Spatially weighted mutual information (SWMI) for registration of digitally reconstructed ex vivo whole mount histology and in vivo prostate MRI

In this work, we present a scheme for the registration of digitally reconstructed whole mount histology (WMH) to pre-operative in vivo multiprotocol prostate MR imagery (T2w and DCE) using spatially weighted mutual information (SWMI). Spatial alignment of ex vivo histological sections to pre-operative in vivo MRI for prostate cancer (CaP) patients undergoing radical prostatectomy is a necessary first step in the discovery of quantitative multiprotocol MRI signatures for CaP. This may be done by spatially mapping delineated extent of disease on ex vivo histopathology onto pre-operative in vivo MRI via image registration. Apart from the challenges in spatially registering multi-modal data (histology and MRI) on account of (a) modality specific differences, (b) deformation due to the endorectal coil and tissue loss on histology, another complication is that the ex vivo histological sections, in the lab, are usually obtained as quadrants. This means they need to be reconstituted as a pseudo-whole mount histologic section (WMHS) prior to registration with MRI. An additional challenge is that most registration techniques rely on availability of the pre-segmented prostate capsule on T2w MRI. The novel contribution of this paper is that it leverages a spatially weighted mutual information (SWMI) scheme to automatically register and map CaP extent from WMHS onto pre-operative, multiprotocol MRI. The SWMI scheme obviates the need for pre-segmentation of the prostate capsule on MRI. Additionally, we leverage a program developed by our group, Histostitcher©, for interactive stitching of individual histology quadrants to digitally reconstruct the pseudo WMHS. Our registration methodology comprises the following main steps, (1) affine registration of T2w and DCE MRI, (2) affine registration of stitched WMHS to multiprotocol T2w and DCE MRI, and (3) multimodal image registration of WMHS to multiprotocol T2w and DCE MRI using SWMI. We quantitatively and qualitatively evaluated all aspects of our methodology in the multimodal registration of a total of 7 corresponding histology and MRI sections from 2 different patients. For the 7 studies, we obtained an average Hausdorff distance of 1.85 mm, mean absolute distance of 0.99 mm, RMS of 1.65 mm, and DICE of 0.83, when comparing the capsular alignment on MRI to histology.

[1]  Guy Marchal,et al.  Multi-modality image registration by maximization of mutual information , 1996, Proceedings of the Workshop on Mathematical Methods in Biomedical Image Analysis.

[2]  Daniel Rueckert,et al.  Groupwise Combined Segmentation and Registration for Atlas Construction , 2007, MICCAI.

[3]  R. Lenkinski,et al.  Elastic registration of multimodal prostate MRI and histology via multiattribute combined mutual information. , 2011, Medical physics.

[4]  Robert E Lenkinski,et al.  Prostate cancer: accurate determination of extracapsular extension with high-spatial-resolution dynamic contrast-enhanced and T2-weighted MR imaging--initial results. , 2007, Radiology.

[5]  Guy Marchal,et al.  Multimodality image registration by maximization of mutual information , 1997, IEEE Transactions on Medical Imaging.

[6]  Max A. Viergever,et al.  Mutual-information-based registration of medical images: a survey , 2003, IEEE Transactions on Medical Imaging.

[7]  Anant Madabhushi,et al.  A Comprehensive Segmentation, Registration, and Cancer Detection Scheme on 3 Tesla In VivoProstate DCE-MRI , 2008, MICCAI.

[8]  Jason W Sohn,et al.  Spatially weighted mutual information image registration for image guided radiation therapy. , 2010, Medical physics.

[9]  Hyunjin Park,et al.  Registration methods for histological slides and ex vivo MRI of prostate , 2007, 2007 IEEE Nuclear Science Symposium Conference Record.

[10]  Anant Madabhushi,et al.  Determining histology-MRI slice correspondences for defining MRI-based disease signatures of prostate cancer , 2011, Comput. Medical Imaging Graph..

[11]  Anant Madabhushi,et al.  HistoStitcher©: An interactive program for accurate and rapid reconstruction of digitized whole histological sections from tissue fragments , 2011, Comput. Medical Imaging Graph..