Classification of diabetic retinopathy using unlabeled data and knowledge distillation.

Over the last decade, advances in Machine Learning and Artificial Intelligence have highlighted their potential as a diagnostic tool in the healthcare domain. Despite the widespread availability of medical images, their usefulness is severely hampered by a lack of access to labeled data. For example, while Convolutional Neural Networks (CNNs) have emerged as an essential analytical tool in image processing, their impact is curtailed by training limitations due to insufficient labeled data availability. Transfer Learning enables models developed for one task to be reused for a second task. Knowledge distillation enables transferring knowledge from a pre-trained model to another. However, it suffers from limitations, and the two models' constraints need to be architecturally similar. Knowledge distillation addresses some of the shortcomings of transfer learning by generalizing a complex model to a lighter model. However, some parts of the knowledge may not be distilled by knowledge distillation sufficiently. In this paper, a novel knowledge distillation approach using transfer learning is proposed. The proposed approach transfers the complete knowledge of a model to a new smaller one. Unlabeled data are used in an unsupervised manner to transfer the new smaller model's maximum amount of knowledge. The proposed method can be beneficial in medical image analysis, where labeled data are typically scarce. The proposed approach is evaluated in classifying images for diagnosing Diabetic Retinopathy on two publicly available datasets, including Messidor and EyePACS. Simulation results demonstrate that the approach effectively transfers knowledge from a complex model to a lighter one. Furthermore, experimental results illustrate that different small models' performance is improved significantly using unlabeled data and knowledge distillation.

Detection of COVID-19 in X-ray images by classification of bag of visual words using neural networks.

Coronavirus disease 2019 (COVID-19) was classified as a pandemic by the World Health Organization in March 2020. Given that this novel virus most notably affects the human respiratory system, early detection may help prevent severe lung damage, save lives, and help prevent further disease spread. Given the constraints on the healthcare facilities and staff, the role of artificial intelligence for automatic diagnosis is critical. The automatic diagnosis of COVID-19 based on medical images is, however, not straightforward. Due to the novelty of the disease, available X-ray datasets are very limited. Furthermore, there is a significant similarity between COVID-19 X-rays and other lung infections. In this paper, these challenges are addressed by proposing an approach consisting of a bag of visual words and a neural network classifier. The proposed method can classify X-ray chest images into non-COVID-19 and COVID-19 with high performance. Three public datasets are used to evaluate the proposed approach. Our best accuracy on the first, second, and third datasets is 96.1, 99.84, and 98 percent. Since detection of COVID-19 is important, sensitivity is used as a criterion. The proposed method's best sensitivities are 90.32, 99.65, and 91 percent on these datasets, respectively. The experimental results show that extracting features with the bag of visual words results in better classification accuracy than the state-of-the-art techniques.

Dense pooling layers in fully convolutional network for skin lesion segmentation.

One of the essential tasks in medical image analysis is segmentation and accurate detection of borders. Lesion segmentation in skin images is an essential step in the computerized detection of skin cancer. However, many of the state-of-the-art segmentation methods have deficiencies in their border detection phase. In this paper, a new class of fully convolutional network is proposed, with new dense pooling layers for segmentation of lesion regions in skin images. This network leads to highly accurate segmentation of lesions on skin lesion datasets, which outperforms state-of-the-art algorithms in the skin lesion segmentation.

Hierarchical Pruning for Simplification of Convolutional Neural Networks in Diabetic Retinopathy Classification.

Convolutional neural networks (CNNs) are widely used in automatic detection and analysis of diabetic retinopathy (DR). Although CNNs have proper detection performance, their structural and computational complexity is troublesome. In this study, the problem of reducing CNN's structural complexity for DR analysis is addressed by proposing a hierarchical pruning method. The original VGG16-Net is modified to have fewer parameters and is employed for DR classification. To have an appropriate feature extraction, pre-trained model parameters on Image-Net dataset are used. Hierarchical pruning gradually eliminates the connections, filter channels, and filters to simplify the network structure. The proposed pruning method is evaluated using the Messidor image dataset which is a public dataset for DR classification. Simulation results show that by applying the proposed simplification method, 35% of the feature maps are pruned resulting in only 1.89% accuracy drop. This simplification could make CNN suitable for implementation inside medical diagnostic devices.

Gland Segmentation in Histopathology Images Using Deep Networks and Handcrafted Features.

Histopathology images contain essential information for medical diagnosis and prognosis of cancerous disease. Segmentation of glands in histopathology images is a primary step for analysis and diagnosis of an unhealthy patient. Due to the widespread application and the great success of deep neural networks in intelligent medical diagnosis and histopathology, we propose a modified version of LinkNet for gland segmentation and recognition of malignant cases. We show that using specific handcrafted features such as invariant local binary pattern drastically improves the system performance. The experimental results demonstrate the competency of the proposed system against the state-of-the-art methods. We achieved the best results in testing on section B images of the Warwick-QU dataset and obtained comparable results on section A images.

Liver Segmentation in Abdominal CT Images Using Probabilistic Atlas and Adaptive 3D Region Growing.

Automatic liver segmentation plays a vital role in computer-aided diagnosis or treatment. Manual segmentation of organs is a tedious and challenging task and is prone to human errors. In this paper, we propose innovative pre-processing and adaptive 3D region growing methods with subject-specific conditions. To obtain strong edges and high contrast, we propose effective contrast enhancement algorithm then we use the atlas intensity distribution of most probable voxels in probability maps along with location before designing conditions for our 3D region growing method. We also incorporate the organ boundary to restrict the region growing. We compare our method with the label fusion of 13 organs on state-of-the-art Deeds registration method and achieved Dice score of 92.56%.

Fetal Ultrasound Image Segmentation for Measuring Biometric Parameters Using Multi-Task Deep Learning.

Ultrasound imaging is a standard examination during pregnancy that can be used for measuring specific biometric parameters towards prenatal diagnosis and estimating gestational age. Fetal head circumference (HC) is one of the significant factors to determine the fetus growth and health. In this paper, a multi-task deep convolutional neural network is proposed for automatic segmentation and estimation of HC ellipse by minimizing a compound cost function composed of segmentation dice score and MSE of ellipse parameters. Experimental results on fetus ultrasound dataset in different trimesters of pregnancy show that the segmentation results and the extracted HC match well with the radiologist annotations. The obtained dice scores of the fetal head segmentation and the accuracy of HC evaluations are comparable to the state-of-the-art.

Deep Neural Network based Polyp Segmentation in Colonoscopy Images using a Combination of Color Spaces.

Colorectal cancer (CRC) is the second leading cause of cancer death. Colorectal polyps cause most colorectal cancer cases. Colonoscopy is considered as the most common method for diagnosis of colorectal polyps, and early detection and segmentation of them can prevent colorectal cancer. On the other hand, today advances in computer systems persuade researchers all around the world to use computer-aided systems to help physicians in their diagnosis. Many modern types of researches and methods have proposed for this goal, and we have aggregated the methods based on previous convolutional neural networks with more recent networks in this paper to improve the quality of segmentation. We also chose the red channel, green channel and the b* component of CIE-L*a*b* as the input of network to leverage the parameters of segmentation result such as dice and sensitivity. LinkNet is used as the convolutional network, and the results show that it is suitable for segmentation. Performance of our method is evaluated on CVC-ColonDB. The results show that our method outperforms previous works in colorectal polyp segmentation field.

Low Complexity CNN Structure for Automatic Bleeding Zone Detection in Wireless Capsule Endoscopy Imaging.

Wireless capsule endoscopy (WCE) is a swallowable device used for screening different parts of the human digestive system. Automatic WCE image analysis methods reduce the duration of the screening procedure and alleviate the burden of manual screening by medical experts. Recent studies widely employ convolutional neural networks (CNNs) for automatic analysis of WCE images; however, these studies do not consider CNN's structural and computational complexities. In this paper, we address the problem of simplifying the CNN's structure. A low complexity CNN structure for bleeding zone detection is proposed which takes a single patch as input and then outputs a segmented patch of the same size. The proposed network is inspired by the FCN paradigm with a simplified structure. Since it is based on image patches, the resulting network benefits from moderate-sized intermediate feature maps. Moreover, the problem of redundant computations in patch-based methods is circumvented by non-overlapping patch processing. The proposed method is evaluated using the publicly available KID dataset for WCE image analysis. Experimental results show that the proposed network has better accuracy and AUC than previous structures while requiring less computational operations.

Lossless Compression of Angiogram Foreground with Visual Quality Preservation of Background.

By increasing the volume of telemedicine information, the need for medical image compression has become more important. In angiographic images, a small ratio of the entire image usually belongs to the vasculature that provides crucial information for diagnosis. Other parts of the image are diagnostically less important and can be compressed with higher compression ratio. However, the quality of those parts affects the overall understanding of the image as well. Existing methods compress foreground and background of angiographic images using different techniques. In this paper, we first utilize a convolutional neural network to segment vessels and then represent a hierarchical block processing algorithm capable of both eliminating the background redundancies and preserving the overall visual quality of angiograms.

Classification of Informative Frames in Colonoscopy Videos Using Convolutional Neural Networks with Binarized Weights.

Colorectal cancer is one of the common cancers in the United States. Polyps are one of the major causes of colonic cancer, and early detection of polyps will increase the chance of cancer treatments. In this paper, we propose a novel classification of informative frames based on a convolutional neural network with binarized weights. The proposed CNN is trained with colonoscopy frames along with the labels of the frames as input data. We also used binarized weights and kernels to reduce the size of CNN and make it suitable for implementation in medical hardware. We evaluate our proposed method using Asu Mayo Test clinic database, which contains colonoscopy videos of different patients. Our proposed method reaches a dice score of 71.20% and accuracy of more than 90% using the mentioned dataset.

Polyp Segmentation in Colonoscopy Images Using Fully Convolutional Network.

Colorectal cancer is one of the highest causes of cancer-related death, especially in men. Polyps are one of the main causes of colorectal cancer, and early diagnosis of polyps by colonoscopy could result in successful treatment. Diagnosis of polyps in colonoscopy videos is a challenging task due to variations in the size and shape of polyps. In this paper, we proposed a polyp segmentation method based on the convolutional neural network. Two strategies enhance the performance of the method. First, we perform a novel image patch selection method in the training phase of the network. Second, in the test phase, we perform effective post-processing on the probability map that is produced by the network. Evaluation of the proposed method using the CVC-ColonDB database shows that our proposed method achieves more accurate results in comparison with previous colonoscopy video-segmentation methods.

Reversible Image Watermarking for Health Informatics Systems Using Distortion Compensation in Wavelet Domain.

Reversible image watermarking guaranties restoration of both original cover and watermark logo from the watermarked image. Capacity and distortion of the image under reversible watermarking are two important parameters. In this study, a reversible watermarking is investigated by focusing on increasing the embedding capacity and reducing the distortion in medical images. We use integer wavelet transform for embedding one bit of watermark in a transform coefficient. We devise a novel approach that when a coefficient is modified in one iteration, the produced distortion is compensated in the next iteration. This distortion compensation method would result in low distortion rate. The proposed method is tested on four types of medical images including MRI of the brain, cardiac MRI, MRI of breast and intestinal polyp images. The maximum capacity of 1.5 BPP is obtained by using a one-level wavelet transform. Experimental results demonstrate that the proposed method is superior to the stateof-the-art works in terms of capacity and distortion.

Left Ventricle Segmentation in Cardiac MR Images Using Fully Convolutional Network.

Medical image analysis, especially segmenting a specific organ, has an important role in developing clinical decision support systems. In cardiac magnetic resonance (MR) imaging, segmenting the left and right ventricles helps physicians diagnose different heart abnormalities. There are challenges for this task, including the intensity and shape similarity between the left ventricle and other organs, inaccurate boundaries, and presence of noise in most of the images. In this paper, we propose an automated method for segmenting the left ventricle in cardiac MR images. We first automatically extract the region of interest and then employ it as an input of a fully convolutional network. We train the network accurately despite the small number of left ventricle pixels in comparison with the whole image. Thresholding on the output map of the fully convolutional network and selection of regions based on their roundness are performed in our proposed post-processing phase. The Dice score of our method reaches 87.24% by applying this algorithm on the York dataset of heart images.

Adaptive Real-Time Removal of Impulse Noise in Medical Images.

Noise is an important factor that degrades the quality of medical images. Impulse noise is a common noise caused by malfunctioning of sensor elements or errors in the transmission of images. In medical images due to presence of white foreground and black background, many pixels have intensities similar to impulse noise and hence the distinction between noisy and regular pixels is difficult. Therefore, it is important to design a method to accurately remove this type of noise. In addition to the accuracy, the complexity of the method is very important in terms of hardware implementation. In this paper a low complexity de-noising method is proposed that distinguishes between noisy and non-noisy pixels and removes the noise by local analysis of the image blocks. All steps are designed to have low hardware complexity. Simulation results show that in the case of magnetic resonance images, the proposed method removes impulse noise with an acceptable accuracy.

Vessel segmentation and catheter detection in X-ray angiograms using superpixels.

Coronary artery disease (CAD) is the leading cause of death around the world. One of the most common imaging methods for diagnosing CAD is the X-ray angiography (XRA). Diagnosing using XRA images is usually challenging due to some reasons such as, non-uniform illumination, low contrast, presence of other body tissues, and presence of catheter. These challenges make the diagnosis task hard and more prone to misdiagnosis. In this paper, we propose a new method for coronary artery segmentation, catheter detection, and centerline extraction in X-ray angiography images. For the segmentation, initially, three different superpixel scales are exploited, and a measure for vesselness probability of each superpixel is determined. A voting mechanism is used for obtaining an initial segmentation map from the three superpixel scales. The initial segmentation is refined by finding the orthogonal line on each ridge pixel of vessel region. The catheter is detected in the first frame of the angiography sequence and is tracked in other frames by fitting a second order polynomial on it. Also, we use the image ridges for extracting the coronary artery centerlines. We evaluated and compared our method with one of the previous well-known coronary artery segmentation methods on two challenging datasets. The results show that our method can segment the vessels and also detect and track the catheter in the XRA sequences. In general, the results assessed by a cardiologist show that 83% of the images processed by our proposed segmentation method were labeled as good or excellent, while this score for the compared method is 48%. Also, the evaluation results show that our method performs 67% faster than the compared method. Graphical abstract Proposed framework for coronary artery detection.

Abnormality detection of mammograms by discriminative dictionary learning on DSIFT descriptors.

Detection and classification of breast lesions using mammographic images are one of the most difficult studies in medical image processing. A number of learning and non-learning methods have been proposed for detecting and classifying these lesions. However, the accuracy of the detection/classification still needs improvement. In this paper we propose a powerful classification method based on sparse learning to diagnose breast cancer in mammograms. For this purpose, a supervised discriminative dictionary learning approach is applied on dense scale invariant feature transform (DSIFT) features. A linear classifier is also simultaneously learned with the dictionary which can effectively classify the sparse representations. Our experimental results show the superior performance of our method compared to existing approaches.

Extraction of skin lesions from non-dermoscopic images for surgical excision of melanoma.

PURPOSE: Computerized prescreening of suspicious moles and lesions for malignancy is of great importance for assessing the need and the priority of the removal surgery. Detection can be done by images captured by standard cameras, which are more preferable due to low cost and availability. One important step in computerized evaluation is accurate detection of lesion's region, i.e., segmentation of an image into two regions as lesion and normal skin. METHODS: In this paper, a new method based on deep neural networks is proposed for accurate extraction of a lesion region. The input image is preprocessed, and then, its patches are fed to a convolutional neural network. Local texture and global structure of the patches are processed in order to assign pixels to lesion or normal classes. A method for effective selection of training patches is proposed for more accurate detection of a lesion's border. RESULTS: Our results indicate that the proposed method could reach the accuracy of 98.7% and the sensitivity of 95.2% in segmentation of lesion regions over the dataset of clinical images. CONCLUSION: The experimental results of qualitative and quantitative evaluations demonstrate that our method can outperform other state-of-the-art algorithms exist in the literature.

Liver segmentation using location and intensity probabilistic atlases.

In a variety of injuries and illnesses, internal organs in the abdominal and pelvic regions, in particular liver, may be compromised. In the current practice, CT scans of liver are visually inspected to investigate the integrity of the organ. However, the size and complexity of the CT images limits the reliability of visual inspection to accurately assess the health of liver. Computer-aided image analysis can create fast and quantitative assessment of liver from the CT, in particular in the environments where access to skilled radiologists may be limited. In this paper we propose a hierarchical method based on probabilistic models of position and intensity of voxels for automated segmentation of liver that achieves the Dice similarity coefficient of higher than 89%.

Lossless compression of RNAi fluorescence images using regional fluctuations of pixels.

RNA interference (RNAi) is considered one of the most powerful genomic tools which allows the study of drug discovery and understanding of the complex cellular processes by high-content screens. This field of study, which was the subject of 2006 Nobel Prize of medicine, has drastically changed the conventional methods of analysis of genes. A large number of images have been produced by the RNAi experiments. Even though a number of capable special purpose methods have been proposed recently for the processing of RNAi images but there is no customized compression scheme for these images. Hence, highly proficient tools are required to compress these images. In this paper, we propose a new efficient lossless compression scheme for the RNAi images. A new predictor specifically designed for these images is proposed. It is shown that pixels can be classified into three categories based on their intensity distributions. Using classification of pixels based on the intensity fluctuations among the neighbors of a pixel a context-based method is designed. Comparisons of the proposed method with the existing state-of-the-art lossless compression standards and well-known general-purpose methods are performed to show the efficiency of the proposed method.