Computer-aided diagnosis of keratoconus through VAE-augmented images using deep learning.

Detecting clinical keratoconus (KCN) poses a challenging and time-consuming task. During the diagnostic process, ophthalmologists are required to review demographic and clinical ophthalmic examinations in order to make an accurate diagnosis. This study aims to develop and evaluate the accuracy of deep convolutional neural network (CNN) models for the detection of keratoconus (KCN) using corneal topographic maps. We retrospectively collected 1758 corneal images (978 normal and 780 keratoconus) from 1010 subjects of the KCN group with clinically evident keratoconus and the normal group with regular astigmatism. To expand the dataset, we developed a model using Variational Auto Encoder (VAE) to generate and augment images, resulting in a dataset of 4000 samples. Four deep learning models were used to extract and identify deep corneal features of original and synthesized images. We demonstrated that the utilization of synthesized images during training process increased classification performance. The overall average accuracy of the deep learning models ranged from 99% for VGG16 to 95% for EfficientNet-B0. All CNN models exhibited sensitivity and specificity above 0.94, with the VGG16 model achieving an AUC of 0.99. The customized CNN model achieved satisfactory results with an accuracy and AUC of 0.97 at a much faster processing speed compared to other models. In conclusion, the DL models showed high accuracy in screening for keratoconus based on corneal topography images. This is a development toward the potential clinical implementation of a more enhanced computer-aided diagnosis (CAD) system for KCN detection, which would aid ophthalmologists in validating the clinical decision and carrying out prompt and precise KCN treatment.

Minimally invasive and invasive liver surgery based on augmented reality training: a review of the literature.

A liver surgeon's knowledge of anatomy is critical. Due to the patient's small field of vision, patient specific, complex nerve system, and other factors, even a minor loss can result in irreversible damage. Surgeons could benefit from the use of augmented reality (AR) technology, which would bring three-dimensional image data into the operating room. AR visualization can improve surgical procedures, facilitate intraoperative planning, and enhance surgical guidance for the anatomy of interest, all of which contribute to the application's minimal invasiveness. This literature review on image guidance in liver surgery provides the reader with information about AR techniques. To ascertain the current state of Augmented reality technology's application in liver surgery, a PubMed and Embase search were conducted using the following keywords: < (Augmented reality) AND (liver surgery) > and < 'Augmented reality' AND 'liver surgery' > (publication date from January 1991 until Jun 2022). The query yielded a total of 205 publications-excluded papers in other languages, virtual reality (VR), and reviews leaving 135 studies for review. After removing duplication, the titles and abstracts of those studies were manually reviewed. Finally, 31 pertinent studies were determined to be pertinent to the subject. Generally, augmented reality technology includes preoperative planning and three-dimensional reconstruction, intraoperative three-dimensional navigation, and registration. Visualization may be aided by virtual three-dimensional reconstruction models of the liver from Computed Tomography/Magnetic Resonance Imaging scans. The results demonstrate that by utilizing augmented reality technology, blood vessels and tumor structures in the liver can be visualized during surgery, allowing for precise navigation during complicated surgical procedures. Augmented reality has been demonstrated to be safe and effective in both minimally invasive and invasive liver surgery. With recent advancements and significant effort by liver surgeons, augmented reality technologies have the potential to increase hepatobiliary surgical procedures dramatically. However, further clinical trials will be necessary to evaluate augmented reality as a tool for reducing post-operative morbidity and mortality. The impact of these cutting-edge computerized image guidance techniques on clinically relevant outcome parameters should be assessed in the future.

Cardiac MR images of thalassemia major patients with myocardial iron overload: a data note.

OBJECTIVE: Patients with thalassemia major (TM) have the highest mortality rate due to heart failure induced by myocardial iron overload. However, T2* weighted MR imaging is currently a gold standard approach for measuring iron overload. Examining ventricular volumes with magnetic resonance imaging (MR imaging) and measuring myocardial iron overload in TM patients allows for an early prediction of heart failure. This dataset includes cardiac MR images of TM patients and the control group with clinical and echocardiographic data. This dataset may be useful to researchers investigating myocardial iron overload. This dataset can also be used for medical image processing applications, such as ventricle segmentation. DATA DESCRIPTION: This study provides open-source cardiac MR images of 50 subjects and clinical and echocardiographic data. From February 2016 to January 2019, all images and clinical data were obtained from the MRI department of a general hospital in Mashhad, Iran. All the images are 16-bit gray-scale and stored in DICOM format. All patient-specific information is removed from image headers to preserve patient privacy. In addition, all images associated with each subject are compressed and saved in the RAR format.

Evaluation of High-Efficiency Image Coding algorithm for dermatology images in teledermatology.

BACKGROUND: Currently, teledermatology assumes a progressively greater role in the modern healthcare system, especially in consultation, diagnosis, or examining lesions and skin cancers. One of the major challenges facing teledermatology systems is determining the optimal image compression method to efficiently reduce the space needed for electronic storage and data transmission. OBJECTIVE: To the objective and subjective assessment of HEIC compression method on dermatological color images and benchmarking the performance of High-Efficiency Image Coding (HEIC) with different algorithms to a feasibility study of the method for teledermatology. METHODS: Twenty-five clinical and five skin histopathology images were taken in department of dermatology, Imam Reza Hospital, Mashhad, Iran. For each image, a set of 24 compressed images with different compression rates, which is composed of eight JPEG, eight JPEG2000, and eight HEIC images, has been prepared. Compressed and original images were shown simultaneously to three dermatologists and one dermatopathologist with different experiences. Each dermatologist scored quality and suitability of compressed images for diagnostic, as well as educational/scientific purposes. An objective evaluation was performed by calculating the mean "distance" of pixel colors and peak signal-to-noise ratio (PSNR). RESULTS: All compression rates for HEIC were objectively better than JPEG and JPEG2000, particularly at PSNR. Moreover, mean "color distance" per pixel for compressed images using HEIC was lower than others. The subjective image quality assessment also confirms the results of objective evaluation. In both educational and clinical diagnostic applications, HEIC compressed images have the highest score. CONCLUSION: In consideration of objective and subjective evaluation, the HEIC algorithm represents an optimal performance in dermatology images compression compared with JPEG and JPEG2000.

Facial skin segmentation using bacterial foraging optimization algorithm.

Nowadays, analyzing human facial image has gained an ever-increasing importance due to its various applications. Image segmentation is required as a very important and fundamental operation for significant analysis and interpretation of images. Among the segmentation methods, image thresholding technique is one of the most well-known methods due to its simplicity, robustness, and high precision. Thresholding based on optimization of the objective function is among the best methods. Numerous methods exist for the optimization process and bacterial foraging optimization (BFO) is among the most efficient and novel ones. Using this method, optimal threshold is extracted and then segmentation of facial skin is performed. In the proposed method, first, the color facial image is converted from RGB color space to Improved Hue-Luminance-Saturation (IHLS) color space, because IHLS has a great mapping of the skin color. To perform thresholding, the entropy-based method is applied. In order to find the optimum threshold, BFO is used. In order to analyze the proposed algorithm, color images of the database of Sahand University of Technology of Tabriz, Iran were used. Then, using Otsu and Kapur methods, thresholding was performed. In order to have a better understanding from the proposed algorithm; genetic algorithm (GA) is also used for finding the optimum threshold. The proposed method shows the better results than other thresholding methods. These results include misclassification error accuracy (88%), non-uniformity accuracy (89%), and the accuracy of region's area error (89%).