Data-driven patient-specific breast modeling: a simple, automatized, and robust computational pipeline

Doctoral thesis (2022)

Background: Breast-conserving surgery is the most acceptable option for breast cancer removal from an invasive and psychological point of view. During the surgical procedure, the imaging acquisition using ... [more ▼]

Background: Breast-conserving surgery is the most acceptable option for breast cancer removal from an invasive and psychological point of view. During the surgical procedure, the imaging acquisition using Magnetic Image Resonance is performed in the prone configuration, while the surgery is achieved in the supine stance. Thus, a considerable movement of the breast between the two poses drives the tumor to move, complicating the surgeon's task. Therefore, to keep track of the lesion, the surgeon employs ultrasound imaging to mark the tumor with a metallic harpoon or radioactive tags. This procedure, in addition to an invasive characteristic, is a supplemental source of uncertainty. Consequently, developing a numerical method to predict the tumor movement between the imaging and intra-operative configuration is of significant interest. Methods: In this work, a simulation pipeline allowing the prediction of patient-specific breast tumor movement was put forward, including personalized preoperative surgical drawings. Through image segmentation, a subject-specific finite element biomechanical model is obtained. By first computing an undeformed state of the breast (equivalent to a nullified gravity), the estimated intra-operative configuration is then evaluated using our developed registration methods. Finally, the model is calibrated using a surface acquisition in the intra-operative stance to minimize the prediction error. Findings: The capabilities of our breast biomechanical model to reproduce real breast deformations were evaluated. To this extent, the estimated geometry of the supine breast configuration was computed using a corotational elastic material model formulation. The subject-specific mechanical properties of the breast and skin were assessed, to get the best estimates of the prone configuration. The final results are a Mean Absolute Error of 4.00 mm for the mechanical parameters E_breast = 0.32 kPa and E_skin = 22.72 kPa. The optimized mechanical parameters are congruent with the recent state-of-the-art. The simulation (including finding the undeformed and prone configuration) takes less than 20 s. The Covariance Matrix Adaptation Evolution Strategy optimizer converges on average between 15 to 100 iterations depending on the initial parameters for a total time comprised between 5 to 30 min. To our knowledge, our model offers one of the best compromises between accuracy and speed. The model could be effortlessly enriched through our recent work to facilitate the use of complex material models by only describing the strain density energy function of the material. In a second study, we developed a second breast model aiming at mapping a generic model embedding breast-conserving surgical drawing to any patient. We demonstrated the clinical applications of such a model in a real-case scenario, offering a relevant education tool for an inexperienced surgeon. [less ▲]

Towards real-time patient-specific breast simulations: from full-field information to surrogate model

Scientific Conference (2022, July)

In breast cancer treatment, surgery is one of the most common practices [DeSantis et al., 2019]. The surgery involves a complex pipeline, principally due to the difference between the imaging and the ... [more ▼]

In breast cancer treatment, surgery is one of the most common practices [DeSantis et al., 2019]. The surgery involves a complex pipeline, principally due to the difference between the imaging and the surgical posture [Mazier et al., 2021]. Indeed, because of the stance difference, the surgeon has to rely on radioactive or invasive markers to predict the tumor position in the surgical setup. Biomechanical simulations could predict such complex tumor displacements but often require patient-specific data (material properties, organs geometries, or loading and boundary conditions). Full-field acquisitions coupled with landmark identifications allow obtaining relative deformation between the different configurations. Having this information and assuming a finite element model, an identification procedure of the model parameters can be carried out. Finally, finding a suitable computational model allowing for a compromise between accuracy and speed, one may consider surrogate models for real-time simulations (20 to 50 FPS). In this work, we obtained the patient-specific geometry through micro-computed tomography in 8 different configurations, including 15 bio-markers. Assessing the displacement of the bio-markers enabled us to infer the relative strains between the different configurations. A heterogeneous neo-Hookean model was assumed for simulating soft tissue behavior. Based on the displacements and the position of the biomarkers, model parameters identification was performed to calibrate the experimental data with the finite element method results. To overcome speed performance issues, Convolutional Neural Network (CNN) trained with a synthetic simulation-based dataset generated by applying different gravity directions is used. Preliminary results show that CNN can predict the displacement of anatomical landmarks to millimetric precision and is 100 times faster than the finite element method, satisfying our real-time objective. Plus, the use of Bayesian inferences involves a longer prediction time but allows a 95% confidence interval of the biomarkers' displacements. For a given precision, contrary to CNNs, optimization methods are computationally expensive and depend on an initialization point. Although CNNs require new training for each patient, optimization algorithms can be applied regardless of the patient's geometry. Through this study, we observed that material properties were playing an essential role but not as much as the anatomical structures e.g. infra-mammary or Copper’s ligaments. [less ▲]

Inverse deformation analysis: an experimental and numerical assessment using the FEniCS Project

in Engineering with Computers (2022)

In this paper, we develop a framework for solving inverse deformation problems using the FEniCS Project finite-element software. We validate our approach with experimental imaging data acquired from a ... [more ▼]

In this paper, we develop a framework for solving inverse deformation problems using the FEniCS Project finite-element software. We validate our approach with experimental imaging data acquired from a soft silicone beam under gravity. In contrast with inverse iterative algorithms that require multiple solutions of a standard elasticity problem, the proposed method can compute the undeformed configuration by solving only one modified elasticity problem. This modified problem has a complexity comparable to the standard one. The framework is implemented within an open-source pipeline enabling the direct and inverse deformation simulation directly from imaging data. We use the high-level unified form language (UFL) of the FEniCS Project to express the finite-element model in variational form and to automatically derive the consistent Jacobian. Consequently, the design of the pipeline is flexible: for example, it allows the modification of the constitutive models by changing a single line of code. We include a complete working example showing the inverse deformation of a beam deformed by gravity as supplementary material. [less ▲]

A rigged model of the breast for preoperative surgical planning

in Journal of Biomechanics (2021), 128

In breast surgical practice, drawing is part of the preoperative planning procedure and is essential for a successful operation. In this study, we design a pipeline to assist surgeons with patient ... [more ▼]

In breast surgical practice, drawing is part of the preoperative planning procedure and is essential for a successful operation. In this study, we design a pipeline to assist surgeons with patient-specific breast surgical drawings. We use a deformable torso model containing the surgical patterns to match any breast surface scan. To be compatible with surgical timing, we build an articulated model through a skinning process coupled with shape deformers to enhance a fast registration process. On one hand, the scalable bones of the skinning account for pose and morphological variations of the patients. On the other hand, pre-designed artistic blendshapes create a linear space for guaranteeing anatomical variations. Then, we apply meaningful constraints to the model to find a trade-off between precision and speed. The experiments were conducted on 7 patients, in 2 different poses (prone and supine) with a breast size ranging from 36A and 42C (US/UK bra sizing). The acquisitions were obtained using the depth camera Structure Sensor, and the breast scans were acquired in less than 1 minute. The result is a registration method converging within a few seconds (3 maximum), reaching a Mean Absolute Error of 2.3 mm for mesh registration and 8.0 mm for breast anatomical landmarks. Compared to the existing literature, our model can be personalized and does not require any database. Finally, our registered model can be used to transfer surgical reference patterns onto any patient in any position. [less ▲]

DIGITAL TWINNING FOR REAL-TIME SIMULATION

Poster (2019, November)