A robust head MRI/CT background removing approach using dynamic morphological operations

Abstract

The objective of the work is to remove the CT table or other noises from Head CT/MRI images, which are often present and effectively decrease the efficiency of further intended image processing operations. At first, depending on the mass an adaptive ellipse-shaped mask is generated to segment out the Region of interest (ROI) area. Then an initial head mask has been generated using iterative morphology. Finally, head mask boundary coordinates are corrected accordingly to fully cover the head portion. The result section shows that the proposed method is more effective on normal brain MRI than skull stripped MRIs. The proposed method performs well in comparison with static morphology-based methods. For MRI with skull images, the score is 97.03% and for skull stripped MRI images the score is about 91%. In the case of brain CT, the proposed method achieved almost 90% Jaccard coefficient similarity index value. These results prove the efficiency and accuracy of the proposed method the results are compared with the existing well-known methods of Atkins et al. (IEEE Trans Med Imaging, 17(1), 98–107, https://doi.org/10.1109/42.668699 , 1998), Liu et al. (EURASIP J Image Video Process, 2017(1), https://doi.org/10.1186/s13640-017-0209-y , 2017), Mol et al. (IOP Conf Ser Mater Sci Eng 396, 012039, https://doi.org/10.1088/1757-899x/396/1/012039 , 2018), Chen et al. (Real-time patient table removal in CT images. Health Information Science, pp 1–8, Springer, https://doi.org/10.1007/978-3-319-48335-1_1 , 2016) and Maureenvan Eijnatten et al. (Comput Methods Progr Biomed 208, 106261, https://doi.org/10.1016/j.cmpb.2021.106261 , 2021). The proposed method can remove any kind of background noises from both CT and MRI images. Here, the value of morphological operators is determined dynamically through iteration. Moreover, this work proposed a novel CT patient table removing approach from Brain CT images often are undesirable part of the medical image.