Computer-aided Detection of Prostate Cancer with MRI: Technology and Applications

Abstract

One in six men will develop prostate cancer in his lifetime. Early detection and accurate diagnosis of the disease can improve cancer survival and reduce treatment costs. Recently, imaging of prostate cancer has greatly advanced since the introduction of multiparametric magnetic resonance imaging (mp-MRI). Mp-MRI consists of T2-weighted sequences combined with functional sequences including dynamic contrast-enhanced MRI, diffusion-weighted MRI, and magnetic resonance spectroscopy imaging. Because of the big data and variations in imaging sequences, detection can be affected by multiple factors such as observer variability and visibility and complexity of the lesions. To improve quantitative assessment of the disease, various computer-aided detection systems have been designed to help radiologists in their clinical practice. This review paper presents an overview of literatures on computer-aided detection of prostate cancer with mp-MRI, which include the technology and its applications. The aim of the survey is threefold: an introduction for those new to the field, an overview for those working in the field, and a reference for those searching for literature on a specific application.

Keywords: MR imaging; Prostate cancer; computer-aided detection; image quantification.

Figure 1

High-resolution T2-weighted MRI. T2-weighted MR images can differentiate the normal intermediate- to high-signal-intensity peripheral zone ( Region 1) from the low-signal-intensity central and transition zones (Region 2).

Figure 2

High-resolution T2-weighted MR images of prostate cancer. (A) There is a low–signal intensity lesion on the right peripheral zone (white arrows) at the mid-gland of the prostate. At prostatectomy, the lesion was classified as a Gleason grade 7 (4+3) prostate adenocarcinoma. (B) An ill-defined homogeneous low–signal intensity area at the left transition zone (white arrows) at mid-gland of the prostate in another patient. TRUS-guided biopsy showed a Gleason grade 8 (4 + 4) prostate adenocarcinoma on the corresponding position (Images from Neto JA, Parente DB: Multiparametric magnetic resonance imaging of the prostate. Magn Reson Imaging Clin N Am 2013, 21:409–426).

Figure 3

Dynamic contrast enhanced MRI (DCE-MRI) of the prostate. (a) Axial T1 GRE unenhanced image. After contrast agent administration, an area with early enhancement is seen on the right in the peripheral zone (b, ROI1) with significant washout in the late-phase image (c). The curve (red) with early enhancement is a typical finding in the case of prostate cancer, while healthy prostate tissue is characterized by a steady slow enhancement (green). High transport constants Ktrans (e) and kep (f) can confirm suspicion of prostate cancer. Prostate adenocarcinoma with a Gleason score of 4+5=9 was diagnosed after prostatectomy (Image from Durmus, T, Baur, A, Hamm, B: Multiparametric magnetic resonance imaging in the detection of prostate cancer. Aktuelle Urol 2014, 45:119–126).

Figure 4

Multiparametric MRI (mp-MRI) of the prostate. Axial T2 TSE (A) and coronal T2 TSE (B) images show a well-defined T2 hypointense lesion in the peripheral zone (arrow) with corresponding high signal on DWI (C) and low signal on the ADC map (D). Biopsy of this region was positive for Gleason 4 + 3 prostate cancer (Images from Yacoub, JH, Oto, A, Miller, FH: MR Imaging of the Prostate. Radiologic Clinics of North America 2014, 52:811–837).

Figure 5

MR spectroscopy (MRS) of prostate cancer. (A) Axial T2-weighted MR images at the level of the prostate mid-gland to apex, shows a large hypointense lesion on the left peripheral zone. (B) A 3D MRS shows a normal spectrum on the right peripheral zone (red box) with normal choline plus creatine-to-citrate ratio of 0.48. In the voxel placed over the lesion on the left peripheral zone (blue box), the curve shows an increased choline peak and the citrate peak is markedly reduced. Random systematic biopsy showed a Gleason grade 9 (4 + 5) prostate adenocarcinoma on the left apex (Images from Neto JA, Parente DB: Multiparametric magnetic resonance imaging of the prostate. Magn Reson Imaging Clin N Am 2013, 21:409–426).

Figure 6

Flowchart for computer aided detection of prostate cancer in mp-MRI.

Figure 7

Prostate segmentation on MR images. Left: 2D MR image and segmentation results where the red curve represents the segmentation from a computer algorithm while the blue curve is the ground truth labeled by a radiologist. Right: 3D visualization after segmentation. The gold region is the prostate surface obtained by the computer algorithm while the red region is the ground truth.

Figure 8

Flowchart for a CAD system based on a multiparametric MRI. The cancer probability map is the final outcome of the algorithm (Image from Shah V, Turkbey B, Mani H, Pang Y, Pohida T, Merino MJ, Pinto PA, Choyke PL, Bernardo M: Decision support system for localizing prostate cancer based on multiparametric magnetic resonance imaging. Med Phys 2012, 39:4093–4103).

Figure 9

Image features for prostate cancer detection. (a) With prostate cancer superposed in green. (b) First order statistics (standard deviation). (c) Sobel-Kirsch feature. (d) second order statistics (contrast inverse moment). (e) Corresponding time-intensity curves for CaP (red) and benign (blue) regions are shown based on DCE-MRI data (Images from Viswanath S, Bloch BN, Rosen M, Chappelow J, Toth R, Rofsky N, Lenkinski R, Genega E, Kalyanpur A, Madabhushi A: Integrating Structural and Functional Imaging for Computer Assisted Detection of Prostate Cancer on Multi-Protocol 3 Tesla MRI. Proc Soc Photo Opt Instrum Eng 2009, 7260:72603I).

Figure 10

Registration between multiparametric MRI and histology.

Figure 11

Registration between MRI and histology. Top: Workflow for pathology-mp-MRI registration in a surgical 3D space. Bottom: 3D deformable registration of virtual whole-mount histology (1), fresh specimen (2), T2 weighted MRI (3), perfusion (4), and diffusion (5) sequences (ADC) applied to prostate cancer (Image from Orczyk C, Rusinek H, Rosenkrantz AB, Mikheev A, Deng FM, Melamed J, Taneja SS: Preliminary experience with a novel method of three-dimensional co-registration of prostate cancer digital histology and in vivo multiparametric MRI. Clin Radiol 2013, 68:e652–658).

Figure 12

MRI and ultrasound fusion for targeted biopsy of the prostate. (A and B) Anterior lesion of the high suspicious lesion identified on mp-MRI. (C) Real-time ultrasound targeting the corresponding lesion. (D and E) 3D models demonstrate the target (blue), prostate (brown), and biopsy cores (tan cylinders). (F) Radical prostatectomy pathology confirmed a 2.3 cm Gleason 8 (4+4) cancer centered in the right anterior prostate (Images from Sonn, GA, Margolis, DJ, Marks, LS: Target detection: Magnetic resonance imaging-ultrasound fusion–guided prostate biopsy. Urologic Oncology: Seminars and Original Investigations 2014, 32:903–911).