Registration of presurgical MRI and histopathology images from radical prostatectomy via RAPSODI.

摘要: PURPOSE Magnetic resonance imaging (MRI) has great potential to improve prostate cancer diagnosis; however, subtle differences between and confounding conditions render MRI interpretation challenging. The tissue collected from patients who undergo radical prostatectomy provides a unique opportunity correlate histopathology images of the with preoperative accurately map extent onto MRI. We seek develop an open-source, easy-to-use platform align presurgical resected prostates in underwent create accurate labels on METHODS Here, we introduce RAdiology Pathology Spatial Open-Source multi-Dimensional Integration (RAPSODI), first open-source framework for registration radiology pathology images. RAPSODI relies three steps. First, it creates three-dimensional (3D) reconstruction specimen as digital representation before gross sectioning. Second, registers corresponding slices. Third, optimized transforms are applied regions outlined project those RESULTS tested phantom study where simulated various conditions, example, shrinkage during fixation. Our experiments showed that can reliably correct multiple artifacts. also evaluated 157 institutions have very different processing scanning. was 907 histpathology-MRI slices achieved Dice coefficient 0.97 ± 0.01 prostate, Hausdorff distance 1.99 ± 0.70 mm boundary, urethra deviation 3.09 ± 1.45 mm, landmark 2.80 ± 0.59 mm registered CONCLUSION robust successfully mapped providing training machine learning methods detect