Computer-aided Diagnosis Tool Research Articles

AN APPROACH TO COMPUTER-AIDED DIAGNOSIS OF HEART DISORDERS USING WAVELETS AND DEEP LEARNING APPLIED TO ELECTROCARDIOGRAMS (EKGS)

Purpose: The purpose of this study was to evaluate the potential of deep learning as a tool for computer-aided diagnosis of heart disorders based on EKG signals, using wavelet transformations to generate images. The research question was whether deep learning algorithms could accurately diagnose heart disorders and provide a valuable complement to traditional EKG views. Methods: We trained five Convolutional Neural Networks (CNNs) using EKG data obtained from the Physionet public database. The algorithms were developed using MATLAB version 2018b and the toolboxes for digital signal processing, neural networks, and wavelets. We evaluated the performance of the CNNs using accuracy, sensitivity, specificity, positive predictive value, and negative predictive value as metrics. Results: The CNNs demonstrated accuracy greater than 90%, and achieved good performance for the other evaluated parameters. We also identified that the representation of EKGs as scalograms showed potential for use as a complement to traditional EKG views. Conclusion: Our findings demonstrate that deep learning is a promising tool for diagnosing heart disorders based on EKG signals, and can be a valuable complement to traditional EKG views. While automated diagnoses should not replace clinical judgment, deep learning can provide additional support to healthcare professionals. Further research should explore the potential of deep learning for medical diagnosis and the use of scalograms as a complementary tool in clinical practice.

From Pixels to Pathology: Employing Computer Vision to Decode Chest Diseases in Medical Images.

Radiology has been a pioneer in the healthcare industry's digital transformation, incorporating digital imaging systems like picture archiving and communication system (PACS) and teleradiology over the past thirty years. This shift has reshaped radiology services, positioning the field at a crucial junction for potential evolution into an integrated diagnostic service through artificial intelligence and machine learning. These technologies offer advanced tools for radiology's transformation. The radiology community has advanced computer-aided diagnosis (CAD) tools using machine learning techniques, notably deep learning convolutional neural networks (CNNs), for medical image pattern recognition. However, the integration of CAD tools into clinical practice has been hindered by challenges in workflow integration, unclear business models, and limited clinical benefits, despite development dating back to the 1990s. This comprehensive review focuses on detecting chest-related diseases through techniques like chest X-rays (CXRs), magnetic resonance imaging (MRI), nuclear medicine, and computed tomography (CT) scans. It examines the utilization of computer-aided programs by researchers for disease detection, addressing key areas: the role of computer-aided programs in disease detection advancement, recent developments in MRI, CXR, radioactive tracers, and CT scans for chest disease identification, research gaps for more effective development, and the incorporation of machine learning programs into diagnostic tools.

3D unsupervised anomaly detection through virtual multi-view projection and reconstruction: Clinical validation on low-dose chest computed tomography

Computer-aided diagnosis for low-dose computed tomography (CT) based on deep learning has recently attracted attention as a first-line automatic testing tool because of its high accuracy and low radiation exposure. We propose a method based on a deep neural network for computer-aided diagnosis called virtual multi-view projection and reconstruction for unsupervised anomaly detection (VMPR-UAD) in low-dose chest CT. Presumably, this is the novel method that only requires data from healthy patients for training to identify three-dimensional (3D) regions containing any anomalies. The method has three key components. Unlike existing computer-aided diagnosis tools that use conventional CT slices as the network input, our method (1) improves the recognition of 3D lung structures by virtually projecting an extracted 3D lung region to obtain two-dimensional (2D) images from diverse views to serve as network inputs, (2) accommodates the input diversity gain for accurate anomaly detection, and (3) achieves 3D anomaly/disease localization through a novel 3D map restoration method using multiple 2D anomaly maps. The proposed method based on unsupervised learning showed a high performance in pneumonia, tuberculosis, and both diseases with patient-level anomaly detection performance of 0.965 area under the curve (AUC) (95% confidence interval (CI); (0.955, 0.972)), 0.948 AUC (95% CI; (0.928, 0.966)), and 0.963 AUC (95% CI; (0.955, 0.970)), respectively. Additionally, our technology visualizes anomalous regions in the 3D perspective. This achieved 93% accuracy in visualizing the location of lung cancer lesions through external validation. These results highlight the potential of a new AI methodology without utilizing disease data learning; this can secure AI model prediction stability by reducing the false negative rate that occurs in various patterns of diseases.

Generative adversarial networks based skin lesion segmentation

Skin cancer is a serious condition that requires accurate diagnosis and treatment. One way to assist clinicians in this task is using computer-aided diagnosis tools that automatically segment skin lesions from dermoscopic images. We propose a novel adversarial learning-based framework called Efficient-GAN (EGAN) that uses an unsupervised generative network to generate accurate lesion masks. It consists of a generator module with a top-down squeeze excitation-based compound scaled path, an asymmetric lateral connection-based bottom-up path, and a discriminator module that distinguishes between original and synthetic masks. A morphology-based smoothing loss is also implemented to encourage the network to create smooth semantic boundaries of lesions. The framework is evaluated on the International Skin Imaging Collaboration Lesion Dataset. It outperforms the current state-of-the-art skin lesion segmentation approaches with a Dice coefficient, Jaccard similarity, and accuracy of 90.1%, 83.6%, and 94.5%, respectively. We also design a lightweight segmentation framework called Mobile-GAN (MGAN) that achieves comparable performance as EGAN but with an order of magnitude lower number of training parameters, thus resulting in faster inference times for low compute resource settings.

Thyroid Nodules Classification using Weighted Average Ensemble and D-CRITIC based TOPSIS Methods for Ultrasound Images.

Thyroid disorders are prevalent worldwide and impact many people. The abnormal growth of cells in the thyroid gland region is very common and even found in healthy people. These abnormal cells can be cancerous or non-cancerous, so early detection of this disease is the only solution for minimizing the death rate or maximizing a patient's survival rate. Traditional techniques to detect cancerous nodules are complex and time-consuming; hence, several imaging algorithms are used to detect the malignant status of thyroid nodules timely. This research aims to develop computer-aided diagnosis tools for malignant thyroid nodule detection using ultrasound images. This tool will be helpful for doctors and radiologists in the rapid detection of thyroid cancer at its early stages. The individual machine learning models are inferior to medical datasets because the size of medical image datasets is tiny, and there is a vast class imbalance problem. These problems lead to overfitting; hence, accuracy is very poor on the test dataset. This research proposes ensemble learning models that achieve higher accuracy than individual models. The objective is to design different ensemble models and then utilize benchmarking techniques to select the best model among all trained models. This research investigates four recently developed image transformer and mixer models for thyroid detection. The weighted average ensemble models are introduced, and model weights are optimized using the hunger games search (HGS) optimization algorithm. The recently developed distance correlation CRITIC (D-CRITIC) based TOPSIS method is utilized to rank the models. Based on the TOPSIS score, the best model for an 80:20 split is the gMLP+ViT model, which achieved an accuracy of 89.70%, whereas using a 70:30 data split, the gMLP+FNet+Mixer-MLP has achieved the highest accuracy of 82.18% on the publicly available thyroid dataset. This study shows that the proposed ensemble models have better thyroid detection capabilities than individual base models for the imbalanced thyroid ultrasound dataset.

A multicenter study on two-stage transfer learning model for duct-dependent CHDs screening in fetal echocardiography

Duct-dependent congenital heart diseases (CHDs) are a serious form of CHD with a low detection rate, especially in underdeveloped countries and areas. Although existing studies have developed models for fetal heart structure identification, there is a lack of comprehensive evaluation of the long axis of the aorta. In this study, a total of 6698 images and 48 videos are collected to develop and test a two-stage deep transfer learning model named DDCHD-DenseNet for screening critical duct-dependent CHDs. The model achieves a sensitivity of 0.973, 0.843, 0.769, and 0.759, and a specificity of 0.985, 0.967, 0.956, and 0.759, respectively, on the four multicenter test sets. It is expected to be employed as a potential automatic screening tool for hierarchical care and computer-aided diagnosis. Our two-stage strategy effectively improves the robustness of the model and can be extended to screen for other fetal heart development defects.

Analysis based on machine and deep learning techniques for the accurate detection of lung nodules from CT images

The imaging modality of computed tomography (CT) is significant for the diagnosis of lung cancer. Doctors, surgeons, radiologists, and oncologists choose CT scanning to find these nodules for diagnosing lung cancer early by using computer-aided diagnosis tools. Early malignant diagnosis is necessary for providing an efficient treatment for patients who are affected by lung cancer and there are multiple methods available for the detection of these modules. Gaining knowledge about the various methods is more important for developing a novel model and hence in this research, 50 recent articles from the last five years are evaluated with a focus on lung nodule identification and a goal of highlighting the problems associated with the current approaches. Based on several variables, including performance measures, publication years and journals, achievements of the methods in numerical evaluations, and other things, the evaluation of various methods is made in this study. On the contrary, a technical analysis that considers the advantages and disadvantages of the approach is also interpreted. This study analyzed several deep learning and related techniques for locating lung nodules using CT scans that provide a deeper insight into their cons and pros. The investigation revealed that the deep learning optimization-based strategies outperformed the convolutional techniques concerning the performance measures like accuracy, sensitivity, specificity, F-score, precision, and recall for the achievements, which vary from 80 to 100%. Lastly, the report examines potential directions for future research as well as challenges to improving lung nodule detection accuracy.

OPTIMAL USE OF BIRAD DESCRIPTORS FOR BREAST CANCER DETECTION BY EMPLOYING DATA MINING WITH CORRELATION OF BIOPSY OUTCOME

Background: Breast cancer (BC) is commonest among cancer patients, required early detection to be cured. Multiple techniques are used for the diagnosis of BC but mammography is a commonly available and effective tool around the world due to its ability to diagnose BC before it becomes palpable, in reward treatment cost also decreases. Computer-aided diagnosis (CAD) is advanced research to support medical practitioners in cancer diagnosis.
 Methods: In comparison to the segmentation technique our research has used “extracted BIRAD characteristics information” from the mammograms. A novel data mining approach for the BIRAD lexicon has been researched in connection with the interactive medical experience. A globally recognized dataset (DDSM) has been used. The characteristics extracted from the mammograms are processed by the data mining method to achieve critical ranges for cancer detection. Multiple classification models with peculiar characteristics have developed to detect BC at the curable stage. 
 Results: A total of 1863 patients with an average age of 61 years were evaluated, 48% indicated with malignancies, and 52% with benign lesions, proved with truth biopsy. Critical ranges, by the development of multiple CAD tools, showed remarkable improvement in the recall (NB; 26.9%) and precision (CART; 17.6%) along with reduced FP rate (CART; 16.7%).
 Conclusion: The proposed method is profitable because of the individuality of CAD with four classification models along with the practitioner’s experience and radiologist’s opinion in the early detection of BC. For avoiding unnecessary biopsies and differentiating benign and malignant tumor the BIRAD features extraction was benevolent.

Application of Deep Learning Methods in a Moroccan Ophthalmic Center: Analysis and Discussion.

Diabetic retinopathy (DR) remains one of the world's frequent eye illnesses, leading to vision loss among working-aged individuals. Hemorrhages and exudates are examples of signs of DR. However, artificial intelligence (AI), particularly deep learning (DL), is poised to impact nearly every aspect of human life and gradually transform medical practice. Insight into the condition of the retina is becoming more accessible thanks to major advancements in diagnostic technology. AI approaches can be used to assess lots of morphological datasets derived from digital images in a rapid and noninvasive manner. Computer-aided diagnosis tools for automatic detection of DR early-stage signs will ease the pressure on clinicians. In this work, we apply two methods to the color fundus images taken on-site at the Cheikh Zaïd Foundation's Ophthalmic Center in Rabat to detect both exudates and hemorrhages. First, we apply the U-Net method to segment exudates and hemorrhages into red and green colors, respectively. Second, the You Look Only Once Version 5 (YOLOv5) method identifies the presence of hemorrhages and exudates in an image and predicts a probability for each bounding box. The segmentation proposed method obtained a specificity of 85%, a sensitivity of 85%, and a Dice score of 85%. The detection software successfully detected 100% of diabetic retinopathy signs, the expert doctor detected 99% of DR signs, and the resident doctor detected 84%.

A Generative Adversarial Network Based Approach for Synthesis of Deep Fake Electrocardiograms

Analyzing the data from an electrocardiogram (ECG) can reveal important details about a patient's heart health. A key component of modern medicine is the use of AI and ML-based computer-aided diagnosis tools to aid in making life-or-death decisions. It is common practice to use them in cardiology for the automatic early diagnosis of a variety of potentially fatal illnesses. The machine learning algorithm's need for a large amount of training data to build the learning model is an empirical challenge in the medical domain. To address this challenge, study into methods for creating synthetic patient data has blossomed. There is a higher risk of privacy invasion due to the need for massive amounts of training data for deep learning automated medical diagnostic systems that may help assess the state of the heart from this signal. To combat this issue, researchers have tried to create artificial ECG readings by analyzing only the statistical distributions of the accessible authentic training data.The primary goal of this study is to learn how generative adversarial networks can be used to create artificial ECG signals for use as training data in a classification task. In this study, we used both GAN and WGAN for generation of artificial ECG signals.

Development and Deployment of a Novel Diagnostic Tool Based on Conventional Ultrasound for Fibrosis Assessment in Chronic Kidney Disease

Accurate identification of risk information about fibrosis severity is crucial for clinical decision-making and clinical management of patients with chronic kidney disease (CKD). This study aimed to develop an ultrasound (US)-derived computer-aided diagnosis tool for identifying CKD patients at high risk of developing moderate-severe renal fibrosis, in order to optimize treatment regimens and follow-up strategies. A total of 162 CKD patients undergoing renal biopsies and US examinations were prospectively enrolled and randomly divided into training (n=114) and validation (n=48) cohorts. A multivariate logistic regression approach was employed to develop the diagnostic tool named S-CKD for differentiating moderate-severe renal fibrosis from mild one in the training cohort by integrating the significant variables, which were screened out from demographic characteristics and conventional US features via the least absolute shrinkage and selection operator regression algorithm. The S-CKD was then deployed as both an online web-based and an offline document-based, easy-to-use auxiliary device. In both the training and validation cohorts, the S-CKD's diagnostic performance was evaluated through discrimination and calibration.The clinical benefit of using S-CKD was revealed by decision curve analysis (DCA) and clinical impact curves. The proposed S-CKD achieved an area under the receiver operating characteristic curve of 0.84 (95% confidence interval (CI): 0.77-0.91) and 0.81 (95% CI: 0.68-0.94) in the training and validation cohorts, respectively, indicating satisfactory diagnosis performance. Results of the calibration curves showed that S-CKD has excellent predictive accuracy (Hosmer-Lemeshow test: training cohort, p=0.497; validation cohort, p=0.205). The DCA and clinical impact curves exhibited a high clinical application value of the S-CKD at a wide range of risk probabilities. The S-CKD tool developed in this study is capable of discriminating between mild and moderate-severe renal fibrosis in patients with CKD and achieving promising clinical benefits, which may aid clinicians in personalizing medical decision-making and follow-up arrangement.

A Temporal Transformer-Based Fusion Framework for Morphological Arrhythmia Classification

By using computer-aided arrhythmia diagnosis tools, electrocardiogram (ECG) signal plays a vital role in lowering the fatality rate associated with cardiovascular diseases (CVDs) and providing information about the patient’s cardiac health to the specialist. Current advancements in deep-learning-based multivariate time series data analysis, such as ECG data classification include LSTM, Bi-LSTM, CNN, with Bi-LSTM, and other sequential networks. However, these networks often struggle to accurately determine the long-range dependencies among data instances, which can result in problems such as vanishing or exploding gradients for longer data sequences. To address these shortcomings of sequential models, a hybrid arrhythmia classification system using recurrence along with a self-attention mechanism is developed. This system utilizes convolutional layers as a part of representation learning, designed to capture the salient features of raw ECG data. Then, the latent embedded layer is fed to a self-attention-assisted transformer encoder model. Because the ECG data are highly influenced by absolute order, position, and proximity of time steps due to interdependent relationships among immediate neighbors, a component of recurrence using Bi-LSTM is added to the encoder model to address this characteristic of the data. The model performance indices such as classification accuracy and F1-score were found to be 99.2%. This indicates that the combination of recurrence along with self-attention-assisted architecture produces improved classification of arrhythmia from raw ECG signal when compared with the state-of-the-art models.

Automated major depressive disorder diagnosis using a dual-input deep learning model and image generation from EEG signals

Major depressive disorder (MDD) is conventionally diagnosed through a questionnaire. Since approaches to diagnose MDD may lead to inaccurate diagnoses, many studies have presented electroencephalogram (EEG)-based machine learning techniques. The present paper introduces a deep learning approach based on image construction from EEGs. Two images are constructed from EEGs based on spectral and functional connectivity features. Afterward, the constructed images are applied to a two-stream convolutional neural network, and the outputs are concatenated. Finally, the concatenating result is applied to a sequential model of long–short-term memory, fully connected, and softmax layers to classify each sample into the MDD and healthy control (HC) classes. To validate the proposed approach, a public EEG dataset was used consisting of EEG data recorded from 34 MDD patients and 30 HC-matched participants. This framework obtained an AC of 98.03%, SE of 98.85%, SP of 97.19%, F1 of 98.07%, and FDR of 2.69% for the random splitting assessment method and achieved an average AC of 99.11%, SE of 98.97%, SP of 99.25%, F1 of 99.13%, and FDR of 0.71% using a 10-fold cross-validation process. Considering the accurate performance of the proposed method, it can be developed as a computer-aided diagnosis tool to diagnose MDD automatically.

Enhancing Multimodal Patterns in Neuroimaging by Siamese Neural Networks with Self-Attention Mechanism.

The combination of different sources of information is currently one of the most relevant aspects in the diagnostic process of several diseases. In the field of neurological disorders, different imaging modalities providing structural and functional information are frequently available. Those modalities are usually analyzed separately, although a joint of the features extracted from both sources can improve the classification performance of Computer-Aided Diagnosis (CAD) tools. Previous studies have computed independent models from each individual modality and combined them in a subsequent stage, which is not an optimum solution. In this work, we propose a method based on the principles of siamese neural networks to fuse information from Magnetic Resonance Imaging (MRI) and Positron Emission Tomography (PET). This framework quantifies the similarities between both modalities and relates them with the diagnostic label during the training process. The resulting latent space at the output of this network is then entered into an attention module in order to evaluate the relevance of each brain region at different stages of the development of Alzheimer's disease. The excellent results obtained and the high flexibility of the method proposed allow fusing more than two modalities, leading to a scalable methodology that can be used in a wide range of contexts.

Mathematical Assessment of Machine Learning Models Used for Brain Tumor Diagnosis

The brain is an intrinsic and complicated component of human anatomy. It is a collection of connective tissues and nerve cells that regulate the principal actions of the entire body. Brain tumor cancer is a serious mortality factor and a highly intractable disease. Even though brain tumors are not considered a fundamental cause of cancer deaths worldwide, about 40% of other cancer types are metastasized to the brain and transform into brain tumors. Computer-aided devices for diagnosis through magnetic resonance imaging (MRI) have remained the gold standard for the diagnosis of brain tumors, but this conventional method has been greatly challenged with inefficiencies and drawbacks related to the late detection of brain tumors, high risk in biopsy procedures, and low specificity. To circumvent these underlying hurdles, machine learning models have recently been developed to enhance computer-aided diagnosis tools for advanced, precise, and automatic early detection of brain tumors. This study takes a novel approach to evaluate machine learning models (support vector machine (SVM), random forest (RF), gradient-boosting model (GBM), convolutional neural network (CNN), K-nearest neighbor (KNN), AlexNet, GoogLeNet, CNN VGG19, and CapsNet) used for the early detection and classification of brain tumors by deploying the multicriteria decision-making method called fuzzy preference ranking organization method for enrichment evaluations (PROMETHEE), based on selected parameters, in this study: prediction accuracy, precision, specificity, recall, processing time, and sensitivity. To validate the results of our proposed approach, we performed a sensitivity analysis and cross-checking analysis with the PROMETHEE model. The CNN model, with an outranking net flow of 0.0251, is considered the most favorable model for the early detection of brain tumors. The KNN model, with a net flow of -0.0154, is the least appealing option. The findings of this study support the applicability of the proposed approach for making optimal choices regarding the selection of machine learning models. The decision maker is thus afforded the opportunity to expand the range of considerations which they must rely on in selecting the preferred models for early detection of brain tumors.

Histopathological diagnosis of colon cancer using micro-FTIR hyperspectral imaging and deep learning

Current studies based on digital biopsy images have achieved satisfactory results in detecting colon cancer despite their limited visual spectral range. Such methods may be less accurate when applied to samples taken from the tumor margin region or to samples containing multiple diagnoses. In contrast with the traditional computer vision approach, micro-FTIR hyperspectral images quantify the tissue-light interaction on a histochemical level and characterize different tissue pathologies, as they present a unique spectral signature. Therefore, this paper investigates the possibility of using hyperspectral images acquired over micro-FTIR absorbance spectroscopy to characterize healthy, inflammatory, and tumor colon tissues. The proposed method consists of modeling hyperspectral data into a voxel format to detect the patterns of each voxel using fully connected deep neural network. A web-based computer-aided diagnosis tool for inference is also provided. Our experiments were performed using the K-fold cross-validation protocol in an intrapatient approach and achieved an overall accuracy of 99% using a deep neural network and 96% using a linear support vector machine. Through the experiments, we noticed the high performance of the method in characterizing such tissues using deep learning and hyperspectral images, indicating that the infrared spectrum contains relevant information and can be used to assist pathologists during the diagnostic process.

SELF: a stacked-based ensemble learning framework for breast cancer classification

Nowadays, breast cancer is the most prevalent and jeopardous disease in women after lung cancer. During the past few decades, a substantial amount of cancer cases have been reported throughout the world. Breast cancer has been a widely acknowledged category of cancer disease in women due to the lack of awareness. According to the world cancer survey report 2020, about 2.3 million cases and 685,000 deaths have been reported worldwide. As, the patient-doctor ratio (PDR) is very high; consequently, there is an utmost need for a machine-based intelligent breast cancer diagnosis system that can detect cancer at its early stage and cure it more efficiently. The plan is to assemble scientists in both the restorative and the machine learning fields to progress toward this clinical application. This paper presents SELF, a stacked-based ensemble learning framework, to classify breast cancer at an early stage from the histopathological images of tumor cells with computer-aided diagnosis tools. In this work, we use the BreakHis dataset with 7909 histopathological images and Wisconsin Breast Cancer Database (WBCD) with 569 instances for the performance evaluation of our proposed framework. We have trained several distinct classifiers on both datasets and selected the best five classifiers on the basis of their accuracy measure to create our ensemble model. We use the stacking ensemble technique and consider the Extra tree, Random Forest, AdaBoost, Gradient Boosting, and KNN9 classifiers as the base learners, and the logistic regression model as a final estimator. We have evaluated the performance of SELF on the BreakHis dataset and WBCD datasets and achieved testing accuracy of approximately 95% and 99% respectively. The result of the other performance parameters on the BreakHis and the WBCD datasets also showed that our proposed framework outperforms with F1-Score, ROC, and MCC scores.

Role of Ensemble Deep Learning for Brain Tumor Classification in Multiple Magnetic Resonance Imaging Sequence Data.

The biopsy is a gold standard method for tumor grading. However, due to its invasive nature, it has sometimes proved fatal for brain tumor patients. As a result, a non-invasive computer-aided diagnosis (CAD) tool is required. Recently, many magnetic resonance imaging (MRI)-based CAD tools have been proposed for brain tumor grading. The MRI has several sequences, which can express tumor structure in different ways. However, a suitable MRI sequence for brain tumor classification is not yet known. The most common brain tumor is 'glioma', which is the most fatal form. Therefore, in the proposed study, to maximize the classification ability between low-grade versus high-grade glioma, three datasets were designed comprising three MRI sequences: T1-Weighted (T1W), T2-weighted (T2W), and fluid-attenuated inversion recovery (FLAIR). Further, five well-established convolutional neural networks, AlexNet, VGG16, ResNet18, GoogleNet, and ResNet50 were adopted for tumor classification. An ensemble algorithm was proposed using the majority vote of above five deep learning (DL) models to produce more consistent and improved results than any individual model. Five-fold cross validation (K5-CV) protocol was adopted for training and testing. For the proposed ensembled classifier with K5-CV, the highest test accuracies of 98.88 ± 0.63%, 97.98 ± 0.86%, and 94.75 ± 0.61% were achieved for FLAIR, T2W, and T1W-MRI data, respectively. FLAIR-MRI data was found to be most significant for brain tumor classification, where it showed a 4.17% and 0.91% improvement in accuracy against the T1W-MRI and T2W-MRI sequence data, respectively. The proposed ensembled algorithm (MajVot) showed significant improvements in the average accuracy of three datasets of 3.60%, 2.84%, 1.64%, 4.27%, and 1.14%, respectively, against AlexNet, VGG16, ResNet18, GoogleNet, and ResNet50.

Ultrasound detection of subquadricipital recess distension

Joint bleeding is a common occurrence for people with hemophilia and, if untreated, can result in hemophilic arthropathy. Ultrasound imaging has recently emerged as an effective tool to diagnose the distension of a joint recess caused by joint bleeding. However, no computer-aided diagnosis tool exists to support the practitioner in the diagnosis process. This paper addresses the problem of automatically detecting the subquadricipital recess of the knee and assessing whether it is distended in ultrasound images collected from patients with hemophilia. After framing the problem, we propose two different approaches: the first one adapts a one-stage object detection algorithm for the task, while the second one is a multi-task approach with a classification and a detection branch. The experimental evaluation, conducted with 483 annotated images, shows that the solution based on object detection alone has a balanced accuracy score of 0.74 with a mean IoU value of 0.66, while the multi-task approach has a higher balanced accuracy value (0.78) at the cost of a slightly lower mean IoU value.

Novel Computer-Aided Diagnosis System for the Early Detection of Alzheimer’s Disease

Aging is a natural process that leads to debility, disease, and dependency. Alzheimer’s disease (AD) causes degeneration of the brain cells leading to cognitive decline and memory loss, as well as dependence on others to fulfill basic daily needs. AD is the major cause of dementia. Computer-aided diagnosis (CADx) tools aid medical practitioners in accurately identifying diseases such as AD in patients. This study aimed to develop a CADx tool for the early detection of AD using the Intelligent Water Drop (IWD) algorithm and the Random Forest (RF) classifier. The IWD algorithm an efficient feature selection method, was used to identify the most deterministic features of AD in the dataset. RF is an ensemble method that leverages multiple weak learners to classify a patient’s disease as either demented (DN) or cognitively normal (CN). The proposed tool also classifies patients as mild cognitive impairment (MCI) or CN. The dataset on which the performance of the proposed CADx was evaluated was sourced from the Alzheimer’s Disease Neuroimaging Initiative (ADNI). The RF ensemble method achieves 100% accuracy in identifying DN patients from CN patients. The classification accuracy for classifying patients as MCI or CN is 92%. This study emphasizes the significance of pre-processing prior to classification to improve the classification results of the proposed CADx tool.