Search

Wireless capsule endoscopy is a non-invasive medical imaging modality used for diagnosing and monitoring digestive tract diseases. However, the analysis of images obtained from wireless capsule endoscopy is a challenging task, as the images are of low resolution and often contain a large number of artifacts. In recent years, deep learning has shown great promise in the analysis of medical images, including wireless capsule endoscopy images. This paper provides a review of the current trends and future directions in deep learning for wireless capsule endoscopy. We focus on the recent advances in transfer learning, attention mechanisms, multi-modal learning, automated lesion detection, interpretability and explainability, data augmentation, and edge computing. We also highlight the challenges and limitations of current deep learning methods and discuss the potential future directions for the field. Our review provides insights into the ongoing research and development efforts in the field of deep learning for wireless capsule endoscopy, and can serve as a reference for researchers, clinicians, and engineers working in this area inspection process.

Wireless capsule endoscopy (WCE) is a huge step forward in diagnosing problems in the digestive system because it can take pictures without hurting the patient. As part of this study, we thoroughly test anomaly detection models that were created with the specific goal of finding WCE disease. The Single Shot Multibox Detector (SSD), EfficientDet, Faster RCNN, RetinaNet, yolov3, Real-Time Multi-Detector (RTMDet), and semi-supervised learning detection models were tested on a translated MS COCO dataset from the Kvasir-Capsule repository. The dataset has a variety of gastrointestinal abnormalities. We used three main evaluation methods: normalized confusion matrices for class-specific accuracy, macro and weighted averages(precision, recall, and F-score). We compared normal and semi-supervised versions of models like Faster RCNN to find out what the benefits of semi-supervised methods are. The models that were tested showed that RTMDet-S and RTMDet-Tiny were the most accurate in their class, finding complex diseases with over 99% accuracy. With an precision, recall, and F-score of 99.67%, SSD300 also had the best total result among statistics measures. Along with that, YOLOv3 was the model that worked the best in terms of operating economy. It achieved a frame rate of 92.1 frames per second while using only 376MiB of memory. There were big improvements in sensitivity and specificity to 98.70% and 99.84% for semi-supervised models like Soft-teacher (Faster RCNN + ResNet).The purpose of this study is to assist researchers and clinicians in improving their diagnostic abilities by providing highly informative information on the most effective object detection models that can be used for the identification of WCE pathology.

Background: Manufacturing markings printed on products play an important role in the handling and use of pharmaceuticals and perishable foods. Currently, optical character recognition, neural networks, deep learning-based methods, and combinations of these methods are used to recognize these codes. Methods: This systematic review was performed to find papers that can answer the following research questions: How have machine vision methods that can recognize product texts evolved over the past eight years? What are the most common difficulties in recognizing product texts? Articles published between 2012 and 2020 were systematically searched from Science Direct/SCOPUS, and Google Scholar in November-December 2020. Ten studies were eligible, with inclusion criteria: details about the recognition method used, performance analysis result, imaging method, product and the text printed on it. Results: Product text recognition methods have evolved significantly over the last two years to tolerate the most common difficulties in the field. This is due to the ability of the deep learning neural network (DNN) architectures such as convolutional neural networks (CNN) to extract and learn salient character features straight from packaging surface images. Four of the most recent methods use two consecutive deep learning networks, one detecting the text area based on an image captured from the product package's surface and the other recognizing the characters within. Furthermore, this paper presents solutions to the most common product text recognition difficulties. Conclusions: There were a limited number of studies that met the eligibility criteria for this systematic review. The study's aim was to evaluate the development of machine vision methods for recognizing manufacturing marking texts printed on the surface of products. The study results demonstrated how methods have evolved over time, beginning with optical character recognition, and advancing to methods which can recognize texts despite the field's most common problems.

In this paper, we focus on presenting a novel AI-based service platform proposal called AIDI (Artificial Intelligence Distribution Interface for healthcare). AIDI proposal is based on our earlier research work in which we evaluated AI-based healthcare services which have been used successfully in practice among healthcare service providers. We have also used our systematic review about AI-based healthcare services benefits in various healthcare sectors. This novel AIDI proposal contains services for health assessment, healthcare evaluation, and cognitive assistant which can be used by researchers, healthcare service provides, clinicians, and consumers. AIDI integrates multiple health databases and data lakes with AI service providers and open access AI algorithms. It also gives healthcare service providers open access to state-of-the-art AI-based diagnosis and analysis services. This paper provides a description of AIDI platform, how it could be developed, what can become obstacles in the development, and how the platform can provide benefits to healthcare when it will be operational in the future.

An important trend in meeting the needs of modern caregiving is providing a well-rounded care delivery. In doing this, access to data of care receiver is important. This, however, has been shown to be possible with connected health with the aid of modern technology. In this paper, we are presenting a design that will expand on the opportunities for better data accessibility and use, by integrating the recommendation system into connected health. In order to ensure a design that meets the needs of care receivers, we conducted two independent surveys with a view to gathering requirements for the design. The result has shown the relevance of timeliness in caregiving along with what the care receivers are interested in such as what their data say about their conditions, the immediate and future expectation of their conditions, and collaborative efforts in managing their conditions. The requirements gathered were used in the design and implementation of the mobile app called Recommendations Sharing Community for Aged and Chronically Ill People (ReSCAP). The benefits of this solution have been real-time recommendations' availability, cost saving, adequate care, and access to health data and information for better care decision-making.