Loading...
Publication
Inteligência artificial para a análise automática de imagens de CBCT e TC: aplicações no planeamento de enxertos ósseos – estudo comparativo de um caso clínico
| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| Projeto de pós-graduação_39548 | 8.75 MB | Adobe PDF |
Authors
Advisor(s)
Abstract(s)
A extração dos molares e pré-molares superiores, juntamente com a resposta do processo alveolar que conduz ao edentulismo posterior, resulta numa reabsorção óssea ao nível dos seios maxilares. Este fenómeno, que leva a uma perda óssea quantitativa, tornou-se um grande desafio para os cirurgiões em implantologia dentária. Por conseguinte, foram desenvolvidas e implementadas técnicas regenerativas, como os enxertos ósseos dos seios maxilares, de forma a inserir um material de preenchimento e aumentar a altura óssea do pavimento sinusal, com o objetivo de possibilitar a reabilitação através da colocação de implantes nos pacientes afetados. As imagens CBCT, assim como as imagens de tomografia computorizada (TC), representam um progresso considerável no campo da implantologia, fornecendo informações tridimensionais detalhadas da região maxilofacial. No entanto, a sua interpretação continua a ser delicada e complexa, podendo levar a interpretações erradas. Nesta tese, tínhamos como objetivo principal comparar a decisão e resultado da reabilitação óssea de uma paciente através da IA, com a reabilitação realizada por um profissional sem recurso a IA. Devíamos encontrar um software de inteligência artificial que integrasse técnicas de processamento de imagem e aprendizagem automática, para promover a análise automática e rigorosa de imagens CBCT e TC e propor planos de tratamento automatizados com simulações virtuais no contexto do aumento pré-implantar do pavimento sinusal de modo a facilitar a tomada de decisão ideal por parte do cirurgião-dentista e otimizar o sucesso deste tratamento complexo. Mas, após várias pesquisas, nenhum software de IA até à data foi suficientemente desenvolvido para permitir a geração de planos de tratamento automatizados para cumprir com sucesso a expectativa inicial do nosso objetivo. Embora não tenhamos cumprido este objetivo, por falta de software que o possa realizar neste momento, surgiu um outro objetivo secundário, a avaliação das capacidades e limitações de um software de IA na análise automática das imagens CBCT e TC de um caso clínico no processo de enxerto ósseo sinusal, a fim de comparar a eficácia dos resultados obtidos com a análise realizada pelo cirurgião-dentista e assim relatar o estado da arte da utilização da IA nos procedimentos de enxerto ósseo sinusal. Para tal, estudámos as imagens CBCT, TC e panorâmicas de um caso clínico antes e após um enxerto ósseo sinusal, que submetemos à análise do nosso software de IA Diagnocat, um dos softwares de IA mais desenvolvidos na área dentária para a análise automática de imagens CBCT, TC e panorâmicas. A IA não substitui a experiência humana, mas revela-se um complemento essencial para ajudar o profissional a tomar decisões informadas e a melhorar os resultados do tratamento. Embora a planificação do enxerto ósseo não tenha podido ser gerada de forma totalmente automatizada pelo software, este forneceu uma segmentação automática e precisa das estruturas anatómicas, bem como uma análise das radiografias panorâmicas. Em conclusão, a automação da análise das imagens CBCT e TC através da IA representa um grande progresso no campo da cirurgia dentária, especialmente no contexto dos enxertos ósseos para o aumento pré-implantar do pavimento sinusal. Ao oferecer uma segmentação precisa e automatizada das estruturas anatómicas, a IA aumenta a eficácia e a precisão da análise radiológica. Este sistema oferece uma solução inovadora e segura para otimizar a prática clínica, reduzindo o tempo necessário para a segmentação manual das estruturas anatómicas e, consequentemente, as complicações pré e pós-operatórias. Nos próximos anos, espera-se que a IA continue a desenvolver-se ainda mais e ofereça muito mais possibilidades graças às tecnologias de ponta, com análises e planos de tratamento totalmente automatizados, o que era o nosso objetivo inicial.
The extraction of upper molars and premolars, along with the response of the alveolar process leading to posterior edentulism, results in bone resorption at the maxillary sinuses. This phenomenon, which causes quantitative bone loss, has become a major challenge for dental implant surgeons. Consequently, regenerative techniques, such as maxillary sinus bone grafts, have been developed and implemented to insert a filling material and increase the bone height of the sinus floor, with the goal of enabling rehabilitation through the placement of implants in affected patients. CBCT images, as well as computed tomography (CT) images, represent a considerable advancement in the field of implantology, providing detailed three-dimensional information about the maxillofacial region. However, their interpretation remains delicate and complex, which can lead to misinterpretations. In this thesis, our main goal was to compare the decision-making and outcome of bone rehabilitation in a patient using AI with the rehabilitation performed by a professional without the aid of AI. We sought to find an artificial intelligence software that integrates image processing and machine learning techniques to promote the automatic and rigorous analysis of CBCT and CT images and to propose automated treatment plans with virtual simulations in the context of pre-implant sinus floor augmentation to facilitate the ideal decision-making by the dental surgeon and optimize the success of this complex treatment. However, after extensive research, no AI software to date was found to be sufficiently developed to generate automated treatment plans that successfully met our initial expectations. Since we did not meet this objective due to the lack of suitable software at this time, a secondary objective emerged: to evaluate the capabilities and limitations of an AI software in the automatic analysis of CBCT and CT images of a clinical case in the sinus bone grafting process, in order to compare the efficacy of the results obtained with those of the analysis performed by the dental surgeon, thus documenting the state of the art of AI use in sinus bone grafting procedures. To do so, we studied CBCT, CT, and panoramic images of a clinical case before and after a sinus bone graft, which we submitted to analysis using our AI software, Diagnocat, one of the most advanced AI tools in the dental field for the automatic analysis of CBCT, CT, and panoramic images. AI does not replace human expertise but proves to be an essential complement to help the professional make informed decisions and improve treatment outcomes. Although the planning of the bone graft could not be fully automated by the software, it provided automatic and precise segmentation of anatomical structures as well as an analysis of panoramic radiographs. In conclusion, the automation of CBCT and CT image analysis through AI represents significant progress in dental surgery, especially in the context of bone grafts for pre-implant sinus floor augmentation. By offering precise and automated segmentation of anatomical structures, AI enhances the efficiency and accuracy of radiological analysis. This system provides an innovative and safe solution to optimize clinical practice, reducing the time required for manual segmentation of anatomical structures and, consequently, pre- and post-operative complications. In the coming years, AI is expected to continue evolving and to offer even more possibilities thanks to cutting-edge technologies, with fully automated analysis and treatment planning, which was our initial objective.
The extraction of upper molars and premolars, along with the response of the alveolar process leading to posterior edentulism, results in bone resorption at the maxillary sinuses. This phenomenon, which causes quantitative bone loss, has become a major challenge for dental implant surgeons. Consequently, regenerative techniques, such as maxillary sinus bone grafts, have been developed and implemented to insert a filling material and increase the bone height of the sinus floor, with the goal of enabling rehabilitation through the placement of implants in affected patients. CBCT images, as well as computed tomography (CT) images, represent a considerable advancement in the field of implantology, providing detailed three-dimensional information about the maxillofacial region. However, their interpretation remains delicate and complex, which can lead to misinterpretations. In this thesis, our main goal was to compare the decision-making and outcome of bone rehabilitation in a patient using AI with the rehabilitation performed by a professional without the aid of AI. We sought to find an artificial intelligence software that integrates image processing and machine learning techniques to promote the automatic and rigorous analysis of CBCT and CT images and to propose automated treatment plans with virtual simulations in the context of pre-implant sinus floor augmentation to facilitate the ideal decision-making by the dental surgeon and optimize the success of this complex treatment. However, after extensive research, no AI software to date was found to be sufficiently developed to generate automated treatment plans that successfully met our initial expectations. Since we did not meet this objective due to the lack of suitable software at this time, a secondary objective emerged: to evaluate the capabilities and limitations of an AI software in the automatic analysis of CBCT and CT images of a clinical case in the sinus bone grafting process, in order to compare the efficacy of the results obtained with those of the analysis performed by the dental surgeon, thus documenting the state of the art of AI use in sinus bone grafting procedures. To do so, we studied CBCT, CT, and panoramic images of a clinical case before and after a sinus bone graft, which we submitted to analysis using our AI software, Diagnocat, one of the most advanced AI tools in the dental field for the automatic analysis of CBCT, CT, and panoramic images. AI does not replace human expertise but proves to be an essential complement to help the professional make informed decisions and improve treatment outcomes. Although the planning of the bone graft could not be fully automated by the software, it provided automatic and precise segmentation of anatomical structures as well as an analysis of panoramic radiographs. In conclusion, the automation of CBCT and CT image analysis through AI represents significant progress in dental surgery, especially in the context of bone grafts for pre-implant sinus floor augmentation. By offering precise and automated segmentation of anatomical structures, AI enhances the efficiency and accuracy of radiological analysis. This system provides an innovative and safe solution to optimize clinical practice, reducing the time required for manual segmentation of anatomical structures and, consequently, pre- and post-operative complications. In the coming years, AI is expected to continue evolving and to offer even more possibilities thanks to cutting-edge technologies, with fully automated analysis and treatment planning, which was our initial objective.
Description
Keywords
Inteligência artificial em medicina dentária CBCT Seios maxilares Enxertos ósseos Implantes Diagnocat Artificial intelligence in dentistry Maxillary sinuses Bone grafts Implants