Cefa Karabağ

Today, unmanned aerial vehicles (UAVs) are heavily dependent on Global Navigation Satellite Systems (GNSSs) for positioning and navigation. However, GNSS signals are vulnerable to jamming and spoofing attacks. This poses serious security risks, especially for military operations and critical civilian missions. In order to solve this problem, an image-based geolocation system has been developed that eliminates GNSS dependency. The proposed system estimates the geographical location of the UAV by matching the aerial images taken by the UAV with previously georeferenced high-resolution satellite images. For this purpose, common visual features were determined between satellite and UAV images and matching operations were carried out using methods based on the homography matrix. Thanks to image processing, a significant relationship has been established between the area where the UAV is located and the geographical coordinates, and reliable positioning is ensured even in cases where GNSS signals cannot be used. Within the scope of the study, traditional methods such as SIFT, AKAZE, and Multiple Template Matching were compared with learning-based methods including SuperPoint, SuperGlue, and LoFTR. The results showed that deep learning-based approaches can make successful matches, especially at high altitudes.

Staining of histological slides with Hematoxylin and Eosin is widely used in clinical and laboratory settings as these dyes reveal nuclear structures as well as cytoplasm and collagen. For cancer diagnosis, these slides are used to recognize tissues and morphological changes. Tissue semantic segmentation is therefore important and at the same time a challenging and time-consuming task. This paper describes a UNet-like deep learning architecture called DRD-UNet , which adds a novel processing block called DRD (Dilation, Residual, and Dense block) to a UNet architecture. DRD is formed by the combination of dilated convolutions (D), residual connections (R), and dense layers (D). DRD-UNet was applied to the multi-class (tumor, stroma, inflammatory, necrosis, and other) semantic segmentation of histological images from breast cancer samples stained with Hematoxylin and Eosin. The histological images were released through the Breast Cancer Semantic Segmentation (BCSS) Challenge. DRD-UNet outperformed the original UNet architecture and 15 other UNet-based architectures on the segmentation of 12,930 image patches extracted from regions of interest that ranged in size between 1036 × 1222 to 6813 × 7360 pixels. DRD-UNet obtained the best performance as measured with Jaccard similarity index, Dice coefficient, in a per-class comparison and accuracy for overall segmentation.

This paper describes a methodology to analyse the complexity of HeLa cells as observed with electron microscopy, in particular the relationship between mitochondria and the roughness of the nuclear envelope as reflected by the invaginations of the surface. For this purpose, several segmentation mitochondria algorithms were quantitatively compared, namely: Topology, Image Processing, Topology and Image Processing, and Deep Learning, which provided the highest accuracy. The invaginations were successfully segmented with one image processing algorithm. Metrics were extracted for both structures and correlations between the mitochondria and invaginations were explored for 25 segmented cells. It was found that there was a positive correlation between the volume of invaginations and the volume of mitochondria, and negative correlations between the number and the mean volume of mitochondria, and between the volume of the cytoplasm and the aspect ratio of mitochondria. These results suggest that there is a relationship between the shape of a cell, its nucleus and its mitochondria; as well as a relationship between the number of mitochondria and their shapes. Whilst these results were obtained from a single cell line and a relatively small number of cells, they encourage further study as the methodology proposed can be easily applied to other cells and settings. Code and data are freely available. HeLa images are available from http://dx.doi.org/10.6019/EMPIAR-10094 , code from https://github.com/reyesaldasoro/MitoEM , and segmented nuclei, cells, invaginations and mitochondria from https://github.com/reyesaldasoro/HeLa_Cell_Data .

This paper describes the quantitative and qualitative evaluation of training data, ground truth and the variability of the shape of HeLa cells as observed with electron microscopy in the process of semantic segmentation. HeLa cells are a widely used immortal cell line, but the principles described here could apply to other cells. From a data set of 8, 000 × 8, 000 × 517 voxels, a region of interest (ROI) of 2, 000 × 2, 000 × 300 was cropped for quantitative evaluation with the qualitative evaluation performed on a larger section of interest. Patches of data and labels were used to train a U-Net architecture from scratch, which was compared also with a traditional image processing algorithm. The correctness of ground truth, that is, the inclusion of one or more nuclei within a region of interest was evaluated with a hand-drawn ground truth of the ROI. The impact of the extent of training data was evaluated by comparing results from 36,000 data and label patches extracted from the odd slices in the central region, to 135,000 patches obtained from every other slice in the set. Accuracy and Jaccard similarity index were higher for the larger training set with multiple nuclei in the ground truth. However, when the U-Nets were used to segment the section of interest beyond the ROI, results were not as good as expected, suggesting that the very complicated nature of the shapes of HeLa cells cannot be learned from a single cell. Data, code and ground truth have been made publicly released through Empair, GitHub and Zenodo.

This paper investigates the classification of radiographic images with eleven convolutional neural network (CNN) architectures (GoogleNet, VGG-19, AlexNet, SqueezeNet, ResNet-18, Inception-v3, ResNet-50, VGG-16, ResNet-101, DenseNet-201 and Inception-ResNet-v2). The CNNs were used to classify a series of wrist radiographs from the Stanford Musculoskeletal Radiographs (MURA) dataset into two classes—normal and abnormal. The architectures were compared for different hyper-parameters against accuracy and Cohen’s kappa coefficient. The best two results were then explored with data augmentation. Without the use of augmentation, the best results were provided by Inception-ResNet-v2 (Mean accuracy = 0.723, Mean kappa = 0.506). These were significantly improved with augmentation to Inception-ResNet-v2 (Mean accuracy = 0.857, Mean kappa = 0.703). Finally, Class Activation Mapping was applied to interpret activation of the network against the location of an anomaly in the radiographs.

This paper investigates the classification of radiographic images with eleven convolutional neural network (CNN) architectures ( GoogleNet, VGG-19, AlexNet, SqueezeNet, ResNet-18, Inception-v3, ResNet-50, VGG-16, ResNet-101, DenseNet-201 and Inception-ResNet-v2 ). The CNNs were used to classify a series of wrist radiographs from the Stanford Musculoskeletal Radiographs (MURA) dataset into two classes - normal and abnormal. The architectures were compared for different hyper-parameters against accuracy and Cohen’s kappa coefficient. The best two results were then explored with data augmentation. Without the use of augmentation, the best results were provided by Inception-Resnet-v2 (Mean accuracy = 0.723 , Mean kappa = 0.506 ). These were significantly improved with augmentation to Inception-Resnet-v2 (Mean accuracy = 0.857 , Mean kappa = 0.703 ). Finally, Class Activation Mapping was applied to interpret activation of the network against the location of an anomaly in the radiographs.