A novel binary feature descriptor to discriminate normal and abnormal chest CT images using dissimilarity measures

New design of adaptive Gabor wavelet filter bank for medical image retrieval

“…Table 1 shows the ARP as well as ARR values for each feature descriptor methods over NEMA database. The experimental results show that the performance of the proposed method has been improved remarkably as compared to the other existing methods [4, 10–12, 22, 27]. It is observed that though NEMA database is a collection of different categories from different parts of the body, the proposed method performs better in retrieving the images in all categories of the dataset.…”

“…The methods such as LBP [4] and Yelampalli et al . [12] are implemented by thresholding the centre pixel from its neighbourhood pixels where as Murala et al . [10] is implemented by considering a neighbourhood.…”

“…The spectrum helps in finding the dominant texture directions of the MRI image by analysing the peaks and the prominent peaks represent the dominant texture orientations present in the medical image. The comparison of ARP and ARR values of the proposed method with existing methods [4, 10–12, 22, 27] for OASIS database is listed in Table 1. It is observed from the Table 1 that the proposed method has an ARP value of 85.24% and an ARR value of 14.05% which are better than the existing methods.…”

“…However, STFT uses fixed window size for which there is a low time–frequency resolution. Therefore, several filters including different wavelet filters have been proposed to provide space‐frequency signal decomposition [11–19].…”

“…[11] used Daubechies wavelet‐based local feature descriptor for the registration of medical images. Same authors used a binary feature descriptor to discriminate normal and abnormal chest computed tomography (CT) images [12]. In their next method [13], the registration of medical images is based on binary feature descriptor and modified affine transform.…”