MAMaria João Almeida

Maria João Almeida

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Maria João Almeida has a total of 100 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Deep learning-based classification of benign anorectal lesions on endoanal ultrasound: a proof-of-concept study

    by Maria João Almeida, Miguel Mascarenhas, Miguel Martins, Francisco Mendes, Joana Mota, Pedro Cardoso, António Pinto da Costa, Pedro Diaz Donoso, Pablo Farina, Matias La Francesca, et al.

  • Open Access

    Real-Time Application of Artificial Intelligence for Automatic Detection of High-Grade Squamous Intraepithelial Lesions During High-Resolution Anoscopy

    by Luis Barroso, Miguel Martins, Maria João Almeida, Joana Mota, Francisco Mendes, Ahsan Javed, Amine Alam, Nadia Fathallah, Pedro Diaz Donoso, Dolores Caffarena, et al.

    Background: High-resolution anoscopy (HRA) is the gold standard for anal cancer screening, but its interpretation is challenging and operator-dependent. Artificial intelligence (AI) may increase diagnostic yield, but most studies have focused on differentiating low-grade and high-grade squamous intraepithelial lesions (LSIL and HSIL, respectively) in still frames, with no clinical application reported. Methods: We describe the first real-time use of deep learning, demonstrated in three patients undergoing HRA for anal cancer screening, at a high-volume American referral center. When an area suggestive of HSIL appeared, a YOLO-based object detection model generated a bounding box. Results: The model detected only histologically confirmed HSIL and did not activate when no lesions or only LSIL were present. Conclusions: This report suggests that real-time AI-enhanced HRA is feasible and may improve lesion detection and differentiation, potentially representing a significant step forward in this demanding field, although multicentric validation studies are still needed.

  • Open Access

    Robotic Capsule Endoscopy: Simultaneous Gastric and Enteric Evaluation in Real-World Practice

    by Hélder Cardoso, Miguel Mascarenhas, Joana Mota, Miguel Martins, Maria João Almeida, Joana Frias, Catarina Cardoso Araújo, Francisco Mendes, Margarida Marques, Patrícia Andrade, et al.

    Background/Objectives: Robotic capsule endoscopy (RCE) is an emerging technology that combines magnetically controlled gastric navigation with conventional capsule enteroscopy (CE), enabling a minimally invasive, comprehensive evaluation of the upper- and mid-gastrointestinal tract. This study aimed to characterize the real-world implementation and diagnostic performance of RCE in a European tertiary referral center. Methods: A retrospective, single-center analysis was conducted on adult patients (≥18 years) who underwent RCE (Omom RC) between June 2023 and July 2025. Eligible patients had a clinical indication for small bowel CE and a concurrent requirement for diagnostic gastroscopy or reassessment of known gastric lesions. The RCE protocol comprised an initial robotic-guided gastric examination followed by passive transit through the small bowel. Results: A total of 85 patients were included (52% female), with a median age of 49 years (IQR 40–64). The most common indications were suspected or established inflammatory bowel disease (57%) and iron deficiency anemia (31%). Gastric preparation was rated at least fair in 98% of cases, with good preparation in 38%. Median gastric transit time was 74 min (IQR 35–106). Relevant gastric findings were identified in 39 cases (46%), namely polyps (18%) and angiectasias (8%, including one with active bleeding), in addition to signs of chronic gastritis. Thirteen patients underwent subsequent endoscopy, resulting in seven therapeutic procedures. Small bowel findings were present in 60 patients (71%), including P3 (active bleeding) in 3% and P2 lesions (angiectasias, ulcers, tumors, varices) in 39%. One moderate adverse event occurred: small bowel capsule retention in a patient with multifocal neuroendocrine tumor and ileostomy, requiring endoscopic intervention. Conclusions: Robotic capsule endoscopy is a feasible tool for dual-region gastrointestinal evaluation. It enables high-quality gastric visualization, facilitates early detection of clinically actionable lesions, and maintains the diagnostic yield expected from standard small bowel CE. These findings support the integration of RCE into diagnostic pathways for patients requiring simultaneous gastric and small bowel assessment.

  • Open Access

    Artificial Intelligence and FLIP Panometry—Automated Classification of Esophageal Motility Patterns

    by Miguel Mascarenhas, Francisco Mendes, João Rala Cordeiro, Joana Mota, Miguel Martins, Maria João Almeida, Catarina Araujo, Joana Frias, Pedro Cardoso, Ismael El Hajra, et al.

    Background/Objectives: Functional lumen imaging probe (FLIP) panometry allows real-time assessment of the esophagogastric junction opening and esophageal body contractile activity during an endoscopic procedure. Despite the development of the Dallas Consensus, FLIP panometry analysis remains complex. Artificial intelligence (AI) models have proven their benefit in high-resolution esophageal manometry; however, data on their role in FLIP panometry are scarce. This study aims to develop an AI model for automatic classification of motility patterns during a FLIP panometry exam. Methods: A total of 105 exams from five centers from both the European and American continents were included. Several machine learning models were trained and evaluated for detection of FLIP panometry patterns. Each exam was classified with an expert consensus-based decision according to the Dallas Consensus, with division into a training and testing dataset in a patient-split design. Models’ performance was evaluated through their accuracy and area under the receiver-operating characteristic curve (AUC-ROC). Results: Pathological planimetry patterns were identified by an AdaBoost Classifier with 84.9% accuracy and a mean AUC-ROC of 0.92. Random Forest identified disorders of the esophagogastric junction opening with 86.7% accuracy and an AUC-ROC of 0.973. The Gradient Boosting Classifier identified disorders of the contractile response with 86.0% accuracy and an AUC-ROC of 0.933. Conclusions: In this study, integrating exams with different probe sizes and demographic contexts, a machine learning model accurately classified FLIP panometry exams according to the Dallas Consensus. AI-driven FLIP panometry could revolutionize the approach to this exam during an endoscopic procedure, optimizing exam accuracy, standardization, and accessibility, and transforming patient management.


  • Open Access

    Artificial Intelligence Driven Diagnosis of Motility Patterns in High-Resolution Esophageal Manometry: A Multicentric Multidevice Study

    by Miguel Mascarenhas, Joana Mota, João Rala Cordeiro, Francisco Mendes, Miguel Martins, Pedro Cardoso, Maria João Almeida, Antonio Pinto da Costa, Ismael Hajra Martinez, Virginia Matallana Royo, et al.

    INTRODUCTION: Esophageal motility disorders (EMDs) are common in clinical practice, with a high symptomatic burden and significant impact on the patients' quality of life. High-resolution esophageal manometry (HREM) is the gold standard for the evaluation of functional esophageal disorders. The Chicago Classification offers a standardized approach to HREM. However, HREM remains a complex procedure, both in data analysis and in accessibility. This study aimed to develop and validate machine learning (ML) models to detect EMDs according to the Chicago Classification. METHODS: We retrospectively analyzed 618 HREM examinations from 3 centers (Spain and the United States) using 2 recording systems. Labels were assigned by expert consensus as either disorder present or absent for 2 categories: esophagogastric junction outflow disorders and peristalsis disorders. Several ML models were trained and evaluated. ML classifiers were developed using an 80/20 patient-level stratified split for training/validation and testing. Model selection was guided by internal evaluation through repeated 10-fold cross-validation. Model performance was assessed by accuracy and area under the receiver-operating characteristic curve (AUC-ROC). RESULTS: The GradientBoostingClassifier model outperformed the remaining ML models with an accuracy of 0.942 ± 0.015 and an AUC-ROC of 0.921 ± 0.041 for identifying disorders of esophagogastric junction outflow. The xGBClassifier model detected disorders of peristalsis with an accuracy of 0.809 ± 0.029 and an AUC-ROC of 0.871 ± 0.027. Performance was consistent across repeated validations, demonstrating model robustness and generalization. DISCUSSION: This multicenter, multidevice study demonstrates that ML models can accurately detect EMDs in HREM. Artificial intelligence-driven HREM may improve diagnosis by standardizing interpretation and reducing interobserver variability.


  • Open Access

    Artificial Intelligence for Lymph Node Detection and Malignancy Prediction in Endoscopic Ultrasound: A Multicenter Study

    by Belén Agudo Castillo, Miguel Mascarenhas Saraiva, António Miguel Martins Pinto da Costa, João Ferreira, Miguel Martins, Francisco Mendes, Pedro Cardoso, Joana Mota, Maria João Almeida, João Afonso, et al.

    Background/Objectives: Endoscopic ultrasound (EUS) is crucial for lymph node (LN) characterization, playing a key role in oncological staging and treatment guidance. EUS criteria for predicting malignancy are imprecise, and histologic diagnosis may have limitations. This multicenter study aimed to evaluate the effectiveness of a novel artificial intelligence (AI)–based system in predicting LN malignancy from EUS images. Methods: This multicenter study included EUS images from nine centers. Lesions were labeled (“malignant” or “benign”) and delimited with bounding boxes. Definitive diagnoses were based on cytology/biopsy or surgical specimens and, if negative, a minimum six-month clinical follow-up. A convolutional neural network (CNN) was developed using the YOLO (You Only Look Once) architecture, incorporating both detection and classification modules. Results: A total of 59,992 images from 82 EUS procedures were analyzed. The CNN distinguished malignant from benign lymph nodes with a sensitivity of 98.8% (95% CI: 98.5–99.2%), specificity of 99.0% (95% CI: 98.3–99.7%), and precision of 99.0% (95% CI: 98.4–99.7%). The negative and positive predictive values for malignancy were 98.8% and 99.0%, respectively. Overall diagnostic accuracy was 98.3% (95% CI: 97.6–99.1%). Conclusions: This is the first study evaluating the performance of deep learning systems for LN assessment using EUS imaging. Our AI-powered imaging model shows excellent detection and classification capabilities, emphasizing its potential to provide a valuable tool to refine LN evaluation with EUS, ultimately supporting more tailored, efficient patient care.


  • Open Access

    Artificial Intelligence and Colposcopy: Detection and Classification of Vulvar HPV-Related Low-Grade and High-Grade Squamous Intraepithelial Lesions

    by Miguel Mascarenhas, Vanitha Sivalingam, Inês Castro, Katie Jones, Miguel Martins, Inês Alencoão, Maria João Carinhas, Joana Mota, Pedro Cardoso, Francisco Mendes, et al.

    Background/Objectives: Accurate identification of vulvar high-grade squamous intraepithelial lesions (HSIL) is essential for preventing progression to invasive squamous cell carcinoma. This study addresses the gap in artificial intelligence (AI) applications for vulvar lesion diagnosis by developing and validating the first convolutional neural network (CNN) model to automatically detect and classify HPV-related vulvar lesions—specifically HSIL and low-grade squamous intraepithelial lesions (LSIL)—based on vulvoscopy images. Methods: This bicentric study included data from 28 vulvoscopies, comprising a total of 9857 annotated frames, categorized using histopathological reports (HSIL or LSIL). The dataset was divided into training, validation, and testing sets for development and assessment of a YOLOv11-based object detection model. Results: The CNN demonstrated a recall (sensitivity) of 99.7% and a precision (positive predictive value) of 99.1% for lesion detection and classification. Conclusions: This is the first AI model developed for detecting and classifying HPV-related vulvar lesions. The integration of such models into vulvoscopy could significantly improve diagnostic accuracy and positively impact women’s health by reducing the need for potentially invasive and anatomy-altering procedures.

  • Open Access

    Artificial Intelligence and Hysteroscopy: A Multicentric Study on Automated Classification of Pleomorphic Lesions

    by Miguel Mascarenhas, Carla Peixoto, Ricardo Freire, Joao Cavaco Gomes, Pedro Cardoso, Inês Castro, Miguel Martins, Francisco Mendes, Joana Mota, Maria João Almeida, et al.

    Background/Objectives: The integration of artificial intelligence (AI) in medical imaging is rapidly advancing, yet its application in gynecologic use remains limited. This proof-of-concept study presents the development and validation of a convolutional neural network (CNN) designed to automatically detect and classify endometrial polyps. Methods: A multicenter dataset (n = 3) comprising 65 hysteroscopies was used, yielding 33,239 frames and 37,512 annotated objects. Still frames were extracted from full-length videos and annotated for the presence of histologically confirmed polyps. A YOLOv1-based object detection model was used with a 70–20–10 split for training, validation, and testing. Primary performance metrics included recall, precision, and mean average precision at an intersection over union (IoU) ≥ 0.50 (mAP50). Frame-level classification metrics were also computed to evaluate clinical applicability. Results: The model achieved a recall of 0.96 and precision of 0.95 for polyp detection, with a mAP50 of 0.98. At the frame level, mean recall was 0.75, precision 0.98, and F1 score 0.82, confirming high detection and classification performance. Conclusions: This study presents a CNN trained on multicenter, real-world data that detects and classifies polyps simultaneously with high diagnostic and localization performance, supported by explainable AI features that enhance its clinical integration and technological readiness. Although currently limited to binary classification, this study demonstrates the feasibility and potential of AI to reduce diagnostic subjectivity and inter-observer variability in hysteroscopy. Future work will focus on expanding the model’s capabilities to classify a broader range of endometrial pathologies, enhance generalizability, and validate performance in real-time clinical settings.


  • Open Access

    Unlocking the Potential of AI in EUS and ERCP: A Narrative Review for Pancreaticobiliary Disease

    by Catarina Cardoso Araújo, Joana Frias, Francisco Mendes, Miguel Martins, Joana Mota, Maria João Almeida, Tiago Ribeiro, Guilherme Macedo, Miguel Mascarenhas

    Artificial Intelligence (AI) is transforming pancreaticobiliary endoscopy by enhancing diagnostic accuracy, procedural efficiency, and clinical outcomes. This narrative review explores AI’s applications in endoscopic ultrasound (EUS) and endoscopic retrograde cholangiopancreatography (ERCP), emphasizing its potential to address diagnostic and therapeutic challenges in pancreaticobiliary diseases. In EUS, AI improves pancreatic mass differentiation, malignancy prediction, and landmark recognition, demonstrating high diagnostic accuracy and outperforming traditional guidelines. In ERCP, AI facilitates precise biliary stricture identification, optimizes procedural techniques, and supports decision-making through real-time data integration, improving ampulla recognition and predicting cannulation difficulty. Additionally, predictive analytics help mitigate complications like post-ERCP pancreatitis. The future of AI in pancreaticobiliary endoscopy lies in multimodal data fusion, integrating imaging, genomic, and molecular data to enable personalized medicine. However, challenges such as data quality, external validation, clinician training, and ethical concerns—like data privacy and algorithmic bias—must be addressed to ensure safe implementation. By overcoming these challenges, AI has the potential to redefine pancreaticobiliary healthcare, improving diagnostic accuracy, therapeutic outcomes, and personalized care.


  • Open Access

    Evaluating ChatGPT-4 for the Interpretation of Images from Several Diagnostic Techniques in Gastroenterology

    by Miguel Mascarenhas Saraiva, Tiago Ribeiro, Belén Agudo, João Afonso, Francisco Mendes, Miguel Martins, Pedro Cardoso, Joana Mota, Maria Joao Almeida, António Costa, et al.

    Background: Several artificial intelligence systems based on large language models (LLMs) have been commercially developed, with recent interest in integrating them for clinical questions. Recent versions now include image analysis capacity, but their performance in gastroenterology remains untested. This study assesses ChatGPT-4’s performance in interpreting gastroenterology images. Methods: A total of 740 images from five procedures—capsule endoscopy (CE), device-assisted enteroscopy (DAE), endoscopic ultrasound (EUS), digital single-operator cholangioscopy (DSOC), and high-resolution anoscopy (HRA)—were included and analyzed by ChatGPT-4 using a predefined prompt for each. ChatGPT-4 predictions were compared to gold standard diagnoses. Statistical analyses included accuracy, sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and area under the curve (AUC). Results: For CE, ChatGPT-4 demonstrated accuracies ranging from 50.0% to 90.0%, with AUCs of 0.50–0.90. For DAE, the model demonstrated an accuracy of 67.0% (AUC 0.670). For EUS, the system showed AUCs of 0.488 and 0.550 for the differentiation between pancreatic cystic and solid lesions, respectively. The LLM differentiated benign from malignant biliary strictures with an AUC of 0.550. For HRA, ChatGPT-4 showed an overall accuracy between 47.5% and 67.5%. Conclusions: ChatGPT-4 demonstrated suboptimal diagnostic accuracies for image interpretation across several gastroenterology techniques, highlighting the need for continuous improvement before clinical adoption.