Theapplication of machine learning in solving complex medical and non-medical challenges keeps growing. The use of CT scanners is very crucial in saving lives hence the need to investigate alternative approaches for solving artefacts found in Computer Tomographic images. In some instances, such faultstakes a while to figure out the types of artefacts and their causes due to large datasheets. It is, therefore, against this background that such a study needs to be investigated. This paper makes use of 180 image datasets and feature detection is applied on each dataset (ring and metal) and both datasets are trained for both 25 and 50 epoch test. After completion of both epochs, unknown datasets are inputted into the model. This research results thus concluded that the model is efficient with an accuracy of 87% and 91% respectively for both 25 and 50 epochs based on 170 image datasets. Furthermore, for 88 image dataset, the results shows the model accuracy of 80% and 90% respectively for both 25 and 50 epochs. As a result, these results demonstrates the stability of the model and shows that the more images with higher resolutions are in a dataset, the higher the accuracy.

Artefact, Machine Learning, CT Scanner, Image Processing, Datasets.