Artificial neural networks thrive in solving the classification problem for a particular rigid task, acquiring knowledge through generalized learning behaviour from a distinct training phase. The resulting network resembles a static entity of knowledge, with endeavours to extend this knowledge without targeting the original task resulting in a catastrophic forgetting. Continual learning shifts this paradigm towards networks that can continually accumulate knowledge over different tasks without the need to retrain from scratch. We focus on task incremental classification, where tasks arrive sequentially and are delineated by clear boundaries. Our main contributions concern: (1) a taxonomy and extensive overview of the state-of-the-art; (2) a novel framework to continually determine the stability-plasticity trade-off of the continual learner; (3) a comprehensive experimental comparison of 11 state-of-the-art continual learning methods; and (4) baselines. We empirically scrutinize method strengths and weaknesses on three benchmarks, considering Tiny Imagenet and large-scale unbalanced iNaturalist and a sequence of recognition datasets. We study the influence of model capacity, weight decay and dropout regularization, and the order in which the tasks are presented, and qualitatively compare methods in terms of required memory, computation time, and storage.

In this paper, a new computer tomography (CT) lung nodule computer-aided detection (CAD) method is proposed for detecting both solid nodules and ground-glass opacity (GGO) nodules (part solid and nonsolid). This method consists of several steps. First, the lung region is segmented from the CT data using a fuzzy thresholding method. Then, the volumetric shape index map, which is based on local Gaussian and mean curvatures, and the ldquodotrdquo map, which is based on the eigenvalues of a Hessian matrix, are calculated for each voxel within the lungs to enhance objects of a specific shape with high spherical elements (such as nodule objects). The combination of the shape index (local shape information) and ldquodotrdquo features (local intensity dispersion information) provides a good structure descriptor for the initial nodule candidates generation. Antigeometric diffusion, which diffuses across the image edges, is used as a preprocessing step. The smoothness of image edges enables the accurate calculation of voxel-based geometric features. Adaptive thresholding and modified expectation-maximization methods are employed to segment potential nodule objects. Rule-based filtering is first used to remove easily dismissible nonnodule objects. This is followed by a weighted support vector machine (SVM) classification to further reduce the number of false positive (FP) objects. The proposed method has been trained and validated on a clinical dataset of 108 thoracic CT scans using a wide range of tube dose levels that contain 220 nodules (185 solid nodules and 35 GGO nodules) determined by a ground truth reading process. The data were randomly split into training and testing datasets. The experimental results using the independent dataset indicate an average detection rate of 90.2%, with approximately 8.2 FP/scan. Some challenging nodules such as nonspherical nodules and low-contrast part-solid and nonsolid nodules were identified, while most tissues such as blood vessels were excluded. The method's high detection rate, fast computation, and applicability to different imaging conditions and nodule types shows much promise for clinical applications.

Although analog semantic communication systems have received considerable attention in the literature, there is less work on digital semantic communication systems. In this paper, we develop a deep learning (DL)-enabled vector quantized (VQ) semantic communication system for image transmission, named VQ-DeepSC. Specifically, we propose a convolutional neural network (CNN)-based transceiver to extract multi-scale semantic features of images and introduce multi-scale semantic embedding spaces to perform semantic feature quantization, rendering the data compatible with digital communication systems. Furthermore, we employ adversarial training to improve the quality of received images by introducing a PatchGAN discriminator. Experimental results demonstrate that the proposed VQ-DeepSC is more robustness than BPG in digital communication systems and has comparable MS-SSIM performance to the DeepJSCC method.

Segmentation of arterial wall boundaries from intravascular images is an important problem for many applications in the study of plaque characteristics, mechanical properties of the arterial wall, its 3-D reconstruction, and its measurements such as lumen size, lumen radius, and wall radius. We present a shape-driven approach to segmentation of the arterial wall from intravascular ultrasound images in the rectangular domain. In a properly built shape space using training data, we constrain the lumen and media-adventitia contours to a smooth, closed geometry, which increases the segmentation quality without any tradeoff with a regularizer term. In addition to a shape prior, we utilize an intensity prior through a nonparametric probability-density-based image energy, with global image measurements rather than pointwise measurements used in previous methods. Furthermore, a detection step is included to address the challenges introduced to the segmentation process by side branches and calcifications. All these features greatly enhance our segmentation method. The tests of our algorithm on a large dataset demonstrate the effectiveness of our approach.

Colorectal cancer is the third most common type of cancer worldwide. However, this disease can be prevented by detection and removal of precursor adenomatous polyps during optical colonoscopy (OC). During OC, the endoscopist looks for colon polyps. While hyperplastic polyps are benign lesions, adenomatous polyps are likely to become cancerous. Hence, it is a common practice to remove all identified polyps and send them to subsequent histological analysis. But removal of hyperplastic polyps poses unnecessary risk to patients and incurs unnecessary costs for histological analysis. In this paper, we develop the first part of a novel optical biopsy application based on narrow-band imaging (NBI). A barrier to an automatic system is that polyp classification algorithms require manual segmentations of the polyps, so we automatically segment polyps in colonoscopic NBI data. We propose an algorithm, Shape-UCM, which is an extension of the gPb-OWT-UCM algorithm, a state-of-the-art algorithm for boundary detection and segmentation. Shape-UCM solves the intrinsic scale selection problem of gPb-OWT-UCM by including prior knowledge about the shape of the polyps. Shape-UCM outperforms previous methods with a specificity of 92%, a sensitivity of 71%, and an accuracy of 88% for automatic segmentation of a test set of 87 images.

Image demoireing is a multi-faceted image restoration task involving both moire pattern removal and color restoration. In this paper, we raise a general degradation model to describe an image contaminated by moire patterns, and propose a novel multi-scale bandpass convolutional neural network (MBCNN) for single image demoireing. For moire pattern removal, we propose a multi-block-size learnable bandpass filters (M-LBFs), based on a block-wise frequency domain transform, to learn the frequency domain priors of moire patterns. We also introduce a new loss function named Dilated Advanced Sobel loss (D-ASL) to better sense the frequency information. For color restoration, we propose a two-step tone mapping strategy, which first applies a global tone mapping to correct for a global color shift, and then performs local fine tuning of the color per pixel. To determine the most appropriate frequency domain transform, we investigate several transforms including DCT, DFT, DWT, learnable non-linear transform and learnable orthogonal transform. We finally adopt the DCT. Our basic model won the AIM2019 demoireing challenge. Experimental results on three public datasets show that our method outperforms state-of-the-art methods by a large margin.