Image registration of in vivo micro-ultrasound and ex vivo pseudo-whole mount histopathology images of the prostate: A proof-of-concept study

• Published in: Biomedical signal processing and control Vol. 96; p. 106657
• Main Authors: Imran, Muhammad, Nguyen, Brianna, Pensa, Jake, Falzarano, Sara M., Sisk, Anthony E., Liang, Muxuan, DiBianco, John Michael, Su, Li-Ming, Zhou, Yuyin, Joseph, Jason P., Brisbane, Wayne G., Shao, Wei
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.10.2024
• Subjects: Histopathology; Image reconstruction; Image registration; Micro-ultrasound; Prostate cancer; Image registration; Histopathology; Prostate cancer; Image reconstruction; Micro-ultrasound
• ISSN: 1746-8094
• DOI: 10.1016/j.bspc.2024.106657

Early diagnosis of prostate cancer greatly improves a patient’s 5-year survival rate. MRI-guided prostate biopsy, the most accurate diagnostic method, offers high sensitivity but moderate specificity. Despite its effectiveness, MRI is underutilized due to high costs and a shortage of expert radiologists. Micro-ultrasound (Micro-US) is a cost-effective alternative with comparable diagnostic accuracy. However, interpreting micro-US is challenging due to subtle grayscale changes distinguishing cancerous from normal tissue. This challenge can be addressed by training urologists with a large dataset of micro-US images containing ground-truth cancer outlines. Such a dataset can be mapped from surgical specimens (histopathology) onto micro-US images through image registration. In this paper, we present a semi-automated approach for registering in vivo micro-US images with ex vivo pseudo-whole mount histopathology images. Our pipeline begins with the reconstruction of pseudo-whole mount histopathology images and a 3-dimensional (3D) micro-US image volume. Each pseudo-whole mount histopathology image is then registered with the corresponding axial micro-US slice using a two-stage registration framework, which uses deep neural networks to first estimate an affine transformation, followed by a non-rigid displacement field. We evaluated our registration pipeline with micro-US and histopathology images from 18 radical prostatectomy patients using 6-fold cross-validation. The results showed a Dice coefficient of 0.97 and a mean landmark error of 2.84 mm, indicating the high accuracy of our registration pipeline. This proof-of-concept study demonstrates the feasibility of accurately aligning micro-US and histopathology images acquired at different orientations. Our code and dataset are publicly available at https://github.com/mirthAI/MUS-Pathology-Registration. •First method for registering micro-ultrasound and whole-mount histopathology images.•Cross-validation on 18 patients demonstrates the accuracy of our registration method.•Cancer labels from image registration enable cancer detection on micro-ultrasound.•Our dataset and code are publicly available.