Design of Refractory Alloys for Desired Thermal Conductivity via AI-Assisted In-Silico Microstructure Realization.

A computational methodology based on supervised machine learning (ML) is described for characterizing and designing anisotropic refractory composite alloys with desired thermal conductivities (TCs). The structural design variables are parameters of our fast computational microstructure generator, which were linked to the physical properties. Based on the Sobol sequence, a sufficiently large dataset of artificial microstructures with a fixed volume fraction (VF) was created. The TCs were calculated using our previously developed fast Fourier transform (FFT) homogenization approach. The resulting dataset was used to train our optimal autoencoder, establishing the intricate links between the material's structure and properties. Specifically, the trained ML model's inverse design of tungsten-30% (VF) copper with desired TCs was investigated. According to our case studies, our computational model accurately predicts TCs based on two perpendicular cut-section images of the experimental microstructures. The approach can be expanded to the robust inverse design of other material systems based on the target TCs.

The feasibility of a dose painting procedure to treat prostate cancer based on mpMR images and hierarchical clustering.

BACKGROUND: We aimed to assess the feasibility of a dose painting (DP) procedure, known as simultaneous integrated boost intensity modulated radiation Therapy (SIB-IMRT), for treating prostate cancer with dominant intraprostatic lesions (DILs) based on multi-parametric magnetic resonance (mpMR) images and hierarchical clustering with a machine learning technique. METHODS: The mpMR images of 120 patients were used to create hierarchical clustering and draw a dendrogram. Three clusters were selected for performing agglomerative clustering. Then, the DIL acquired from the mpMR images of 20 patients were categorized into three groups to have them treated with a DP procedure being composed of three planning target volumes (PTVs) determined as PTV1, PTV2, and PTV3 in treatment plans. The DP procedure was carried out on the patients wherein a total dose of 80, 85 and 91 Gy were delivered to the PTV1, PTV2, and PTV3, respectively. Dosimetric and radiobiologic parameters [Tumor Control Probability (TCP) and Normal Tissue Complication Probability (NTCP)] of the DP procedure were compared with those of the conventional IMRT and Three-Dimensional Conformal Radiation Therapy (3DCRT) procedures carried out on another group of 20 patients. A post-treatment follow-up was also made four months after the radiotherapy procedures. RESULTS: All the dosimetric variables and the NTCPs of the organs at risks (OARs) revealed no significant difference between the DP and IMRT procedures. Regarding the TCP of three investigated PTVs, significant differences were observed between the DP versus IMRT and also DP versus 3DCRT procedures. At post-treatment follow-up, the DIL volumes and apparent diffusion coefficient (ADC) values in the DP group differed significantly (p-value < 0.001) from those of the IMRT. However, the whole prostate ADC and prostate-specific antigen (PSA) indicated no significant difference (p-value > 0.05) between the DP versus IMRT. CONCLUSIONS: The results of this comprehensive clinical trial illustrated the feasibility of our DP procedure for treating prostate cancer based on mpMR images validated with acquired patients' dosimetric and radiobiologic assessment and their follow-ups. This study confirms significant potential of the proposed DP procedure as a promising treatment planning to achieve effective dose escalation and treatment for prostate cancer. TRIAL REGISTRATION: IRCT20181006041257N1; Iranian Registry of Clinical Trials, Registered: 23 October 2019, https://en.irct.ir/trial/34305 .

A novel design of printable tunable stiffness metamaterial for bone healing.

A tunable stiffness bone rod was designed, optimized, and 3D printed to address the shortcomings of existing bone fixation devices, such as stress shielding and bone nonunion in the healing of fractured bones. Current bone plates/rods have constant and high stiffness. High initial stiffness prevents the micromotion of newly formed bone and results in poor bone healing. Our novel design framework provides surgeons with a ready-for-3D-printing, patient-specific design, optimized to have the desired force-displacement response with a stopping mechanism for preventing further deformation under higher-than-normal loads, such as falling. The computational framework is a design optimization based on the multi-objective genetic algorithm (GA) optimization with the FE simulation to quantify the objectives: tuning the varied stiffness while minimizing the maximum von Mises stress of the model to avoid plastic and permanent deformation of the bone rod. The computational framework for optimum design of tunable stiffness metamaterial presented in this paper is not specific for a tibia bone rod, and it can be used for any application where bilinear stiffness is desirable.

A novel machine learning based computational framework for homogenization of heterogeneous soft materials: application to liver tissue.

Real-time simulation of organs increases comfort and safety for patients during the surgery. Proper generalized decomposition (PGD) is an efficient numerical method with coordinate errors below 1 mm and response time below 0.1 s that can be used for simulated surgery. For input of this approach, nonlinear mechanical properties of each segment of the liver need to be calculated based on the geometries of the patient's liver extracted using medical imaging techniques. In this research work, a map of the mechanical properties of the liver tissue has been estimated with a novel combined method of the finite element (FE) optimization. Due to the existence of major-size vessels in the liver that makes the surrounding tissue anisotropic, the simulation of hyperelastic material with two different sections is computationally expensive. Thus, a homogenized, anisotropic, and hyperelastic model with the nearest response to the real heterogeneous model was developed and presented. Because of various possibilities of the vessel orientation, position, and size, homogenization has been carried out for adequate samples of heterogeneous models to train artificial neural networks (ANNs) as machine learning tools. Then, an unknown sample of heterogeneous material was categorized and mapped to its homogenized material parameters with the trained networks for the fast and low-cost generalization of our combined FE optimization method. The results showed the efficiency of the proposed novel machine learning based technique for the prediction of effective material properties of unknown heterogeneous tissues.