Three-dimensional deep learning to automatically generate cranial implant geometry

Abstract

We present a 3D deep learning framework that can generate a complete cranial model using a defective one. The Boolean subtraction between these two models generates the geometry of the implant required for surgical reconstruction. There is little or no need for post-processing to eliminate noise in the implant model generated by the proposed approach. The framework can be used to meet the repair needs of cranial imperfections caused by trauma, congenital defects, plastic surgery, or tumor resection. Traditional implant design methods for skull reconstruction rely on the mirror operation. However, these approaches have great limitations when the defect crosses the plane of symmetry or the patient's skull is asymmetrical. The proposed deep learning framework is based on an enhanced three-dimensional autoencoder. Each training sample for the framework is a pair consisting of a cranial model converted from CT images and a corresponding model with simulated defects on it. Our approach can learn the spatial distribution of the upper part of normal cranial bones and use flawed cranial data to predict its complete geometry. Empirical research on simulated defects and actual clinical applications shows that our framework can meet most of the requirements of cranioplasty.

Clinical applications of machine learning in predicting 3D shapes of the human body: a systematic review

“…1 ). Following abstract and title screening, 7664 articles were removed, leaving 85 articles for full-text screening.12 articles met the inclusion criteria [ 10 – 21 ] including one identified through hand searching [ 14 ]. All 12 articles were published between 2015 to 2022.…”

“…All 12 articles were published between 2015 to 2022. Six studies were identified as unspecific cohort studies [ 10 , 11 , 15 , 18 , 19 , 21 ], five were retrospective cohort studies [ 12 , 14 , 16 , 17 , 20 ], and one was a case study [ 13 ]. 10 studies described approval from a relevant human ethics committee, however this was not mentioned in two articles [ 10 , 15 ].…”

“…10 studies described approval from a relevant human ethics committee, however this was not mentioned in two articles [ 10 , 15 ]. These studies predicted 3D shapes of a wide range of body regions, including the face [ 11 , 14 , 18 – 20 ], brain [ 10 , 15 ], head [ 16 , 21 ], forearm [ 12 ], aorta [ 13 ] and breast [ 17 ]. Sample sizes ranged from n = 7 to 209.…”

“…The programming language was mentioned by 58% of the included papers. Python [ 14 , 18 , 20 ], C++ [ 11 , 16 , 19 ], and MATLAB [ 11 , 18 , 20 , 21 ] were three languages used for building and running programs, with MATLAB the most prevalent. However, none of the articles published source code which would bring difficulties to open science and method replication.…”

Re: Corrección Tardía de Deformidades de Órbita Usando Implantes Personalizados 3D

“…Específicamente en el manejo de trauma craneofacial, Wang et al (2022) desarrollaron un algoritmo compuesto por dos redes neuronales convolucionales, el cual se entreno y validó, permitiendo detectar y clasificar diferentes tipos de fracturas mandibulares a través de Tomografías Computarizadas. Por otro lado, Wu et al (2022), generaron un modelo de capas neuronales con la capacidad de aprender la distribución espacial del segmento superior de los huesos craneales y utilizar los datos para predecir su geometría completa, para así reconstruir digitalmente los segmentos afectados por trauma y de esa forma generar la prótesis de reemplazo automáticamente e imprimirla de inmediato con la utilización de una impresora 3D, evitando pasos extra como la impresión del biomodelo previamente expuesto.…”