Search Results (301)

The rapid advancement of machine learning and artificial intelligence (AI) has created new opportunities to enhance diagnostic accuracy in dermatology, particularly within primary care settings. Computer-aided diagnosis (CAD) systems have demonstrated potential to support General Practitioners (GPs) by enabling earlier and more consistent identification of skin diseases. This review critically examines the literature on explainable artificial intelligence (XAI) for skin disease classification, with a specific focus on the evolution of explainability frameworks and the methodological implications of dataset selection. A comprehensive review of studies published between 2020 and 2025 was conducted across multiple academic databases, encompassing research on skin lesion detection, classification, and monitoring. The analysis reveals that deep learning architectures, particularly those leveraging transfer learning with models such as EfficientNet, ResNet, and Xception, frequently report high classification accuracies—often exceeding 90% when evaluated on single benchmark datasets. However, studies employing multiple datasets consistently demonstrate more stable and generalisable performance, albeit with modest reductions in reported accuracy, highlighting a critical trade-off between performance optimisation and real-world robustness. The review further identifies a clear temporal progression in the adoption of XAI techniques. Early studies relied on a broader range of post hoc explainability while later work increasingly consolidated around Grad-CAM, SHAP, and related attribution techniques, followed by gradual diversification into more specialised frameworks such as TCAVs (Testing with Concept Activation Vectors) and Prototype-based Networks. Despite these advances, the lack of clinically grounded explanations, limited integration of ethical considerations, and reliance on non-clinical imagery continue to constrain clinical applicability which we have explored using a GRADE-style narrative. Notably, evidence suggests that CAD systems can improve GP diagnostic accuracy for conditions such as melanoma and seborrhoeic keratosis; however, sustained clinical adoption remains contingent on transparent, reliable, and context-aware explainability mechanisms. Full article

Background/Objectives: Oral cancer is one of the most common types of cancer, with high mortality rates if not detected early. Traditional diagnostic methods based on clinical examination rely on experience, leading to delays in early and reliable diagnosis. In recent years, medical imaging and AI-based computer-aided diagnostic systems have shown promising results in the automated identification of oral cancer. In particular, the efficient management of high-dimensional feature spaces in machine learning and deep learning approaches directly impacts classification performance. In this context, metaheuristic-based feature selection technics is a critical component because of eliminating redundant and irrelevant features. To address these challenges, this study proposes a metaheuristic-based feature selection method to reduce feature dimensionality and enhance the classification performance of oral cancer detection. Methods: This study proposes an improved Honey Badger Algorithm-based feature selection approach for the automated detection of oral cancer. In the proposed method, the distance vector used in the HBA method has been redefined to improve the balance between exploration and exploitation. Additionally, a new Cauchy mutation-based migration strategy was integrated into the proposed method to increase diversity in the search space and avoid getting stuck in local minima. The continuous-valued iHBA method was discretized with a modified sin–cos transfer function for feature selection. Oral cancer images were filtered using the CLAHE method, and after extracting deep features with the ResNet50 architecture, the proposed metaheuristic-based method was used to select discriminative features. Results: The proposed method was first tested for reliability and limitations through repeated runs on problems with different characteristics, such as unimodal and multimodal classical test functions. Then, the method was applied to extract significant features for oral cancer detection using a Histopathological Imaging Database containing 1224 histopathological oral tissue images at 100× and 400× magnification levels from 230 patients. The proposed approach was assessed in terms of accuracy, precision, recall, F1-score, and convergence curves in comparison with various classical feature selection techniques, such as wrapper-based, filter-based, and embedded-based methods, as well as other metaheuristic-based methods. The experimental results demonstrated that the suggested strategy outperformed both traditional feature selection techniques and alternative metaheuristic approaches. Conclusions: The effectiveness of the proposed method in improving diagnostic accuracy was evaluated through comprehensive experimental analyses. The obtained findings show that the proposed iHBA-based feature selection approach can reduce feature dimensionality, eliminate redundant and irrelevant features, and improve the classification performance of oral cancer detection. Therefore, the proposed method provides an effective and competitive computer-aided diagnostic framework for the automated classification of histopathological oral cancer images. Full article

Background/Objectives: Hysteroscopy is the gold standard for visualization and treatment of intrauterine pathology. Because hysteroscopic interpretation remains operator-dependent, artificial intelligence (AI) has been evaluated as a tool to improve consistency, lesion recognition, and decision support. We aimed to systematically review AI, machine learning (ML), deep learning (DL), or computer-aided diagnosis (CAD) applications in hysteroscopy. Methods: A systematic search of PubMed/MEDLINE and EBSCOhost was performed from database inception to 8 March 2026, supplemented by targeted searches. Risk of bias was assessed using QUADAS-2 (diagnostic), PROBAST (prognostic), RoB2, and structured technical quality domains. Results: Nineteen primary studies were included, covering five areas: diagnostic classification and object detection (n = 8), real-time lesion detection and localization (n = 4), segmentation and visual-field support (n = 3), operative guidance (n = 1), and prognostic or decision-support applications (n = 3). Performance was highest in narrowly defined binary tasks and in large multicenter systems (e.g., ECCADx: AUC 0.979 internal, 0.975 external) and in prognostic fertility-prediction models after hysteroscopic adhesiolysis (AUC up to 0.992). Broader multiclass classification of heterogeneous lesions showed uneven and lower performance. Most studies were single-center, retrospective, and lacked external validation. Only one randomized study linked AI support to measurable procedural outcomes. Conclusions: The available studies indicate good technical performance in selected hysteroscopic tasks, particularly binary classification, focal lesion detection, and postoperative fertility stratification. Current evidence, however, remains limited by retrospective design, operator-dependent image acquisition, inconsistent validation, and scarce outcome-based clinical testing. In the short term, the most likely role of these systems is to support image interpretation, improve visual quality control, highlight suspicious lesions, and integrate hysteroscopic findings with complementary clinical data. Full article

Background/Objectives: The early identification of skin cancer by standard RGB dermoscopy is a clinical difficulty because of the complex visual differences between impacted lesions and healthy tissue. Methods: For the biomedical challenge, a novel approach to signal processing and image reconstruction is introduced in this study, called the spectrum-aided visual enhancer (SAVE). The proposed SAVE mechanism aims at reconstructing the diagnostically relevant spectral information from the conventional RGB dermoscopic images using the principles of hyperspectral imaging (HSI) and band selection (BS). After quality control and pre-processing, the images in the ISIC2019 dataset were selected, with 865 images that contain basal cell carcinoma (BCC), seborrheic keratosis (SK), and actinic keratosis (AK) lesions. To reduce data leakage, the dataset was split into training, validation, and testing subsets of 70%, 20%, and 10%, respectively. Five supervised deep learning object detection models were trained and tested on the conventional RGB image dataset and on the SAVE-enhanced dataset. Five supervised deep learning object detection models, namely, YOLOv8, YOLOv10, YOLOv11, SSDLite, and SSD, were trained and tested on the conventional RGB image dataset and the SAVE-enhanced dataset. Additional repeated experimental assessments and statistical comparisons were also carried out to evaluate the improvement in performance. Results: The experimental results showed that the SAVE-based pre-processing always yielded better performance in terms of lesion detection than conventional RGB image processing. The SAVE framework for SSD was evaluated and compared with all other evaluated models and was found to be the most successful, with an accuracy of 96%, a precision of 97%, a recall of 96%, and an F1 score of 96%. Conclusions: The results indicate that the proposed SAVE framework could be a promising RGB-compatible spectral enhancement technique for boosting skin cancer detection and computer-aided dermatologic analysis with the aid of AI. Full article

Lung cancer is the deadliest cause of cancer-related deaths worldwide, and early and accurate detection is key to improving patient outcomes. IQ-OTH/NCCD CT scan images are used in this study to present an optimized computer-aided diagnosis (CAD) framework for lung cancer detection. In order to extract deep features and improve diagnostic accuracy, a weighted geometric mean (WGM) ensemble of pretrained convolutional neural networks (CNNs) called the LCD-VRD model—comprising VGG16, ResNet50V2, and DenseNet121—provides robust feature extraction and strong generalization capabilities for accurately classifying normal, benign, and malignant (cancerous) cases. To actively mitigate data imbalance and reduce model overfitting, real-time data augmentation alongside rigorous class weighting was implemented. The results show that, with 97.27% accuracy and a 97.24% F1-score, the WGM ensemble of these models performs exceptionally well. Furthermore, Gradient-weighted Class Activation Mapping (Grad-CAM) visualization was investigated on CT images to provide an exploratory qualitative visualization of the image regions associated with model predictions. While the proposed framework shows promise as an effective tool for automated lung cancer diagnosis, its validation is currently limited to the IQ-OTH/NCCD dataset. External dataset evaluation will be essential to fully establish robustness and clinical applicability. Full article

Background/Objectives: Accurate differentiation of benign melanocytic nevi from invasive melanoma in dermato-oncology directly informs biopsy decisions and oncological management. Vision–language models (VLMs) are increasingly explored for image-based skin cancer assessment, but their diagnostic reliability and robustness to adversarial input manipulation remain insufficiently characterized. We evaluated three contemporary VLMs for diagnostic performance and susceptibility to single-word adversarial input manipulation (prompt injection) on dermoscopic images of histopathologically confirmed lesions. Methods: Fifty-two dermoscopic images (26 benign melanocytic nevi, 26 invasive melanomas) were analyzed using Claude Opus 4.7, Gemini 3.1 Pro, and GPT-5.4 under four conditions: an unmodified baseline and three adversarial conditions with a single opposite-of-ground-truth label embedded as a visual overlay, in image metadata, or both. Three independent rounds per image × model × condition yielded 1872 classifications across 52 lesions (independent diagnostic units) and 16,848 structured-output observations in total. Results: Baseline diagnostic accuracy ranged from 58.3% to 62.2%, with asymmetric sensitivity and specificity, including a pronounced benign-labeling bias in one model that missed 22 of 26 invasive melanomas. All adversarial conditions reduced accuracy to near-zero levels (0.0–1.9%; all p < 10⁻⁷ after Bonferroni correction). Repeated queries produced identical incorrect outputs in 98–100% of cases (Fleiss’ κ 0.97–1.00). Non-diagnostic outputs remained largely unchanged, and self-reported confidence did not decrease. Conclusions: Contemporary VLMs show limited baseline performance and marked vulnerability to minimal adversarial input in dermoscopic skin cancer assessment. The failure selectively alters the malignancy decision while preserving surrounding outputs and confidence, indicating that, within the conditions evaluated here, these systems do not currently appear suitable for unsupervised clinical use in dermato-oncology in the absence of input-integrity safeguards and qualified human oversight. Full article

Objective: Chiari Type I Malformation (CM1) is a structural abnormality of the hindbrain that can cause a range of neurological symptoms and often requires radiological confirmation using magnetic resonance imaging (MRI). The aim of this study was to develop and evaluate a deep feature-based machine learning framework for the automated detection of CM1 from sagittal MRI images. Materials and Methods: The cohort comprised 550 adults: 250 patients with CM1 (168 women, 82 men; age range, 18–65 years) and 300 healthy control participants (210 women, 90 men; age range, 18–65 years). A total of 764 T2-weighted sagittal MR images (384 CM1, 380 healthy) acquired from two different 1.5T MRI scanners (Siemens Magnetom Altea and Symphony) between 2020 and 2024 were retrospectively analyzed. Deep features were extracted using ResNet-50 and MobileNetV2 architectures and subsequently classified using Support Vector Machines (SVM), Logistic Regression (LR), Random Forest (RF), XGBoost, and voting-based ensemble models. Model performance was assessed through patient-level 5-fold cross-validation using accuracy, sensitivity, specificity, F1-score, PPV, NPV, and AUC metrics. Code and trained models are available from the corresponding author upon reasonable request; imaging data are not publicly available due to patient privacy and institutional restrictions. Results: Across patient-level five-fold cross-validation, models built on ResNet-50 deep features demonstrated extremely high and stable diagnostic performance. The final soft-voting ensemble classifier based on ResNet-50 achieved perfect mean performance, with accuracy, balanced accuracy, sensitivity, specificity, F1-score, and AUC all equal to 1.000 ± 0.000 across folds. Other ResNet-based classifiers also achieved near-perfect results. MobileNetV2-based models also demonstrated strong performance but showed slightly lower stability compared with ResNet-based models, with mean accuracies ranging from 0.984 to 0.993 and mean AUC values between 0.99947 and 0.99984 across classifiers. Conclusions: The proposed deep feature-based machine learning framework demonstrated excellent performance for the automated detection of Chiari Type I Malformation from sagittal MRI images. In particular, the ResNet-50–based soft-voting ensemble model achieved perfect classification performance in cross-validation testing, suggesting that deep feature representations combined with machine learning classifiers may serve as a promising computer-aided diagnostic tool for supporting radiological evaluation of CM1. Full article

Background/Objectives: Esophageal cancer is one of the major global causes of cancer mortality, and the 5-year survival rate remains below 20% because many cases are detected late. In this study, a Spectral-Aided Vision Enhancer (SAVE) algorithm was utilized to convert conventional white-light endoscopic images (WLI) into hyperspectral-like narrow-band imaging (NBI) images for machine-learning classification of Dysplasia, Normal, and Squamous Cell Carcinoma (SCC). Methods: A total of 762 WLI images obtained from Kaohsiung Medical University were augmented to 1074 using the Al bumentations library, employing vertical flipping, horizontal flipping, and rotations. The SAVE conversion pipeline employs a 24-patch Macbeth color checker for calibration, γ-correction, CIE XYZ transformation, and multivariate regression to interpolate spectral bands, yielding an average color difference of 2.79 (CIEDE2000) from true NBI. The training outcomes and performance metrics illustrate the versatility of the machine learning/deep learning models—Random Forest (RF), Support Vector Machine (SVM), and Convolutional Neural Network (CNN)—which were trained and evaluated on both the original WLI and SAVE datasets. Performance metrics were analyzed based on precision, recall, accuracy, and F1-score. Results: The CNN sample achieved an accuracy of 100 percent on SAVE data, compared to 93 percent for WLI. The accuracy of RF improved, with WLI at 91% and SAVE at 96%, while SVM increased from 79% to 84%. These improvements indicate the diagnostically valuable spectral variations that can be amplified with SAVE, resulting in significant enhancements in pre-cancer/SCC sensitivity. Conclusions: The proposed SAVE method demonstrates significant potential for enhancing endoscopic imaging and advancing computer-aided diagnosis in esophageal cancer screening, with applicability in other gastrointestinal imaging scenarios as well. Full article

Background: Accurate colposcopy assessment is essential for detection of cervical precancerous lesions. However, the diagnostic performance depends heavily on the examiner’s experience and subjective interpretation. Recent advances in artificial intelligence offer new opportunities for automated image analysis. In particular, deep learning models have shown promise for colposcopy image analysis intended for precancer screening. However, their limited interpretability restricts their translational value in clinical gynecology. This study leverages a custom dataset of expert-annotated digital colposcopic images to quantify the diagnostic strengths of these architectures. Methods: A comparative analysis was provided of the well-established convolutional neural network YOLOv11 and the modern transformer-based RF-DETR architecture, both of which were trained for the segmentation of 10 distinct classes. Target objects included anatomical structures, medical instruments and colposcopic findings. Results: Our results demonstrate that the YOLO architecture provides better performance for anatomical structures, whereas the RF-DERT reaches higher scores for colposcopy findings. These findings demonstrate transformer architecture superiority in more nuanced segmentation of clinically relevant findings, providing an interesting framework for computer-aided decision support systems in colposcopy. Conclusions: Both YOLOv11 and RF-DETR enable effective segmentation of colposcopic images, with performance dependent on class size and characteristics. Best results were achieved for large anatomical structures, while small and underrepresented findings remained challenging due to class imbalance. YOLO offers greater stability and efficiency, whereas RF-DETR performs better on more complex cases. Limitations include data imbalance, variable image quality, and annotation inconsistencies; future work should address these issues to improve generalization. Full article

Accurate and efficient classification of hematological malignancies from peripheral blood smear (PBS) images remains challenging due to the scarcity of annotated datasets, staining variability, and subtle morphological differences among blood cancer subtypes. To address these limitations, this study proposes an Advanced Lightweight Deep Learning (ALDL) framework for the multi-class classification of Acute Lymphoblastic Leukemia (ALL) across four clinically significant stages: Benign, Pro-B, Pre-B, and Early Pre-B. The framework integrates EfficientNetV2-S with Convolutional Block Attention Modules (CBAM) to enhance spatial and channel-wise feature refinement. At the same time, Focal Loss is employed to mitigate class imbalance by prioritizing hard-to-classify samples. A robust preprocessing pipeline, including CLAHE contrast enhancement, Reinhard stain normalization, and data augmentation, improves feature visibility and dataset generalization. Lesion segmentation is performed using RGB-based thresholding and watershed overlay, followed by lesion-level cropping to ensure consistency across inputs. Experimental evaluations on the ALL-DB dataset demonstrate the superior performance of the proposed method, achieving an average accuracy of 96.11%, an F1-score of 95.99%, and an AUC of 0.9875. Comparative analyses against MobileNetV3, ResNet50, DenseNet121, VGG16, and InceptionV3 confirm that the proposed segmentation-guided EfficientNetV2-S + CBAM + Focal Loss framework consistently outperforms conventional CNN architectures across both 70:30 and 60:40 train–test splits. Furthermore, a detailed investigation of color spaces (RGB, HSV, LAB, and HED) indicates that RGB yields the most reliable segmentation and classification results. At the same time, HED enhances lesion visualization at the expense of higher computational cost. The proposed ALDL framework demonstrates strong potential for real-world application as a computer-aided diagnostic (CAD) system for early leukemia detection, offering improved diagnostic reliability, reduced error rates, and practical scalability for clinical environments. Full article

Brain tumor detection from magnetic resonance imaging (MRI) is fundamental to computer-aided diagnosis, yet automated models must remain robust to heterogeneous imaging conditions. Despite strong recent progress, many deep learning and transformer-based approaches primarily optimize performance accuracy without explicitly improving feature selectivity and spatial localization, and they typically produce deterministic output without uncertainty estimates, which limits reliability. To overcome these limitations, we introduce UA-EffNet-DA, an uncertainty-aware EfficientNet framework that addresses these limitations through three complementary components. First, EfficientNet-B4 serves as an efficient backbone for discriminative feature extraction. Second, lightweight dual attention modules, comprising channel and spatial attention in parallel, are applied to the model to emphasize what and where discriminative features to focus within MRI slices. Third, Monte Carlo dropout is employed during inference to quantify predictive uncertainty and enable confidence-aware decision. Experiments on two public benchmarks demonstrate strong performance, yielding accuracies of 98.73% on the Figshare dataset and 99.23% on the Kaggle dataset. In addition, explainable AI analysis using Gradient-weighted Class Activation Mapping (Grad-CAM) further indicates that the proposed model concentrates on diagnostically relevant tumor regions rather than background structures, supporting transparent decision-making. Ablation studies confirm the complementary contribution of dual attention refinement and uncertainty-aware inference. Overall, the proposed UA-EffNet-DA framework offers an efficient and interpretable approach for brain tumor detection that supports more reliable clinical decision support through uncertainty-aware predictions. Full article

Early detection of dental caries in children is critical to prevent irreversible tooth damage and guarantee optimal oral health outcomes. However, interpreting pediatric panoramic radiographs throughout the mixed dentition stage remains a very challenging task due to overlap in anatomical structures and developmental variability. This complexity underscores the need for well curated, representative datasets that enable the development of reliable computer-aided diagnostic models. Herein, this study introduces the Mixed Dentition Orthopantomogram Dataset, a newly developed, publicly available dataset of children that was carefully labeled by dental specialists to identify proximal and occlusal caries regions in the range of 3–12 years. To evaluate the dataset’s applicability for artificial intelligence research, we benchmarked it using both classification and segmentation models. A patch-based classifier achieved an average AUC of 0.89 and Recall 0.85 in distinguishing healthy and carious regions. For segmentation, we evaluated U-Net and Attention U-Net with multiple loss functions, and the Attention U-Net trained with Focal loss achieved the best Dice score of 0.94. Collectively, these findings support the dataset’s utility for pediatric caries analysis and demonstrate the viability of deep learning approaches for mixed dentition panoramic imaging. Full article

Artificial intelligence (AI) is increasingly used to support tuberculosis (TB) screening and diagnosis, particularly through computer-aided detection (CAD) applied to chest radiography (CXR). However, the programmatic value of AI depends not only on diagnostic accuracy but also on implementation context, threshold calibration, and integration into diagnostic pathways. We conducted a narrative, state-of-the-art review of AI applications across the TB diagnosis pathway. Evidence was synthesized from World Health Organization policy documents, independent validation initiatives, and peer-reviewed studies published between 2010 and 2026, with a structured selection process aligned with PRISMA principles. CAD for CXR is the most mature AI application and is recommended by WHO for TB screening and triage among individuals aged ≥15 years in specific contexts. Across studies, CAD-CXR demonstrates sensitivity comparable to human readers, although performance varies by product, population, and imaging conditions, necessitating local threshold calibration. Evidence from implementation studies suggests improvements in screening efficiency and potential cost-effectiveness in high-burden settings. Other AI modalities, including computed tomography (CT)-based imaging analysis, point-of-care ultrasound interpretation, cough or stethoscope sound analysis, clinical risk models, and genomic resistance prediction show promising but heterogeneous results, with most requiring further independent validation and prospective evaluation. AI has the potential to strengthen TB screening and diagnostic pathways, but its impact depends on integration into health systems and evaluated using patient- and program-level outcomes rather than accuracy alone. A differentiated approach is needed, with responsible scale-up of policy-endorsed tools alongside rigorous evaluation of emerging technologies to support effective and equitable TB care. Full article

Diabetic retinopathy is a prevalent condition and a major public health concern due to its detrimental impact on eyesight. Diabetes is a root cause of its development and damages small blood vessels caused by prolonged high blood sugar levels. The degenerative consequences of diabetic retinopathy are irrevocable if not diagnosed in the early stages of its progression. This ailment triggers the development of retinal lesions, which can be identified for diagnosis and prognosis. However, lesion detection is challenging due to their similarity in intensity profiles to other retinal features, inconsistent sizes, and random locations. This research evaluates a custom deep learning network for classifying retinal images and compares it with the state-of-the-art classifiers. The novel preprocessing method is introduced to reduce the complexity of the diagnostic process and to enhance classification performance by adaptively enhancing images. Despite being a shallow network, the proposed model yields competitive results with an accuracy of 87.66% and an F1-score of 0.78. The evaluation metrics indicate that class imbalance affects the performance of the proposed model despite using the weighted cross-entropy loss. The future contribution will be the inclusion of generative adversarial networks for generating synthetic images to balance the dataset. This research aims to develop a robust computer-aided diagnostic system as a second interpreter for ophthalmologists during the diagnosis and prognosis stages. Full article

Objective: The integration of artificial intelligence (AI) into computer-aided detection (CAD) is a major innovation in lung cancer diagnosis. However, its reliability in detecting the earliest radiographic sign—faint ground-glass opacities (GGOs) indicating pre-invasive adenocarcinoma—remains a critical, unquantified gap. This study aimed to perform a rigorous failure analysis to define the specific conditions under which commercial AI/CAD systems fail in a low-dose CT (LDCT) screening setting. Methods: In this retrospective diagnostic accuracy study, a primary cohort of 100 patients and an external validation cohort of 50 patients with moderate/low-risk nodules on LDCT were included. An expert reference standard was established by a consensus panel of three thoracic radiologists. Two independent, commercially deployed AI/CAD systems from different vendors (Vendor A & Vendor B) processed all cases. Nodules confirmed by experts but missed by AI were analyzed. Their morphology was categorized, and their mean CT attenuation (HU) was measured via manual region-of-interest placement. Results: The AI systems demonstrated significant and comparable false negative rates in the combined cohort: 12.7% for Vendor A and 14.7% for Vendor B. The vast majority of missed nodules were GGOs (92.3% and 78.6%, respectively, in the primary cohort). Crucially, quantitative analysis revealed a consistent density threshold for AI failure: the mean CT value of missed GGOs was −737 ± 51.50 HU for Vendor A and −727 ± 70.07 HU for Vendor B. This algorithmic blind spot was fully corroborated by the external validation cohort (−741 ± 48.2 HU and −733 ± 62.5 HU, respectively). Anatomical complexity (juxta-pleural/endobronchial location) was a secondary failure factor. Conclusions: This study identifies a quantifiable “−730 HU blind spot” as a common limitation of current commercial AI/CAD systems in diagnosing early lung adenocarcinoma. This finding represents a pivotal advancement in understanding AI’s role in diagnostics: it is not infallible. To innovate and safeguard screening efficacy, radiologists must adopt a human–AI collaborative model with mandated manual verification targeting low-attenuation opacities, ensuring this diagnostic innovation fulfills its promise while mitigating the risks of overdiagnosis. Full article