Public reporting burden for this collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, gathering and maintaining the data needed, and completing and reviewing this collection of information. Send comments regarding this burden estimate or any other aspect of this collection of information, including suggestions for reducing this burden to Department of Defense, Washington Headquarters Services, Directorate for Information Operations and Reports (0704-0188), 1215 Jefferson Davis Highway, Suite 1204, Arlington, VA 22202-4302. Respondents should be aware that notwithstanding any other provision of law, no person shall be subject to any penalty for failing to comply with a collection of information if it does not display a currently valid OMB control number. PLEASE DO NOT RETURN YOUR FORM TO THE ABOVE ADDRESS.
REPORT DATE (DD-MM-YYYY)
01-07-2006
REPORT TYPE
Annual
DATES COVERED (From -To
PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) 8. PERFORMING ORGANIZATION REPORT NUMBERUniversity of Michigan Ann Arbor, Michigan 48109-0904
SPONSORING / MONITORING AGENCY NAME(S) AND ADDRESS(ES) 10. SPONSOR/MONITOR'S ACRONYM(S)
U.S. Army Medical Research and Materiel Command Fort Detrick, Maryland 21702-5012
SPONSOR/MONITOR'S REPORT NUMBER(S)
DISTRIBUTION / AVAILABILITY STATEMENTApproved for Public Release; Distribution Unlimited
SUPPLEMENTARY NOTES
ABSTRACT:The goal of this project is to develop a computer-aided diagnosis (CAD) system for automatic interval change analysis of microcalcification clusters on mammograms. Based on our regional registration method and a search program cluster candidates were detected within the local area on the prior. The cluster on the current image is then paired with the candi-dates to form true (TP-TP) or false (TP-FP) pairs and a correspondence classifier is designed to reduce the (TP-FP). A temporal classifier (TC) based on current and prior information is used if a cluster is detected in the prior, and a current classifier (CurC) based on current information alone is used if no prior cluster is detected. For the TC an LDA, SVM and NN were used. 175 temporal pairs of mammograms were used for evaluation. The registration stage identified 85% (149/175) of the TP-TP pairs with 15 false matches within the 164 image pairs that had detected clusters. The TC based on LDA, SVM and NN achieved a test Az of 0.83, 0.82, 0.84, respectively, for the 164 pairs for classifying the clusters as malignant or benign. For the 11 clusters without detection on the prior, the test Az by the CurC was 0.72. Four radiologists participated in pilot observer study using our CAD. The average Az in estimating the likelihood of malignancy was 0.70 without CAD and improved to 0.77 with CAD(p=0.04).
SUBJECT TERMS
ABSTRACTWe are developing an automated system for analysis of microcalcification clusters on serial mammograms. Our automated system consists of two stages: (1) automatic registration of corresponding clusters on tempo...