Paper Introduction

1.	Introduction
In the current medical situation, the design and manufacturing of cranial implants are performed through third-party softwares by some external suppliers. This involves costly commercial softwares and trained professional designers [1]. The three-dimensional reconstruction of the craniofacial region is a challenge if involving a very large or complicated defect, it must requires a sophisticated proprietary image processing and expensive computer-aided design software. Previously, it was very difficult to design large and complicated defects as it must be performed by 2D imaging manually, this process was very time take and cost effective [2]. In the recent years, additive manufacturing technology becomes very advanced, an implant that exactly fits the defects can be manufactured pre-operatively from the radiographic data obtained from CT scans.
Patient specific implants can be produced in any size with any shape and accurate fit using manual CAD technology [3]. A lot of work is already being done by many researchers using manual CAD/CAM methods in the field of design and manufacturing of implants. The research may vary from cranial, facial to hip implant design [4].
Design of Automatic Implants is a complicated task, however, many researchers and experts are working to develop some algorithms to help doctors in the designing of cranial implants automatically from CT scan data [5]. Artificial Intelligence (AI) techniques play a very important role in the field of medical science. It is now very useful to save time and money as a doctor can easily design an Implant sitting in an Operation Room.
Knoops et al. trained a model and validated on 4261 volunteers with machine learning frameworks. 95.5% patients were diagnosed with sensitivity [6]. Another researcher has developed an algorithm to integrate facial and intraoral images of anterior teeth using active contour model. The sophisticated landmark detection algorithm is able to detect facial key points. As a result, the proposed algorithm can recognize 96% of the teeth from the selected image set [7].
Li Jinniang et al. proposed a patch-based training strategy to design cranial implant design and is beneficial with high resolution data and spatially sparse and the training cold be used with other application such as 3D shape learning tasks [8].