ColonGen: an efficient polyp segmentation system for generalization improvement using a new comprehensive dataset.

Colorectal cancer (CRC) is one of the most common causes of cancer-related deaths. While polyp detection is important for diagnosing CRC, high miss rates for polyps have been reported during colonoscopy. Most deep learning methods extract features from images using convolutional neural networks (CNNs). In recent years, vision transformer (ViT) models have been employed for image processing and have been successful in image segmentation. It is possible to improve image processing by using transformer models that can extract spatial location information, and CNNs that are capable of aggregating local information. Despite this, recent research shows limited effectiveness in increasing data diversity and generalization accuracy. This paper investigates the generalization proficiency of polyp image segmentation based on transformer architecture and proposes a novel approach using two different ViT architectures. This allows the model to learn representations from different perspectives, which can then be combined to create a richer feature representation. Additionally, a more universal and comprehensive dataset has been derived from the datasets presented in the related research, which can be used for improving generalizations. We first evaluated the generalization of our proposed model using three distinct training-testing scenarios. Our experimental results demonstrate that our ColonGen-V1 outperforms other state-of-the-art methods in all scenarios. As a next step, we used the comprehensive dataset for improving the performance of the model against in- and out-of-domain data. The results show that our ColonGen-V2 outperforms state-of-the-art studies by 5.1%, 1.3%, and 1.1% in ETIS-Larib, Kvasir-Seg, and CVC-ColonDB datasets, respectively. The inclusive dataset and the model introduced in this paper are available to the public through this link: https://github.com/javadmozaffari/Polyp_segmentation .

A survey on deep learning models for detection of COVID-19.

The spread of the COVID-19 started back in 2019; and so far, more than 4 million people around the world have lost their lives to this deadly virus and its variants. In view of the high transmissibility of the Corona virus, which has turned this disease into a global pandemic, artificial intelligence can be employed as an effective tool for an earlier detection and treatment of this illness. In this review paper, we evaluate the performance of the deep learning models in processing the X-Ray and CT-Scan images of the Corona patients' lungs and describe the changes made to these models in order to enhance their Corona detection accuracy. To this end, we introduce the famous deep learning models such as VGGNet, GoogleNet and ResNet and after reviewing the research works in which these models have been used for the detection of COVID-19, we compare the performances of the newer models such as DenseNet, CapsNet, MobileNet and EfficientNet. We then present the deep learning techniques of GAN, transfer learning, and data augmentation and examine the statistics of using these techniques. Here, we also describe the datasets introduced since the onset of the COVID-19. These datasets contain the lung images of Corona patients, healthy individuals, and the patients with non-Corona pulmonary diseases. Lastly, we elaborate on the existing challenges in the use of artificial intelligence for COVID-19 detection and the prospective trends of using this method in similar situations and conditions. Supplementary Information: The online version contains supplementary material available at 10.1007/s00521-023-08683-x.

MedViT: A robust vision transformer for generalized medical image classification.

Convolutional Neural Networks (CNNs) have advanced existing medical systems for automatic disease diagnosis. However, there are still concerns about the reliability of deep medical diagnosis systems against the potential threats of adversarial attacks since inaccurate diagnosis could lead to disastrous consequences in the safety realm. In this study, we propose a highly robust yet efficient CNN-Transformer hybrid model which is equipped with the locality of CNNs as well as the global connectivity of vision Transformers. To mitigate the high quadratic complexity of the self-attention mechanism while jointly attending to information in various representation subspaces, we construct our attention mechanism by means of an efficient convolution operation. Moreover, to alleviate the fragility of our Transformer model against adversarial attacks, we attempt to learn smoother decision boundaries. To this end, we augment the shape information of an image in the high-level feature space by permuting the feature mean and variance within mini-batches. With less computational complexity, our proposed hybrid model demonstrates its high robustness and generalization ability compared to the state-of-the-art studies on a large-scale collection of standardized MedMNIST-2D datasets.

Automatic breast lesion detection in ultrafast DCE-MRI using deep learning.

PURPOSE: We propose a deep learning-based computer-aided detection (CADe) method to detect breast lesions in ultrafast DCE-MRI sequences. This method uses both the 3D spatial information and temporal information obtained from the early-phase of the dynamic acquisition. METHODS: The proposed CADe method, based on a modified 3D RetinaNet model, operates on ultrafast T1 weighted sequences, which are preprocessed for motion compensation, temporal normalization, and are cropped before passing into the model. The model is optimized to enable the detection of relatively small breast lesions in a screening setting, focusing on detection of lesions that are harder to differentiate from confounding structures inside the breast. RESULTS: The method was developed based on a dataset consisting of 489 ultrafast MRI studies obtained from 462 patients containing a total of 572 lesions (365 malignant, 207 benign) and achieved a detection rate, sensitivity, and detection rate of benign lesions of 0.90 (0.876-0.934), 0.95 (0.934-0.980), and 0.81 (0.751-0.871) at four false positives per normal breast with 10-fold cross-testing, respectively. CONCLUSIONS: The deep learning architecture used for the proposed CADe application can efficiently detect benign and malignant lesions on ultrafast DCE-MRI. Furthermore, utilizing the less visible hard-to-detect lesions in training improves the learning process and, subsequently, detection of malignant breast lesions.

Differentiating benign and malignant mass and non-mass lesions in breast DCE-MRI using normalized frequency-based features.

PURPOSE: In this study, we propose a new computer-aided diagnosis (CADx) to distinguish between malign and benign mass and non-mass lesions in breast DCE-MRI. For this purpose, we introduce new frequency textural features. METHODS: In this paper, we propose novel normalized frequency-based features. These are obtained by applying the dual-tree complex wavelet transform to MRI slices containing a lesion for specific decomposition levels. The low-pass and band-pass frequency coefficients of the dual-tree complex wavelet transform represent the general shape and texture features, respectively, of the lesion. The extraction of these features is computationally efficient. We employ a support vector machine to classify the lesions, and investigate modified cost functions and under- and oversampling strategies to handle the class imbalance. RESULTS: The proposed method has been tested on a dataset of 80 patients containing 103 lesions. An area under the curve of 0.98 for the mass and 0.94 for the non-mass lesions is obtained. Similarly, accuracies of 96.9% and 89.8%, sensitivities of 93.8% and 84.6% and specificities of 98% and 92.3% are obtained for the mass and non-mass lesions, respectively. CONCLUSION: Normalized frequency-based features can characterize benign and malignant lesions efficiently in both mass- and non-mass-like lesions. Additionally, the combination of normalized frequency-based features and three-dimensional shape descriptors improves the CADx performance.

Localized-atlas-based segmentation of breast MRI in a decision-making framework.

Breast-region segmentation is an important step for density estimation and Computer-Aided Diagnosis (CAD) systems in Magnetic Resonance Imaging (MRI). Detection of breast-chest wall boundary is often a difficult task due to similarity between gray-level values of fibroglandular tissue and pectoral muscle. This paper proposes a robust breast-region segmentation method which is applicable for both complex cases with fibroglandular tissue connected to the pectoral muscle, and simple cases with high contrast boundaries. We present a decision-making framework based on geometric features and support vector machine (SVM) to classify breasts in two main groups, complex and simple. For complex cases, breast segmentation is done using a combination of intensity-based and atlas-based techniques; however, only intensity-based operation is employed for simple cases. A novel atlas-based method, that is called localized-atlas, accomplishes the processes of atlas construction and registration based on the region of interest (ROI). Atlas-based segmentation is performed by relying on the chest wall template. Our approach is validated using a dataset of 210 cases. Based on similarity between automatic and manual segmentation results, the proposed method achieves Dice similarity coefficient, Jaccard coefficient, total overlap, false negative, and false positive values of 96.3, 92.9, 97.4, 2.61 and 4.77%, respectively. The localization error of the breast-chest wall boundary is 1.97 mm, in terms of averaged deviation distance. The achieved results prove that the suggested framework performs the breast segmentation with negligible errors and efficient computational time for different breasts from the viewpoints of size, shape, and density pattern.

Fast opposite weight learning rules with application in breast cancer diagnosis.

Classification of breast abnormalities such as masses is a challenging task for radiologists. Computer-aided Diagnosis (CADx) technology may enhance the performance of radiologists by assisting them in classifying patterns into benign and malignant categories. Although Neural Networks (NN) such as Multilayer Perceptron (MLP) have drawbacks, namely long training times, a considerable number of CADx systems employ NN-based classifiers. The reason being that they provide high accuracy when they are appropriately trained. In this paper, we introduce three novel learning rules called Opposite Weight Back Propagation per Pattern (OWBPP), Opposite Weight Back Propagation per Epoch (OWBPE), and Opposite Weight Back Propagation per Pattern in Initialization (OWBPI) to accelerate the training procedure of an MLP classifier. We then develop CADx systems for the diagnosis of breast masses employing the traditional Back Propagation (BP), OWBPP, OWBPE and OWBPI algorithms on MLP classifiers. We quantitatively analyze the accuracy and convergence rate of each system. The results suggest that the convergence rate of the proposed OWBPE algorithm is more than 4 times faster than that of the traditional BP. Moreover, the CADx systems which use OWBPE classifier on average yield an area under Receiver Operating Characteristic (ROC), i.e. Az, of 0.928, a False Negative Rate (FNR) of 9.9% and a False Positive Rate (FPR) of 11.94%.

Classification of benign and malignant masses based on Zernike moments.

In mammography diagnosis systems, high False Negative Rate (FNR) has always been a significant problem since a false negative answer may lead to a patient's death. This paper is directed towards the development of a novel Computer-aided Diagnosis (CADx) system for the diagnosis of breast masses. It aims at intensifying the performance of CADx algorithms as well as reducing the FNR by utilizing Zernike moments as descriptors of shape and margin characteristics. The input Regions of Interest (ROIs) are segmented manually and further subjected to a number of preprocessing stages. The outcomes of preprocessing stage are two processed images containing co-scaled translated masses. Besides, one of these images represents the shape characteristics of the mass, while the other describes the margin characteristics. Two groups of Zernike moments have been extracted from the preprocessed images and applied to the feature selection stage. Each group includes 32 moments with different orders and iterations. Considering the performance of the overall CADx system, the most effective moments have been chosen and applied to a Multi-layer Perceptron (MLP) classifier, employing both generic Back Propagation (BP) and Opposition-based Learning (OBL) algorithms. The Receiver Operational Characteristics (ROC) curve and the performance of resulting CADx systems are analyzed for each group of features. The designed systems yield Az=0.976, representing fair sensitivity, and Az=0.975 demonstrating fair specificity. The best achieved FNR and FPR are 0.0% and 5.5%, respectively.