Stochastic Sequential Modeling: Toward Improved Prostate Cancer Diagnosis Through Temporal-Ultrasound
- PDF / 1,896,370 Bytes
- 12 Pages / 593.972 x 792 pts Page_size
- 9 Downloads / 144 Views
Annals of Biomedical Engineering ( 2020) https://doi.org/10.1007/s10439-020-02585-y
Original Article
Stochastic Sequential Modeling: Toward Improved Prostate Cancer Diagnosis Through Temporal-Ultrasound LAYAN NAHLAWI ,1 FARHAD IMANI,2 MENA GAED,3 JOSE A. GOMEZ,3 MADELEINE MOUSSA,3 ELI GIBSON,4 AARON FENSTER,5 AARON WARD,6 PURANG ABOLMAESUMI,2 PARVIN MOUSAVI,1 and HAGIT SHATKAY1,7 1 School of Computing, Queen’s University, Kingston, ON K7L 2N8, Canada; 2Department of Electrical and Computer Engineering, University of British Columbia, Vancouver, BC, Canada; 3London Health Sciences Centre, London, ON, Canada; 4 University College London, London, UK; 5Robarts Research Institute, London, ON, Canada; 6Department of Medical Physics, Western University, London, ON, Canada; and 7Department of Computer and Information Sciences, University of Delaware, Newark, DE, USA
(Received 11 November 2019; accepted 27 July 2020) Associate Editor Agata A. Exner oversaw the review of this article.
Abstract—Prostate cancer (PCa) is a common, serious form of cancer in men that is still prevalent despite ongoing developments in diagnostic oncology. Current detection methods lead to high rates of inaccurate diagnosis. We present a method to directly model and exploit temporal aspects of temporal enhanced ultrasound (TeUS) for tissue characterization, which improves malignancy prediction. We employ a probabilistic-temporal framework, namely, hidden Markov models (HMMs), for modeling TeUS data obtained from PCa patients. We distinguish malignant from benign tissue by comparing the respective log-likelihood estimates generated by the HMMs. We analyze 1100 TeUS signals acquired from 12 patients. Our results show improved malignancy identification compared to previous results, demonstrating over 85% accuracy and AUC of 0.95. Incorporating temporal information directly into the models leads to improved tissue differentiation in PCa. We expect our method to generalize and be applied to other types of cancer in which temporal-ultrasound can be recorded. Keywords—Image guided diagnosis, Hidden Markov models, Time-domain analysis, TRUS-guided biopsies, Tissue characterization.
Address correspondence to Layan Nahlawi, School of Computing, Queen’s University, Kingston, ON K7L 2N8, Canada. Electronic mail: [email protected] Parvin Mousavi and Hagit Shatkay have contributed equally to this work and sharing last authorship.
INTRODUCTION Prostate cancer (PCa) is the most commonly diagnosed form of cancer in men, second only to skin cancer. The number of new cases in the USA alone during 2019 is estimated at 174,650.1 A definitive diagnosis is obtained through histopathology analysis of prostate-tissue specimen collected during core needle biopsy under trans-rectal ultrasound (TRUS) guidance after initial clinical-assessment.7 Some centers use magnetic resonance (MR) and MR-TRUS fusion for guiding biopsies.22,28 TRUS-guided biopsies often lead to a high rate (~ 40%) of false negatives for cancer diagnosis.27 Extensive heterogeneity in morphology
Data Loading...
