Cancer lesion screening in MRI for prostate biopsy

• Published in: 2016 IEEE 13th International Conference on Signal Processing (ICSP) pp. 23 - 28
• Main Authors: Yunjie Chen, Liangbin Zhang, Wenjun Ma, Su Zhang, Fen Gao, Yunkai Zhu, Yaqing Chen
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.11.2016
• Subjects: Algorithm design and analysis; Artificial neural networks; Biopsy; convex analysis of mixtures with compartmental model; lesion detection; Lesions; Magnetic resonance imaging; prostate biopsy; Prostate cancer
• ISBN: 9781509013449; 150901344X
• ISSN: 2164-5221
• DOI: 10.1109/ICSP.2016.7877789

Magnetic resonance imaging (MRI) provides reliable image information for cancer lesion screening before prostate biopsy is performed by radiologists. This study attempts to apply an algorithm for prostate based on artificial neural networks (ANN-PT) to T2 weighted image (T2WI) as well as diffusion weighted image (DWI) and convex analysis of mixture with compartmental model (CAM-CM) to dynamic contrast-enhanced MRI (DCE-MRI) for tumor detection before prostate biopsy. ANN-PT act as a classifier trained in advance by a labeled sample set, which means ANN-PT can screen lesions according to prior knowledge. CAM-CM considers each pixel in DCE-MRI as a weighted composition of distinct tissues and estimates the time intensity curves (TICs). These curves can indicate whether the tissues are malignant or benign and specify the distribution of distinct tissues. Our method is evaluated through 32 patients, especially compared two patients which have lesions be detected not by radiologist but by biopsy. The result of our method shows that it may improve the efficiency and accuracy of prostate biopsy.