SPIE Membership Get updates from SPIE Newsroom
  • Newsroom Home
  • Astronomy
  • Biomedical Optics & Medical Imaging
  • Defense & Security
  • Electronic Imaging & Signal Processing
  • Illumination & Displays
  • Lasers & Sources
  • Micro/Nano Lithography
  • Nanotechnology
  • Optical Design & Engineering
  • Optoelectronics & Communications
  • Remote Sensing
  • Sensing & Measurement
  • Solar & Alternative Energy
  • Sign up for Newsroom E-Alerts
  • Information for:
    Advertisers
SPIE Photonics West 2018 | Call for Papers

SPIE Journals OPEN ACCESS

SPIE PRESS

SPIE PRESS




Print PageEmail PageView PDF

Biomedical Optics & Medical Imaging

Biomechanical modeling to enhance imaging-guided prostate biopsy

Finite element technology and practical insights can enhance the accuracy of magnetic resonance-guided ultrasound biopsy of the prostate.
27 May 2016, SPIE Newsroom. DOI: 10.1117/2.1201604.006432

Prostate cancer is the second leading cause of cancer death in American men, and is currently confirmed by systematic trans-rectal ultrasound (TRUS)-guided biopsy. However, this approach samples only a fraction of the prostate, and often results in missing and undergrading cancer, which in turn leads to under- and overtreatment.1 Multiparametric magnetic resonance imaging (mpMRI)-guided in-bore biopsy has been shown to be highly accurate in detecting and localizing aggressive cancer.2, 3 A further promising option is to enhance TRUS biopsies by means of image fusion MR guidance (which aligns MR and TRUS anatomically to provide a clinician with MR information during TRUS biopsy).4 The technology of image fusion of mpMRI and TRUS poses several interesting technical challenges, which we discuss here.

Purchase SPIE Field Guide to MicroscopyImage fusion technology is subject to registration errors that limit the accuracy of this approach and set a lower size limit on targets that can be successfully biopsied. Furthermore, the target area is determined not by the size of the whole tumor but by that of the aggressive area within that tumor (see Figure 1, bottom right panel, where the area is marked in red). The aggressive area is highly visible on mpMRI (see Figure 1, bottom left panel, the dark area in the apparent diffusion coefficient image), but the actual tumor might be larger on pathology (Figure 1, top right panel, marked by blue lines). Sampling such a heterogeneous lesion, but not at the most aggressive area, causes the diagnosis to undergrade the tumor, and may result in a wrong treatment decision. We investigated how often lesions are heterogeneous in a population of 51 patients with 62 peripheral-zone tumors. We found that more than half of the tumors were heterogeneous, thus containing a high-grade tumor focal spot. Using the same population, we performed a biopsy simulation experiment to compute the required target registration error (TRE), given a requested hit rate. The results of this experiment showed that we required a TRE of 1.9mm to correctly establish the highest Gleason grade component in 95% of the tumors with a single MR-guided TRUS biopsy.5 However, this measure of accuracy is theoretical. In clinical practice, factors such as the correct placement of the biopsy gun and needle deflection would possibly lead to additional inaccuracies. Multiple biopsies may decrease the TRE requirement.6


Figure 1. Prostate imaging demonstrating that actual tumor size on pathology is often underestimated on magnetic resonance images (MRI). Top left: Pathology section with annotated whole tumor area (in blue). Top right: T2-weighted MRI with pathology annotation overlay. Bottom left: Apparent diffusion coefficient MRI with pathology annotation overlay. Bottom right: Cutout of MR tumor area, as shown on pathology and on MRI in red.

Commercially available fusion technology has a TRE in the range of 3–6mm, which would fail to target the smaller tumors or focal spots in our representative population. Therefore, we investigated improvements to the most accurate surface-based deformable MR-TRUS registration. The current approach for the deformable registration of MR and TRUS images is to align prostate surfaces. However, this type of surface-based registration assumes an unrealistic, highly elastic model of internal prostate deformation. To improve this approach, we extended surface-based registration with realistic biomechanical modeling, and successfully constrained the internal deformation of the prostate for trans-perineal ultrasound.7, 8 We then applied this method to the more conventional trans-rectal approach. First, we assumed that we had a segmentation of the prostate in MR and ultrasound. We developed methods to automatically, properly condition MR and TRUS tetrahedral mesh models (see Figure 2). Next, we forced the external vertices of one model to match those of the other. The internal vertices displaced as a result of realistic biomechanical modeling, and the external and internal vertices drove a thin plate spline non-rigid transformation. We collected MR and TRUS images from 10 patients, and used anatomical landmarks to determine the registration accuracy. The median TRE of a surface-based registration with biomechanical regularization was 2.76mm. This was significantly better than the median TRE of 3.47mm for regular surface-based registration without biomechanical regularization. We concluded that biomechanical finite element modeling has the potential to improve the accuracy of multimodal prostate registration and could help to improve the effectiveness of MR-guided TRUS biopsy procedures.


Figure 2. Mesh models of a prostate. Right: Whole prostate model. Left: Model with a cross section take-out to show the 3D tetrahedral elements that are used in the biomechanical modeling software.

MR guidance enhances MR-guided TRUS by increasing the visibility of prostate lesions (see Figure 3), making it possible to target these lesions accurately. To biopsy TRUS ‘invisible’ lesions, the operator has to rely solely on accurate registration. Therefore, to have a better understanding of the visibility of prostate cancer on TRUS images using prior knowledge of MRI, we determined the proportion of lesions that become visible on TRUS after fusion. We collected mpMRI images from 34 patients and assessed these according to the prostate imaging reporting and data system (PI-RADS) guidelines.9 In total, we detected and scored 56 lesions on mpMRI. We asked five observers to determine the visibility of each of the prostate lesions on 3D TRUS images and assign a visibility score on a five-point scale. Our results showed that 43% of all MR lesions were considered visible by the majority of the observers. For PI-RADS 4 and 5 lesions the results were 55 and 62%, respectively. For clinical application, this means that more than half of the potential biopsy targets are visible on TRUS images during MR fusion. For these cases, and especially for the most aggressive PI-RADS 5 cases (of which almost two-thirds were visible), MR-guided TRUS fusion biopsy will benefit from TRUS lesion visibility.


Figure 3. Images of the prostate showing the cross section of a tumor (identified by the blue arrows). Top: MRI. Bottom: Trans-rectal ultrasound (TRUS) aligned using fusion technology. The dark region on TRUS augmented by MR fusion enables clear guidance for TRUS biopsy.

In summary, MR-guided prostate biopsy can miss the aggressive area of a tumor. We have described a method that enhances such biopsies by means of biomechanical modeling and knowledge about TRUS appearance of tumors. Many prostate lesions are benign and require no biopsy. MR-guided TRUS active surveillance could provide a cost-effective means to avoid biopsy and remain vigilant. We are now investigating technology to make this feasible.


Henkjan Huisman, Wendy van de Ven
Radboud University Nijmegen Medical Centre
Nijmegen, The Netherlands

Henkjan Huisman is assistant professor of computer-assisted prostate imaging diagnostics and interventions.


References:
1. F. H. Schröder, J. Hugosson, M. J. Roobol, T. L. J. Tammela, S. Ciatto, V. Nelen, M. Kwiatkowski, et al., Screening and prostate-cancer mortality in a randomized European study, N. Engl. J. Med. 360, p. 1320-1328, 2009.
2. T. Hambrock, D. M. Somford, C. Hoeks, S. A. Bouwense, H. Huisman, D. Yakar, I. M. van Oort, J. A. Witjes, J. J. Fütterer, J. O. Barentz, Magnetic resonance imaging guided prostate biopsy in men with repeat negative biopsies and increased prostate specific antigen, J. Urol. 183, p. 520-527, 2010.
3. M. R. Pokorny, M. de Rooij, E. Duncan, F. H. Schröder, R. Parkinson, J. R. Barentz, L. C. Thompson, Prospective study of diagnostic accuracy comparing prostate cancer detection by transrectal ultrasound-guided biopsy versus magnetic resonance (MR) imaging with subsequent MR-guided biopsy in men without previous prostate biopsies, Eur. Urol. 66, p. 22-29, 2014.
4. M. M. Siddiqui, S. Rais-Bahrami, B. Turkbey, A. K. George, J. Rothwax, N. Shakir, C. Okoro, et al., Comparison of MR/ultrasound fusion-guided biopsy with ultrasound-guided biopsy for the diagnosis of prostate cancer, J. Am. Med. Assoc. 313, p. 390-397, 2015.
5. W. J. M. van de Ven, C. A. Hulsbergen-van de Kaa, T. Hambrock, J. O. Barentsz, H. J. Huisman, Simulated required accuracy of image registration tools for targeting high-grade cancer components with prostate biopsies, Eur. Radiol. 23, p. 1401-1407, 2013.
6. P. R. Martin, D. W. Cool, C. Romagnoli, A. Fenster, A. D. Ward, Magnetic resonance imaging-targeted, 3D transrectal ultrasound-guided fusion biopsy for prostate cancer: quantifying the impact of needle delivery error on diagnosis, Med. Phys. 41, p. 073504, 2014.
7. Y. Hu, H. U. Ahmed, Z. Taylor, C. Allen, M. Emberton, D. Hawkes, D. Barratt, MR to ultrasound registration for image-guided prostate interventions, Med. Image Anal. 16, p. 687-703, 2012.
8. Y. Hu, E. Gibson, H. U. Ahmed, C. M. Moore, M. Emberton, D. C. Barratt, Population-based prediction of subject-specific prostate deformation for MR-to-ultrasound image registration, Med. Image Anal. 26, p. 332-344, 2015.
9. J. C. Weinreb, J. O. Barentz, P. L. Choyke, F. Cornud, M. A. Haider, K. J. Macura, D. Margolis, et al., PI-RADS prostate imaging—reporting and data system: 2015, version 2, Eur. Urol. 69, p. 16-40, 2016.