Biomedical Image Classification Research Articles

Assessing large multimodal models for one-shot learning and interpretability in biomedical image classification.

Image classification plays a pivotal role in analyzing biomedical images, serving as a cornerstone for both biological research and clinical diagnostics. We demonstrate that large multimodal models (LMMs), like GPT-4, excel in one-shot learning, generalization, interpretability, and text-driven image classification across diverse biomedical tasks. These tasks include the classification of tissues, cell types, cellular states, and disease status. LMMs stand out from traditional single-modal classification approaches, which often require large training datasets and offer limited interpretability.

Editorial for Vol.32, No.2

The June 2024 (Vol. 32, No. 2) issue of CIT. Journal of Computing and Information Technology brings four papers from the areas of natural language processing, image processing, network security, and biomedical image classification.

CelloType: A Unified Model for Segmentation and Classification of Tissue Images.

Cell segmentation and classification are critical tasks in spatial omics data analysis. We introduce CelloType, an end-to-end model designed for cell segmentation and classification of biomedical microscopy images. Unlike the traditional two-stage approach of segmentation followed by classification, CelloType adopts a multi-task learning approach that connects the segmentation and classification tasks and simultaneously boost the performance of both tasks. CelloType leverages Transformer-based deep learning techniques for enhanced accuracy of object detection, segmentation, and classification. It outperforms existing segmentation methods using ground-truths from public databases. In terms of classification, CelloType outperforms a baseline model comprised of state-of-the-art methods for individual tasks. Using multiplexed tissue images, we further demonstrate the utility of CelloType for multi-scale segmentation and classification of both cellular and non-cellular elements in a tissue. The enhanced accuracy and multi-task-learning ability of CelloType facilitate automated annotation of rapidly growing spatial omics data.

Biomedical image classification using seagull optimization with deep learning for colon and lung cancer diagnosis

<p>Traditional health care relies on biomedical image categorization to identify and treat various medical conditions. In machine learning and medical imaging, biomedical image classification for colon and lung cancer diagnosis is significant. The work focuses on building novel models and algorithms to accurately detect and categorize tumorous lesions using computer tomography (CT) scans and histopathology slides. These systems use image processing, deep learning (DL), and convolutional neural networks (CNN) to assist medical professionals diagnose cancer sooner and improve patient outcomes. Biomedical image classification using seagull optimization with deep learning (BIC-SGODL) addresses colon and lung cancer diagnosis. The BIC-SGODL method improves cancer diagnosis using hyperparameter optimized DL model. BIC-SGODL utilizes DenseNet to learn complicated features. The convolutional long short-term memory (CLSTM) standard captures spatiotemporal information in sequential picture data. Finally, the SGO method adjusts hyperparameters to improve model performance and generalization. BIC-SGODL performs well with biomedical image dataset simulations. Thus, medical picture cancer diagnosis may be automated using BIC-SGODL.</p>

MobileDenseNeXt: Investigations on biomedical image classification

Background and objectiveWe are living in the information era. Therefore, intelligence-based researchers are hot-topic such as artificial intelligence. In the artificial intelligence research area, machine learning and deep learning models have frequently used to create intelligence assistants and deep learning is the shining star of the AI. Specifically, in the computer vision, numerous deep learning models have been proposed, leading to a competition between transformers and convolutional neural networks (CNNs). Since the introduction of Vision Transformers (ViT), many transformer models have been advocated for computer vision, often overshadowing CNNs. Therefore, it is crucial to propose CNNs to showcase their prowess in image classification. This research introduces a lightweight CNN named MobileDenseNeXt. Materials and methodsThe proposed MobileDenseNeXt comprises four main blocks: (i) input, (ii) main, (iii) average pooling-based downsampling, and (iv) output. This research also incorporates convolution-based residual blocks and uses a depth concatenation layer to increase the number of filters. For downsampling, an average pooling operation has been employed, similar to the original DenseNet. Furthermore, the swish activation function is utilized in the presented CNN. MobileDenseNeXt has approximately 1.4 million learnable parameters, categorizing it as a lightweight CNN model. Additionally, a deep feature engineering approach has been developed using MobileDenseNeXt, incorporating two feature extractors with global average pooling and dropout layers, along with 10 feature selectors, to demonstrate the transfer learning capabilities of MobileDenseNeXt. ResultsThe recommended models achieved over 95% test classification accuracy on the used three datasets, unequivocally demonstrating the high image classification proficiency of the proposed MobileDenseNeXt. Moreover, to show general classification ability of the proposed model, MobileDenseNeXt was trained on the CIFAR10 dataset and reached 98.62% accuracy. ConclusionsThis research not only highlights the efficiency and effectiveness of MobileDenseNeXt in biomedical image classification but also highlights the competitive potential of this model for computer vision.

Classification of rotation-invariant biomedical images using equivariant neural networks

Transmission electron microscopy (TEM) is an imaging technique used to visualize and analyze nano-sized structures and objects such as virus particles. Light microscopy can be used to diagnose diseases or characterize e.g. blood cells. Since samples under microscopes exhibit certain symmetries, such as global rotation invariance, equivariant neural networks are presumed to be useful. In this study, a baseline convolutional neural network is constructed in the form of the commonly used VGG16 classifier. Thereafter, it is modified to be equivariant to the p4 symmetry group of rotations of multiples of 90° using group convolutions. This yields a number of benefits on a TEM virus dataset, including higher top validation set accuracy by on average 7.6% and faster convergence during training by on average 23.1% of that of the baseline. Similarly, when training and testing on images of blood cells, the convergence time for the equivariant neural network is 7.9% of that of the baseline. From this it is concluded that augmentation strategies for rotation can be skipped. Furthermore, when modelling the accuracy versus amount of TEM virus training data with a power law, the equivariant network has a slope of − 0.43 compared to − 0.26 of the baseline. Thus the equivariant network learns faster than the baseline when more training data is added. This study extends previous research on equivariant neural networks applied to images which exhibit symmetries to isometric transformations.

Can deep neural networks work with amplitude and phase input of defocused images?

Deep neural network (DNN) models, particularly convolutional neural networks (CNNs), have demonstrated remarkable performance in biomedical image classification due to their ability to automatically learn features from large datasets. One common challenge in the preparation of large, microscopic datasets for DNN tasks is sample defocusing, potentially impairing the model performance. To handle defocusing, computational imaging, or specifically quantitative phase imaging (QPI), performs digital refocusing by using both the phase and the amplitude of the complex optical field. This leads us to investigate whether feeding the complex field into DNN would potentially address the defocusing problem as in-focus information is implicitly encoded in the complex field. In this paper, we assess the feasibility of employing neural networks to directly process full amplitude and phase data from a defocus plane without digital refocusing. Our specific focus lies in understanding the tolerance for defocus in image classification neural networks when amplitude and phase are taken as inputs. To accomplish this, we used Fourier ptychography microscopy (FPM) to acquire in-focus phase and amplitude images of two distinct object types - normal red blood cells and echinocytes. We then digitally propagate the complex field to generate progressively defocused images of the samples to serve as training and testing datasets for image classification neural networks. While the digitally defocused images contain sufficient information to recover the original in-focus images, we observed that current standard implementations of deep learning models are unable to effectively utilize the defocused field to distinguish between the two cell types. We conclude that the physical-model-based digital refocusing capability of QPI remains indispensable for overcoming defocusing issues in current standard DNN models.

Intelligent deep learning supports biomedical image detection and classification of oral cancer.

Oral cancer is a malignant tumor that usually occurs within the tissues of the mouth. This type of cancer mainly includes tumors in the lining of the mouth, tongue, lips, buccal mucosa and gums. Oral cancer is on the rise globally, especially in some specific risk groups. The early stage of oral cancer is usually asymptomatic, while the late stage may present with ulcers, lumps, bleeding, etc. The objective of this paper is to propose an effective and accurate method for the identification and classification of oral cancer. We applied two deep learning methods, CNN and Transformers. First, we propose a new CANet classification model for oral cancer, which uses attention mechanisms combined with neglected location information to explore the complex combination of attention mechanisms and deep networks, and fully tap the potential of attention mechanisms. Secondly, we design a classification model based on Swim transform. The image is segmented into a series of two-dimensional image blocks, which are then processed by multiple layers of conversion blocks. The proposed classification model was trained and predicted on Kaggle Oral Cancer Images Dataset, and satisfactory results were obtained. The average accuracy, sensitivity, specificity and F1-Socre of Swin transformer architecture are 94.95%, 95.37%, 95.52% and 94.66%, respectively. The average accuracy, sensitivity, specificity and F1-Score of CANet model were 97.00%, 97.82%, 97.82% and 96.61%, respectively. We studied different deep learning algorithms for oral cancer classification, including convolutional neural networks, converters, etc. Our Attention module in CANet leverages the benefits of channel attention to model the relationships between channels while encoding precise location information that captures the long-term dependencies of the network. The model achieves a high classification effect with an accuracy of 97.00%, which can be used in the automatic recognition and classification of oral cancer.

Implementing vision transformer for classifying 2D biomedical images

In recent years, the growth spurt of medical imaging data has led to the development of various machine learning algorithms for various healthcare applications. The MedMNISTv2 dataset, a comprehensive benchmark for 2D biomedical image classification, encompasses diverse medical imaging modalities such as Fundus Camera, Breast Ultrasound, Colon Pathology, Blood Cell Microscope etc. Highly accurate classifications performed on these datasets is crucial for identification of various diseases and determining the course of treatment. This research paper presents a comprehensive analysis of four subsets within the MedMNISTv2 dataset: BloodMNIST, BreastMNIST, PathMNIST and RetinaMNIST. Each of these selected datasets is of diverse data modalities and comes with various sample sizes, and have been selected to analyze the efficiency of the model against diverse data modalities. The study explores the idea of assessing the Vision Transformer Model’s ability to capture intricate patterns and features crucial for these medical image classification and thereby transcend the benchmark metrics substantially. The methodology includes pre-processing the input images which is followed by training the ViT-base-patch16-224 model on the mentioned datasets. The performance of the model is assessed using key metrices and by comparing the classification accuracies achieved with the benchmark accuracies. With the assistance of ViT, the new benchmarks achieved for BloodMNIST, BreastMNIST, PathMNIST and RetinaMNIST are 97.90%, 90.38%, 94.62% and 57%, respectively. The study highlights the promise of Vision transformer models in medical image analysis, preparing the way for their adoption and further exploration in healthcare applications, aiming to enhance diagnostic accuracy and assist medical professionals in clinical decision-making.

Classification of Biomedical Images with Mined Statistical Features and Dynamic Programming

Background:: In the research and practice of medical sciences, accurate classification of biomedical images with computer programs may provide an important basis for the study and diagnosis of many diseases. Methods:: This paper proposes a new statistical approach that can accurately classify biomedical images based on their statistical features. In the first step of the proposed approach, a number of SIFT features of different types are computed for each pixel in a biomedical image and a statistical feature that describes the distribution of each type of SIFT features is obtained for the image. In the second step, a dynamic programming approach is used to efficiently analyze the dependence among different statistical features associated with an image and compute the probability for an image to belong to each possible class; the class with the largest probability is determined as the result of classification. Results:: Experimental results show that the proposed approach can lead to classification results with accuracy higher than that of a few state-of-the-art approaches for the classification of biomedical images Conclusion:: The proposed approach can achieve classification accuracy comparable to that of several state-of-the-art classification approaches. It is thus potentially useful for applications where large models are not appropriate for classification tasks due to limitations in computational or communication resources

Early Detection and Diagnosis of Oral Cancer Using Deep Neural Network

Oral squamous cell carcinoma (OSCC) is a malignancy that destroys the ability of the tissues surrounding the mouth to develop layers and membranes. Automated early diagnosis of oral histopathological images has allowed for the successful diagnosis of oral cancer, thanks to recent advancements in Deep Learning (DL) for biomedical image classification. By using a convolutional neural network (CNN) model based on deep learning for the initial analysis of oral squamous cell carcinoma (OSCC), this work aims to automate the classification of benign and malignant oral biopsy histopathological images. For this study, the CNN model Inception-Resnet-V2 is selected using the transfer learning approach. To enhance OSCC detection, additional layers are incorporated into this pretrained model. By mining a repository of oral cancer histopathology images, we can gauge how well these tweaked models perform. We examine the modified structure of the pre-trained Inception-Resnet-V2 model and suggest a DL-CNN model that uses it. With an accuracy of 91.78%, it has outperformed in terms of performance metrics.

Early Detection and Diagnosis of Oral Cancer Using Deep Neural Network

Oral squamous cell carcinoma (OSCC) is a malignancy that destroys the ability of the tissues surrounding the mouth to develop layers and membranes. Automated early diagnosis of oral histopathological images has allowed for the successful diagnosis of oral cancer, thanks to recent advancements in Deep Learning (DL) for biomedical image classification. By using a convolutional neural network (CNN) model based on deep learning for the initial analysis of oral squamous cell carcinoma (OSCC), this work aims to automate the classification of benign and malignant oral biopsy histopathological images. For this study, the CNN model Inception-Resnet-V2 is selected using the transfer learning approach. To enhance OSCC detection, additional layers are incorporated into this pretrained model. By mining a repository of oral cancer histopathology images, we can gauge how well these tweaked models perform. We examine the modified structure of the pre-trained Inception-Resnet-V2 model and suggest a DL-CNN model that uses it. With an accuracy of 91.78%, it has outperformed in terms of performance metrics.

On Using Computer Linguistic Models in the Classification of Biomedical Images

On Using Computer Linguistic Models in the Classification of Biomedical Images

Extreme Learning Machine for Biomedical Image Classification: A Multi-Case Study

In the current realm of biomedical image classification, the predominant choice remains deep learning networks, particularly convolutional neural network (CNN) models. However, deep learning suffers from a notable drawback in terms of its high training cost, mainly due to intricate data models. A recent alternative, known as the Extreme Learning Machine (ELM), has emerged as a promising solution. Empirical investigations have indicated that ELM can offer satisfactory predictive performance for a wide array of classification tasks, while significantly reducing training costs when compared to deep learning networks trained using back propagation.This research paper introduces a methodology designed to evaluate the suitability of employing the Extreme Learning Machine for biomedical classification tasks. Our study encompasses binary and multiclass classification across four distinct scenarios, involving the analysis of biomedical images obtained from both dermatoscopes and blood cell microscopes. The findings underscore the effectiveness of the Extreme Learning Machine, showcasing its successful utilization in the classification of biomedical images.

Performance Analysis of Five U-Nets on Cervical Cancer Datasets

Image segmentation is crucial for precise analysis and classification of biomedical images, especially in the realm of cervical cancer detection. The accuracy of segmentation significantly influences the efficacy of subsequent image classification processes. While traditional algorithms exist for image segmentation, recent advancements in convolutional neural networks, particularly U-Nets have showcased exceptional effectiveness, especially in the realm of biomedical imaging. This research focuses on evaluating the accuracy of various U-Net architectures applied to three distinct cervical cancer datasets i.e., DSB containing 1340 images, SipakMed containing 1849 images and Intel Images for Screening containing 2000 images datasets taken from 2018 Data Science Bowl. The investigated U-Net architectures comprise the fundamental U-Net, Attention U-Net, Double U-Net, Spatial Attention U-Net, and Residual U-Net. The performance of the u-nets is judged on the metrics: Recall, Precision, F1, Jaccard and Accuracy. It is observed that Basic U-Net on the DSB dataset provides highest value on these metrices and accuracy obtained is 96%. The reason of high accuracy for DSB dataset can be attributed to the contrast of the images which by using co-occurrence matrix is calculated as 145.

A Computational Deep Fuzzy Network-Based Neuroimaging Analysis for Brain Hemorrhage Classification.

The boundaries and regions between individual classes in biomedical image classification are hazy and overlapping. These overlapping features make predicting the correct classification result for biomedical imaging data a difficult diagnostic task. Thus, in precise classification, it is frequently necessary to obtain all necessary information before making a decision. This paper presents a novel deep-layered design architecture based on Neuro-Fuzzy-Rough intuition to predict hemorrhages using fractured bone images and head CT scans. To deal with data uncertainty, the proposed architecture design employs a parallel pipeline with rough-fuzzy layers. In this case, the rough-fuzzy function functions as a membership function, incorporating the ability to process rough-fuzzy uncertainty information. It not only improves the deep model's overall learning process, but it also reduces feature dimensions. The proposed architecture design improves the model's learning and self-adaptation capabilities. In experiments, the proposed model performed well, with training and testing accuracies of 96.77% and 94.52%, respectively, in detecting hemorrhages using fractured head images. The comparative analysis shows that the model outperforms existing models by an average of 2.6 ±0.90% on various performance metrics.

One-Vote Veto: Semi-Supervised Learning for Low-Shot Glaucoma Diagnosis.

Convolutional neural networks (CNNs) are a promising technique for automated glaucoma diagnosis from images of the fundus, and these images are routinely acquired as part of an ophthalmic exam. Nevertheless, CNNs typically require a large amount of well-labeled data for training, which may not be available in many biomedical image classification applications, especially when diseases are rare and where labeling by experts is costly. This article makes two contributions to address this issue: 1) It extends the conventional Siamese network and introduces a training method for low-shot learning when labeled data are limited and imbalanced, and 2) it introduces a novel semi-supervised learning strategy that uses additional unlabeled training data to achieve greater accuracy. Our proposed multi-task Siamese network (MTSN) can employ any backbone CNN, and we demonstrate with four backbone CNNs that its accuracy with limited training data approaches the accuracy of backbone CNNs trained with a dataset that is 50 times larger. We also introduce One-Vote Veto (OVV) self-training, a semi-supervised learning strategy that is designed specifically for MTSNs. By taking both self-predictions and contrastive predictions of the unlabeled training data into account, OVV self-training provides additional pseudo labels for fine-tuning a pre-trained MTSN. Using a large (imbalanced) dataset with 66,715 fundus photographs acquired over 15 years, extensive experimental results demonstrate the effectiveness of low-shot learning with MTSN and semi-supervised learning with OVV self-training. Three additional, smaller clinical datasets of fundus images acquired under different conditions (cameras, instruments, locations, populations) are used to demonstrate the generalizability of the proposed methods.

Intelligent biomedical image classification in a big data architecture using metaheuristic optimization and gradient approximation

AbstractMedical imaging has experienced significant development in contemporary medicine and can now record a variety of biomedical pictures from patients to test and analyze the illness and its severity. Computer vision and artificial intelligence may outperform human diagnostic ability and uncover hidden information in biomedical images. In healthcare applications, fast prediction and reliability are of the utmost importance parameters to assure the timely detection of disease. The existing systems have poor classification accuracy, and higher computation time and the system complexity is higher. Low-quality images might impact the processing method, leading to subpar results. Furthermore, extensive preprocessing techniques are necessary for achieving accurate outcomes. Image contrast is one of the most essential visual parameters. Insufficient contrast may present many challenges for computer vision techniques. Traditional contrast adjustment techniques may not be adequate for many applications. Occasionally, these technologies create photos that lack crucial information. The primary contribution of this work is designing a Big Data Architecture (BDA) to improve the dependability of medical systems by producing real-time warnings and making precise forecasts about patient health conditions. A BDA-based Bio-Medical Image Classification (BDA-BMIC) system is designed to detect the illness of patients using Metaheuristic Optimization (Genetic Algorithm) and Gradient Approximation to improve the biomedical image classification process. Extensive tests are conducted on publicly accessible datasets to demonstrate that the suggested retrieval and categorization methods are superior to the current methods. The suggested BDA-BMIC system has average detection accuracy of 94.6% and a sensitivity of 97.3% in the simulation analysis.

A quantum convolutional network and ResNet (50)-based classification architecture for the MNIST medical dataset

Biomedical image classification is crucial for both computer vision tasks and clinical care. The conventional method requires a significant amount of time and effort for extracting and selecting classification features. Deep Neural Networks (DNNs) and Quantum Convolutional Neural Networks (QCNN) are emerging techniques in machine learning that have demonstrated their efficacy for various classification tasks. Because of the complexity of their designs, the results of such models may also be challenging to interpret. In this paper, we propose an architecture called Medical Quantum Convolutional Neural Network (MQCNN), based on the QCNN model and a modified ResNet (50) pre-trained model, for enhancing the biomedical image classification in the MNIST medical dataset. During the training phase, the weights are updated using the Adam optimizer, while ResNet (50) is used to reduce the computational cost. MQCNN is compared to the QCNN model, the ResNet (50) pre-trained model, and some other related works on the Medical MNIST dataset. The results showed that MQCNN model achieves 99.6% accuracy, 99.7% precision, 99.6% recall, and 99.7% F1 score, and outperforms the ResNet (50) pre-trained model, the QCNN model, and the other compared related works.

Automated biomedical image classification using multi-scale dense dilated semi-supervised u-net with cnn architecture

Automated biomedical image classification using multi-scale dense dilated semi-supervised u-net with cnn architecture