3D digital breast cancer models with multimodal fusion algorithms

Abstract

Breast cancer image fusion consists of registering and visualizing different sets of a patient synchronized torso and radiological images into a 3D model. Breast spatial interpretation and visualization by the treating physician can be augmented with a patient-specific digital breast model that integrates radiological images. But the absence of a ground truth for a good correlation between surface and radiological information has impaired the development of potential clinical applications. A new image acquisition protocol was designed to acquire breast Magnetic Resonance Imaging (MRI) and 3D surface scan data with surface markers on the patient’s breasts and torso. A patient-specific digital breast model integrating the real breast torso and the tumor location was created and validated with a MRI/3D surface scan fusion algorithm in 16 breast cancer patients. This protocol was used to quantify breast shape differences between different modalities, and to measure the target registration error of several variants of the MRI/3D scan fusion algorithm. The fusion of single breasts without the biomechanical model of pose transformation had acceptable registration errors and accurate tumor locations. The performance of the fusion algorithm was not affected by breast volume. Further research and virtual clinical interfaces could lead to fast integration of this fusion technology into clinical practice.

Personalized 3D Breast Cancer Models with Automatic Image Segmentation and Registration

“…Additionally, the distribution of correct quadrants for those 6 patients is: 3 tumors in the Upper Outer, 2 in the Lower Outer and 1 in the intersection of Upper Interior and Outer. The validation of the fusion agorithm [5] already suggested potential limitations of the fusion algorithm to deal with large lateral breast deformations caused by MRI coils, and showed slighlty larger TRE for breast surface markers limiting the outer quadrants. Some solutions for these limitations were also advanced: a short MRI breast acquisition in the supine position (after the normal MRI acquisition protocol in prone, the patient would change position to supine) would eliminate the medial displacement of tissues caused by the MRI coils and potentially facilitate the creation of 3D digital breast models both in supine or upright postion.…”

“…for a subset of 10 patients. These patients integrated the dataset used in [5] to evaluate the benefits of incorporating a biomechanical model (BM) of pose transformation on the fusion algorithm pipeline. In detail, for this study both manual annotations of breast contour from MRI-T1w and breast tissue including malignant tumor(s) from DCE-MRI were used for validation.…”

“…However, the MRI/3D fusion algorithm from [5] was designed to take advantage of a GT annotation that includes a wider portion of the torso on the coronal plane, including quasi-rigid anatomic strucutres that facilitate the matching of breast surfaces from MRI (Fig.2, GT) and 3D surface scans ( Fig.1, 3D Scan). This motivated an adaptation of the stop criteria imposed on the segmentation algorithm and the addition of an extra post-processing step of the 3D volumetric mask to remove undesired internal structures segmented with the breast contour.…”

“…In [4,5], a multi-modal image registration combines MRI/3D surface scans into a patient-specific breast model that enables precise non-invasive pre-operative location of the tumor and provides a proper 3D upright representation for computer assisted surgery planning [10]. This two-step strategy comprises ICP for initial alignment of interior and surface data, and free form deformation (FFD) for the fine matching of the breast anterior surfaces in both representations.…”

“…In [4], we proposed a multimodal fusion algorithm that combines prone MRI with 3D surface scans in upright, that was thoroughly validated [5] in its ability to create adequate personalized 3D models of the patient's breast with precise tumor location. However, this strategy relied on manual annotations of the breast anterior contour on the T1-weighted MRI (MRI-T1w) sequence, where the 3D anatomical landmarks and structures are most visible.…”