The purpose of our prostate tissue-typing studies is to develop more sensitive and specific imaging methods for identifying and characterizing cancerous prostate tissue and thereby improving the effectiveness of biopsy guidance, therapy targeting, and treatment monitoring. We acquired ultrasonic radio-frequency echo-signal data, and clinical variables, e.g., prostate-specific antigen, during biopsy examinations; computed spectra of the radio-frequency signals in each biopsied region; and trained neural network classifiers using biopsy results as the gold standard. Lookup tables returned scores for cancer likelihood on a pixel-by-pixel basis from spectral parameter and PSA values to generate tissue-type images (TTIs). ROC curves based on a leave-one-patient-out approach were used to minimize the chance of biased classification. The ROC-curve area for neural-network-based classification derived from the values of ultrasound spectral parameters plus PSA level was 0.84 +/− 0.02; in comparison, the ROC-curve area for B-mode-based classification was only 0.64 +/− 0.03. Furthermore, the sensitivity of neural-network-based classification was more than 50% better than the sensitivity of B-mode-based classification at ROC specificity values corresponding to a B-mode, biopsy-guidance sensitivity of 50% to 60%.
[1]
Ernest J. Feleppa,et al.
Application of spectrum analysis and neural-network classification to imaging for targeting and monitoring treatment of prostate cancer
,
2001,
2001 IEEE Ultrasonics Symposium. Proceedings. An International Symposium (Cat. No.01CH37263).
[2]
Ernest J. Feleppa,et al.
Emerging Ultrasound Technologies for Early Markers of Disease
,
2003,
Disease markers.
[3]
E J Feleppa,et al.
Diagnostic spectrum analysis in ophthalmology: a physical perspective.
,
1986,
Ultrasound in medicine & biology.
[4]
Andrew Kalisz,et al.
Recent Developments in Tissue-Type Imaging (TTI) for Planning and Monitoring Treatment of Prostate Cancer
,
2004,
Ultrasonic imaging.
[5]
C. Metz.
ROC Methodology in Radiologic Imaging
,
1986,
Investigative radiology.
[6]
E J Feleppa,et al.
Spectrum-Analysis and Neural Networks for Imaging to Detect and Treat Prostate Cancer
,
2001,
Ultrasonic imaging.