Show simple item record

dc.contributor.advisorWilliam Wells, III.en_US
dc.contributor.authorChan, Ian, 1978-en_US
dc.contributor.otherMassachusetts Institute of Technology. Dept. of Electrical Engineering and Computer Science.en_US
dc.date.accessioned2006-03-24T16:19:15Z
dc.date.available2006-03-24T16:19:15Z
dc.date.copyright2002en_US
dc.date.issued2002en_US
dc.identifier.urihttp://hdl.handle.net/1721.1/29723
dc.descriptionThesis (M.Eng.)--Massachusetts Institute of Technology, Dept. of Electrical Engineering and Computer Science, 2002.en_US
dc.descriptionIncludes bibliographical references (leaves 26-28).en_US
dc.description.abstractA multi-channel statistical classifier to detect prostate cancer was developed by combining information from 3 different MR methodologies: T2-weighted, T2-mapping, and Line Scan Diffusion lmaging(LSDI). From these MR sequences, 4 sets of image intensities were obtained: T2-weighted(T2W) from T2-weighted imaging, Apparent Diffusion Coefficient(ADC) from LSDI, and Proton Density (PD) and T2 (T2Map) from T2-mapping imaging. Manually- segmented tumor labels from a radiologist were validated by biopsy results to serve as tumor "ground truth." Textural features were derived from the images using co-occurrence matrix and discrete cosine transform. Anatomical location of voxels was described by a cylindrical coordinate system. Statistical jack-knife approach was used to evaluate our classifiers. Single-channel maximum likelihood(ML) classifiers were based on 1 of the 4 basic image intensities. Our multi-channel classifiers: support vector machine (SVM) and fisher linear discriminant(FLD), utilized 5 different sets of derived features. Each classifer generated a summary statistical map that indicated tumor likelihood in the peripheral zone(PZ) of the gland. To assess classifier accuracy, the average areas under the receiver operator characteristic (ROC) curves were compared. Our best FLD classifier achieved an average ROC area of 0.839 (±0.064) and our best SVM classifier achieved an average ROC area of 0.761 (±0.043). The T2W intensity maximum likelihood classifier, our best single-channel classifier, only achieved an average ROC area of 0.599 (± 0.146). Compared to the best single-channel ML classifier, our best multi-channel FLD and SVM classifiers have statistically superior ROC performance with P-values of 0.0003 and 0.0017 respectively from pairwise 2-sided t-test. By integrating information from the multiple images and capturing the textural and anatomical features in tumor areas, the statistical summary maps can potentially improve the accuracy of image-guided prostate biopsy and enable the delivery of localized therapy under image guidance.en_US
dc.description.statementofresponsibilityby Ian Chan.en_US
dc.format.extent28 leavesen_US
dc.format.extent2371531 bytes
dc.format.extent2371339 bytes
dc.format.mimetypeapplication/pdf
dc.format.mimetypeapplication/pdf
dc.language.isoengen_US
dc.publisherMassachusetts Institute of Technologyen_US
dc.rightsM.I.T. theses are protected by copyright. They may be viewed from this source for any purpose, but reproduction or distribution in any format is prohibited without written permission. See provided URL for inquiries about permission.en_US
dc.rights.urihttp://dspace.mit.edu/handle/1721.1/7582
dc.subjectElectrical Engineering and Computer Science.en_US
dc.titleDetection of prostate cancer using multi-parametric magnetic resonanceen_US
dc.typeThesisen_US
dc.description.degreeM.Eng.en_US
dc.contributor.departmentMassachusetts Institute of Technology. Department of Electrical Engineering and Computer Science
dc.identifier.oclc54038602en_US


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record