD function with a sensitivity of 80.5% at 3.8 FPs per breast. The CAD schemes are still being improved for all the systems along with an increase in the number of image databases. Clinical examinations will be started soon, and commercialized CAD systems for the above subjects will appear by the completion of this project.