Region-based multifocus image fusion for the precise acquisition of Pap smear images

Abstract

A multifocus image fusion method to obtain a single focused image from a sequence of microscopic high-magnification Papanicolau source (Pap smear) images is presented. These images, captured each in a different position of the microscope lens, frequently show partially focused cells or parts of cells, which makes them unpractical for the direct application of image analysis techniques. The proposed method obtains a focused image with a high preservation of original pixels information while achieving a negligible visibility of the fusion artifacts. The method starts by identifying the best-focused image of the sequence; then, it performs a mean-shift segmentation over this image; the focus level of the segmented regions is evaluated in all the images of the sequence, and best-focused regions are merged in a single combined image; finally, this image is processed with an adaptive artifact removal process. The combination of a region-oriented approach, instead of block-based approaches, and a minimum modification of the value of focused pixels in the original images achieve a highly contrasted image with no visible artifacts, which makes this method especially convenient for the medical imaging domain. The proposed method is compared with several state-of-the-art alternatives over a representative dataset. The experimental results show that our proposal obtains the best and more stable quality indicators.

Medical Image Fusion Based on Low-Level Features

“…The regions with greater energy or larger changes of pixels are considered to be in focus during the fusion process. The spatial domain methods mainly include pixel-based methods [6][7][8] and region-based methods [9][10][11]. These methods are simple and fast.…”

“…The spatial domain methods deal with pixels or regions in spatial domains directly based on the pixel intensities [ 6 ]. The fundamental problem of spatial domain methods is the selection of the clearest image pixels or regions from the source images in order to construct the fused image.…”

Multi Focus Image Fusion with Region-Center Based Kernel

“…Then, the optimization method of kernel processing and subsampling sample is applied in [6]. The other method predicts the focus region through the mean shift algorithm to create the precise segmentation map [7]. Then, by using the image's information saliency, the method is supposed to keep the detailed information of multi-focus images in constructing the fused image [8].…”

Nucleus Segmentation of Cervical Cytology Images Based on Depth Information

“…Cervical nucleus segmentation involves certain challenges, such as the overlapping cervical cells, uneven staining, poor contrast, and presence of neutrophils. Many scholars have conducted extensive research in the nucleus segmentation, and some algorithms have been proposed, including threshold methods [3]- [5], watershed methods [16]- [19], morphological algorithms [6]- [8], deformable models [9]- [13], template-matching algorithms [14], [15], graph-based segmentation algorithms [20], [21], region-based methods [22]- [25], clustering methods [26]- [29], and neural network methods [30]- [32].…”

“…The clustering algorithm is an effective object segmentation method used by CAD system. Tello-Mijares et al combined the multi-focus fusion algorithm with a mean-shift clustering segmentation algorithm to segment the bestfocused cervical images [26], [27]. Fuzzy C-Means (FCM) algorithm and its improved algorithm were used extensively in segmenting brain magnetic resonance images.…”