3D volume reconstruction from serial breast specimen radiographs for mapping between histology and 3D whole specimen imaging

摘要: PURPOSE: In breast imaging, radiological in-vivo images, such as X-ray mammography and Magnetic Resonance Imaging (MRI), are used for tumour detection, diagnosis size determination. After excision, the specimen is typically sliced into slabs a small subset sampled. Histopathological imaging of stained samples gold standard characterisation microenvironment. A 3D volume reconstruction whole from 2D could facilitate bridging gap between histology imaging. This task challenging however, due to large deformation that tissue undergoes after surgery significant undersampling obtained in histology. this work we present method reconstruct coherent digital radiographs slabs. METHODS: To volume, propose use multiple target neighbouring slices, when deforming each slab radiograph rather than performing pairwise registrations. The algorithm combines neighbourhood slice information with Free Form Deformations, which enables exible, non-linear be computed subject constraint obtained. provides adequate constraints, without need any additional regularisation terms. RESULTS: validated on clinical mastectomy using quantitative assessment smoothness comparison image acquired validation before slicing. Additionally, registration error 5 mm (comparable thickness 4 mm) was five cases. manual annotations four observers standard, inter-observer variability 3.4 mm. Finally, illustrate how reconstructed volumes can map images sample (either MRI or CT). CONCLUSIONS: Qualitative has illustrated benefit our proposed methodology serial radiographs. knowledge first been applied cases, goal reconstructing sample. part pipeline mapping ex-vivo ultimately breast.