Matheus Ferreira de Carvalho

Dr. Matheus Ferreira de Carvalho

orcid

Research Focus & Metrics

Academic Career Length

3Years (2024-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
26 September 2024 in Clinical and Translational Gastroenterology
Verified
Journal Article
Open access
Miguel Mascarenhas SaraivaMiguel Mascarenhas Saraiva
Miguel Mascarenhas Saraiva
MRMariano Gonzalez-Haba Ruiz
Mariano Gonzalez-Haba Ruiz
JWJessica Widmer
Jessica Widmer
FMFrancisco Mendes
Francisco Mendes
TGTamas A. Gonda
Tamas A. Gonda
Guilherme MacedoGuilherme Macedo
Guilherme Macedo
INTRODUCTION: Endoscopic ultrasound (EUS) allows for characterization and biopsy of pancreatic lesions. Pancreatic cystic neoplasms (PCN) include mucinous (M-PCN) and nonmucinous lesions (NM-PCN). Pancreatic ductal adenocarcinoma (P-DAC) is the commonest pancreatic solid lesion (PSL), followed by pancreatic neuroendocrine tumor (P-NET). Although EUS is preferred for pancreatic lesion evaluation, its diagnostic accuracy is suboptimal. This multicentric study aims to develop a convolutional neural network (CNN) for detecting and distinguishing PCN (namely M-PCN and NM-PCN) and PSL (particularly P-DAC and P-NET). METHODS: A CNN was developed with 378 EUS examinations from 4 international reference centers (Centro Hospitalar Universitário São João, Hospital Universitario Puerta de Hierro Majadahonda, New York University Hospitals, Hospital das Clínicas Faculdade de Medicina da Universidade de São Paulo). About 126.000 images were obtained—19.528 M-PCN, 8.175 NM-PCN, 64.286 P-DAC, 29.153 P-NET, and 4.858 normal pancreas images. A trinary CNN differentiated normal pancreas tissue from M-PCN and NM-PCN. A binary CNN distinguished P-DAC from P-NET. The total data set was divided into a training and testing data set (used for model's evaluation) in a 90/10% ratio. The model was evaluated through its sensitivity, specificity, positive and negative predictive values, and accuracy. RESULTS: The CNN had 99.1% accuracy for identifying normal pancreatic tissue, 99.0% and 99.8% for M-PCN and NM-PCN, respectively. P-DAC and P-NET were distinguished with 94.0% accuracy. DISCUSSION: Our group developed the first worldwide CNN capable of detecting and differentiating the commonest PCN and PSL in EUS images, using examinations from 4 centers in 2 continents, minimizing the impact of the demographic bias. Larger multicentric studies are needed for technology implementation.

28 August 2025 in American Journal of Gastroenterology
Verified
Journal Article
Closed access
Miguel Mascarenhas SaraivaMiguel Mascarenhas Saraiva
Miguel Mascarenhas Saraiva
JFJ.P.S. Ferreira
J.P.S. Ferreira
JAJoão Afonso
João Afonso
FMFrancisco Mendes
Francisco Mendes
WSWilliam Preston Sonnier
William Preston Sonnier
Guilherme MacedoGuilherme Macedo
Guilherme Macedo
INTRODUCTION: Capsule endoscopy (CapE) is a minimally invasive procedure designed for small bowels' evaluation. However, prolonged reading times and a risk of missing clinically significant findings limit its potential. Prospective clinical validation studies of artificial intelligence (AI) for CapE remain scarce. Furthermore, existing studies focus on lesion detection, without addressing lesion differentiation. METHODS: The aim of a multicenter prospective validation study was to compare AI-assisted reading with conventional CapE reading. Three hundred thirty CapE videos from 3 devices across 7 centers and 4 countries were included. After conventional reading reports, AI-assisted reading was performed by an independent expert using a deep learning model to detect and differentiate pleomorphic small bowel lesions. Both reports were reviewed by an expert from an independent center, which decided in discrepant cases. AI-assisted and standard readings were evaluated through their accuracy, sensitivity, specificity, positive and negative predictive value, and small bowel lesion detection rate. RESULTS: AI-assisted reading detected 605 of 635 lesions identified by expert-based consensus, whereas standard reading identified 354 lesions. AI-assisted reading outperformed standard reading, with small bowel lesion detection rate of 96.1% vs 76.3% and sensitivity of 97.5% vs 78.2%. AI-assisted reading had a mean examination reading time of 203 seconds per examination. DISCUSSION: This was the first multicentric study proving AI-assisted CapE reading superiority compared with conventional reading. The inclusion of videos from multiple devices addresses the interoperability challenge, whereas including patients from 4 countries and 2 different continents assures a diverse demographic context. AI achieved gastroenterologist-level identification of small-bowel lesions, surpassing conventional reading methods in both lesion detection and characterization.

22 October 2025 in Cancers
Verified
Journal Article
Open access
PDF file
BCBelen Agudo Castillo
Belen Agudo Castillo
Miguel Mascarenhas SaraivaMiguel Mascarenhas Saraiva
Miguel Mascarenhas Saraiva
ACAntónio Miguel Martins Pinto da Costa
António Miguel Martins Pinto da Costa
JFJoão Ferreira
João Ferreira
MMMiguel Martins
Miguel Martins
MRMariano Gonzalez-Haba Ruiz
Mariano Gonzalez-Haba Ruiz
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.

24 July 2024 in Inflammatory Bowel Diseases
Verified
Journal Article
Closed access
MCMatheus Ferreira de Carvalho
Matheus Ferreira de Carvalho
ACAlexandre de Sousa Carlos
Alexandre de Sousa Carlos
AKAngelo So Taa Kum
Angelo So Taa Kum
ABAlexandre Moraes Bestetti
Alexandre Moraes Bestetti
IGIgor Logetto Caetité Gomes
Igor Logetto Caetité Gomes
EMEduardo Guimarães Hourneaux de Moura
Eduardo Guimarães Hourneaux de Moura
Background: Crohn’s disease (CD) is a chronic inflammatory condition of the gastrointestinal tract. Current drugs are not effective for treating the primary fibrotic component of CD. Recommended invasive treatments include endoscopic balloon dilation, surgery with resection, or strictureplasty. This meta-analysis compared invasive treatment techniques for CD-related strictures in the pediatric population. Methods: The MEDLINE, EMBASE, and LILACS databases were searched from inception to December 2023. This meta-analysis was performed as per the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines. The outcome was measured by the rate of stenosis recurrence and the adverse events of the techniques. Eligibility criteria were studies that involved the analysis of the recurrence rate of stenosis in pediatric patients with CD requiring surgical intervention after undergoing any of the previously proposed therapies. Additionally, adverse events were analyzed. Results: Three studies comprising a total of 106 endoscopic balloon dilations demonstrated a combined rate of stricture recurrence in patients with CD requiring surgical intervention of 0.171 (95% confidence interval [CI], 0.110-0.255). Three studies comprising a total of 49 surgical resections demonstrated a pooled event rate of 0.100 (95% CI, 0.038-0.240). Finally, 2 studies comprising a total of 38 strictureplasties demonstrated a pooled event rate of 0.347 (95% CI, 0.070-0.789). Concerning adverse events, the most common occurrences were found after surgical resection. Conclusions: Surgical resection is the best option for treating strictures in pediatric CD patients in terms of recurrence rate, compared with endoscopic balloon dilation and strictureplasty. Nevertheless, surgical resection is associated with more adverse events.

7 January 2025 in Endoscopy International Open
Verified
Journal Article
Open access
PDF file
EMEduardo Guimarães Hourneaux de Moura
Eduardo Guimarães Hourneaux de Moura
MCMatheus Ferreira de Carvalho
Matheus Ferreira de Carvalho
VOVictor Lira de Oliveira
Victor Lira de Oliveira
MBMateus Bond Boghossian
Mateus Bond Boghossian
ANAntonio Afonso Miranda Neto
Antonio Afonso Miranda Neto
MFMateus Pereira Funari
Mateus Pereira Funari
Achalasia is a condition related to failure of relaxation of the lower esophageal sphincter (LES). Treatment is based on reducing LES pressure. Although treatment is traditionally surgical, poor candidates for this modality may be treated with peroral endoscopic myotomy (POEM). However, POEM is associated with a relatively high incidence of gastroesophageal reflux disease (GERD). For cases refractory to proton pump inhibitors (PPIs), transoral incisionless fundoplication (TIF) is one of the endoscopic therapies proposed.This was a pilot single-center prospective cohort study including 10 patients with post-POEM GERD refractory to clinical management who underwent endoscopic treatment with the TIF procedure between February and November 2021. We included patients ≥ 18 years old who developed GERD after POEM.Technical success was achieved in all 10 cases treated with TIF. In 6- and 12-month follow-up, seven patients (70%) reduced PPI use. Two patients (20%) had no esophagitis initially, increasing to five (55%) at 6 months and four (44%) at 12 months. Symptom evaluation and GERD-HRQL questionnaire showed a significant score reduction at 6 months and a downward trend at 12 months. Mean Eckardt score showed a decreasing trend, although mean dysphagia score showed a slight tendency to increase in 1 year. The procedure was considered safe, with no adverse events.Use of TIF seems to be a feasible alternative for treating GERD after POEM, improving both clinical and endoscopic parameters and pHmetry in a considerable percentage of cases.

1 December 2025 in Journal of Coloproctology
Verified
Conference Paper
Open access
MCMatheus Ferreira de Carvalho
Matheus Ferreira de Carvalho
MSMiguel Mascarenhas Saraiva
Miguel Mascarenhas Saraiva
JFJoão Ferreira
João Ferreira
MFMarley Ribeiro Feitosa
Marley Ribeiro Feitosa
MVMaisa Vasconcellos
Maisa Vasconcellos
GMGuilherme Macedo
Guilherme Macedo
Autor correspondente: MATHEUS FERREIRA DE CARVALHO - matheusfcarvalho1@gmail.com Introdução Diagnosticar lesões anais como fissuras e lacerações continua sendo um desafio clínico devido à anatomia complexa do canal anal. Embora o ultrassom endoanal (EAUS) ofereça visualização de alta resolução dessas estruturas, a interpretação depende fortemente da experiência do especialista. A inteligência artificial (IA), especialmente os modelos de aprendizado profundo, pode oferecer um caminho para ferramentas diagnósticas mais padronizadas e acessíveis. Objetivo Este estudo bicêntrico explora o desenvolvimento e validação de um modelo baseado em redes neurais convolucionais (CNN) para a classificação automática de fissuras anais e lacerações internas e externas usando imagens de EAUS. Métodos Um estudo retrospectivo bicêntrico analisou 16.747 quadros de 241 exames radiais de EAUS realizados entre abril de 2022 e janeiro de 2024 em dois centros – Manoph (Porto, Portugal; n = 238) e Centro Privado de Cirurgia e Coloproctologia (Buenos Aires, Argentina; n = 3). Os quadros foram categorizados como fissuras (595), lacerações externas (2.174), lacerações internas (1838) e normais (n = 12140), com validação por três especialistas. Os dados foram divididos em 80% para treinamento e 20% para teste. As métricas de desempenho incluíram sensibilidade, especificidade e acurácia. Resultados Para lacerações externas, a CNN atingiu 86,6% de sensibilidade, 97,6% de especificidade e 95,8% de acurácia. Para lacerações internas, atingiu 83,1% de sensibilidade, 97,8% de especificidade e 89,1% de acurácia. Para fissuras anais, obteve sensibilidade de 91,7%, especificidade de 96,7% e acurácia de 96,1%. Para quadros não rotulados, atingiu 97,8% de sensibilidade, 95,5% de especificidade e acurácia de 96,2%. Conclusão Este estudo apresenta o primeiro modelo transatlântico assistido por IA para classificação automática de lesões anais benignas usando EAUS. O algoritmo alcançou excelentes métricas diagnósticas e demonstra o potencial da IA para melhorar a consistência diagnóstica, reduzir a dependência do operador e apoiar a tomada de decisão clínica em proctologia. Apesar de ser limitado por um conjunto de dados relativamente pequeno e geograficamente restrito, este trabalho estabelece as bases para uma integração mais ampla das tecnologias de IA na imagem e no cuidado anorretal. Article published online: 26 May 2026 © 2026. The Author(s). This is an open access article published by Thieme under the terms of the Creative Commons Attribution 4.0 International License, permitting copying and reproduction so long as the original work is given appropriate credit (https://creativecommons.org/licenses/by/4.0/) Thieme Revinter Publicações Ltda. Rua Rego Freitas, 175, loja 1, República, São Paulo, SP, CEP 01220-010, Brazil