Academic literature on the topic 'Confocal laser endomicroscopy'

Aim. To evaluate information content of confocal laser endomicroscopy with targeted biopsy in verifying etiology of extrahepatic bile duct strictures.Material and methods. There were 28 patients with extrahepatic bile duct strictures who underwent retrograde intervention with confocal laser endomicroscopy and targeted biopsy. Data of confocal laser endomicroscopy and biopsy were compared with final postoperative and histological diagnosis. Follow-up within 1–4 years after endoscopic treatment was also considered.Results. Diagnostic sensitivity, specificity and overall accuracy of the method in differential diagnosis of common bile duct strictures were 91.7%, 93.7% and 92.8%, respectively. Complication (acute edematous pancreatitis) occurred in 1 (3.6%) case.Conclusion. Confocal laser endomicroscopy is new effective method for in vivo microscopic assessment of mucous membrane. Despite technical complexity, the method is not associated with advanced morbidity and, accordingly, has no additional contraindications in comparison with ERCP.

Probe-based confocal laser endomicroscopy (pCLE) is an emerging imaging tool that allows real-time in situ morphological imaging at cellular and subcellular resolution. Its ability to image morphological features of epithelial surfaces of the gastrointestinal tract, biliary tree and respiratory tree rendered differentiation of macroscopically inconspicuous neoplastic and non-neoplastic tissues possible in real-time. However, its role outwith the endoluminal environment for surgical applications has been comparatively sparsely investigated and little reported on its ability to characterise morphological features beyond endoluminal applications. This thesis aims to systematically evaluate the potential pCLE has in visualization of soft tissue morphology in applications pertaining to breast conserving surgery (BCS), parathyroid surgery and thyroid surgery; whereby morphological information regarding cavity wall margin status, tissue-specific entity and viability status of preserved parathyroid glands (PG), respectively, could potentially guide decision-making intraoperatively. The perceptions that pCLE imaging is confined to endoluminal mucosal surfaces, the inability of pCLE to perform image acquisition through sterile transparent sheaths and the inability of surgeons to interpret pCLE images were interrogated using three small feasibility studies. Firstly, in a study carried out on a live, anaesthetised, porcine model, pCLE image acquisition of morphological architecture of soft tissues of the neck e.g. thyroid, lymph nodes, adipose, skeletal and smooth muscles, were shown to be feasible in an intraoperative field and the presence dried blood on the tissue surface did not impede the consistency of morphological architecture visualization. Secondly, we demonstrated that utilization of a sterile transparent sheath did not impede pCLE image acquisition and that the quality of images obtained was comparable to that of without the sheath. Thirdly, we have shown that surgeons with little or no histopathology background were able to acquire the relevant pattern recognition skills to interpret pCLE images following a training session utilizing a validated pCLE morphological classification from colorectal lesions. Building upon these discoveries, we elucidated the potential of pCLE to image neoplastic and non-neoplastic breast morphology with the envisaged application of identifying residual cancerous foci intraoperatively, thereby guiding operative decision making based on real-time breast cavity scanning during BCS. Preliminary ex vivo analyses from 71 freshly excised, acriflavine-stained neoplastic and non-neoplastic tissues samples from 50 breast cancer patients show excellent correlation with histopathology findings. In particular, the glandular structures, adipocytes and collagen fibres of non-neoplastic breast tissues were readily visualised on pCLE images. These were distinguishable from the markedly haphazard and hypercellular architecture exhibited by invasive and non-invasive carcinoma. We developed a classification based on description of pCLE morphology unique to neoplastic and non-neoplastic breast morphology and validated this with 17 histopathologists and surgeons through a systematic pattern recognition training session based on this classification where they were subsequently subjected to objective assessment of 50 pCLE images while blinded to histopathology results. The overall mean accuracy of pCLE image interpretation for histopathologists and surgeons were 94% and 92%, respectively. The overall inter-observer agreement was ‘almost perfect’ (κ=0.81) for the former and ‘substantial’ (κ=0.77), for the latter. We explored the role of intravenous fluorescein sodium (FS) in a prospective, cross-sectional, observational study of 10 patients undergoing BCS where they received between 1.5ml to 3.5ml of intravenous bolus of 10% fluorescein sodium (FS) intraoperatively. Ex vivo analyses of FS-stained breast samples showed that dense fibrous tissue response evoked by infiltrating tumor cells were readily visualised as fluorescent regions with haphazardly arranged, amorphous-looking collagen fibres. However, the lack of nuclei visualization rendered differentiation of neoplastic from non-neoplastic tissues impossible. Nevertheless, the uniformity that FS staining imparts to all tissue layers facilitated creation of longer and meaningful pCLE mosaics. These findings could have important implications where tissue deformation could result in AH-stained layers intermittently fail to coincide with the optical slice imaged at the respective depth. The promising findings of AH-stained breast tissues were found to be potentially relevant in parathyroid surgery. Similar analyses on freshly excised AH-stained parathyroid specimens from 35 patients undergoing parathyroidectomy for primary and secondary hyperparathyroidism showed nest-like arrangements of parenchymal cells, fibrovascular septum and microfollicles of diseased PGs were readily identifiable on CE images and these were consistent with histopathological findings. Following pattern recognition training based on an in-house developed classification system, these were distinguishable from epithelial-lined thyroid follicles and polygonal-shaped adipocytes with mean accuracies of 94% and 93% for histopathologists and surgeons, respectively, and high overall inter-observer agreement, κ=0.82. Where intraoperative identification of diseased PGs presents a challenge especially in multi-glandular disease and re-operative surgery, pCLE could potentially facilitate its recognition. Finally, the role for pCLE imaging of PG vasculature was explored by means of an intraoperative clinical study utilising a sterile-transparent draped pCLE probe on 20 patients undergoing thyroid and parathyroid surgery. Utilising intravenous FS, branched-vessels including capillary networks were readily visualised. Vascular flow on viable glands was depicted by unidirectional, high velocity thrusts of dark-coloured erythrocytes within hyperfluorescent vessels or diffusely in the parenchyma whereas these were absent on non-viable (post-excision) glands. Further analysis on preserved PGs showed that absence of blood flow was found in patients who had sub-optimal post-operative parathyroid function. Given that visual assessment of tissue discolouration is not a reliable method of determining parathyroid gland viability during thyroidectomy, information regarding viability of preserved PGs decisions could potentially aid decisions pertinent to autotransplantation remains challenging. This thesis significantly expands upon the potential intraoperative applications of pCLE. Evidently, these findings are preliminary and warrant further evaluation in well-powered clinical trials. However a systematic approach to investigate the optimal trade-offs between the optical resolution requirements of tissue morphology visualization and deployability of pCLE probe holds the key to successful clinical translation. In particular, evaluation of a robust mechatronically enhanced platform equipped with the flexibility to cater for tissue surface deformation and precision mechanisms that generates accurate spatio-temporal localisation in real-time to aid intraoperative decision making constitutes the next stage of research priorities.

abstract: Intraoperative diagnosis in neurosurgery has traditionally relied on frozen and formalin-fixed, paraffin-embedded section analysis of biopsied tissue samples. Although this technique is considered to be the “gold standard” for establishing a histopathologic diagnosis, it entails a number of significant limitations such as invasiveness and the time required for processing and interpreting the tissue. Rapid intraoperative diagnosis has become possible with a handheld confocal laser endomicroscopy (CLE) system. Combined with appropriate fluorescent stains or labels, CLE provides an imaging technique for real-time intraoperative visualization of histopathologic features of the suspected tumor and healthy tissues. This thesis scrutinizes CLE technology for its ability to provide real-time intraoperative in vivo and ex vivo visualization of histopathological features of the normal and tumor brain tissues. First, the optimal settings for CLE imaging are studied in an animal model along with a generational comparison of CLE performance. Second, the ability of CLE to discriminate uninjured normal brain, injured normal brain and tumor tissues is demonstrated. Third, CLE was used to investigate cerebral microvasculature and blood flow in normal and pathological conditions. Fourth, the feasibility of CLE for providing optical biopsies of brain tumors was established during the fluorescence-guided neurosurgical procedures. This study established the optimal workflow and confirmed the high specificity of the CLE optical biopsies. Fifth, the feasibility of CLE was established for endoscopic endonasal approaches and interrogation of pituitary tumor tissue. Finally, improved and prolonged near wide-field fluorescent visualization of brain tumor margins was demonstrated with a scanning fiber endoscopy and 5-aminolevulinic acid. These studies suggested a novel paradigm for neurosurgery-pathology workflow when the noninvasive intraoperative optical biopsies are used to interrogate the tissue and augment intraoperative decision making. Such optical biopsies could shorten the time for obtaining preliminary information on the histological composition of the tissue of interest and may lead to improved diagnostics and tumor resection. This work establishes a basis for future in vivo optical biopsy use in neurosurgery and planning of patient-related outcome studies. Future studies would lead to refinement and development of new confocal scanning technologies making noninvasive optical biopsy faster, convenient and more accurate.
Dissertation/Thesis
Doctoral Dissertation Neuroscience 2020

abstract: Rapid intraoperative diagnosis of brain tumors is of great importance for planning treatment and guiding the surgeon about the extent of resection. Currently, the standard for the preliminary intraoperative tissue analysis is frozen section biopsy that has major limitations such as tissue freezing and cutting artifacts, sampling errors, lack of immediate interaction between the pathologist and the surgeon, and time consuming. Handheld, portable confocal laser endomicroscopy (CLE) is being explored in neurosurgery for its ability to image histopathological features of tissue at cellular resolution in real time during brain tumor surgery. Over the course of examination of the surgical tumor resection, hundreds to thousands of images may be collected. The high number of images requires significant time and storage load for subsequent reviewing, which motivated several research groups to employ deep convolutional neural networks (DCNNs) to improve its utility during surgery. DCNNs have proven to be useful in natural and medical image analysis tasks such as classification, object detection, and image segmentation. This thesis proposes using DCNNs for analyzing CLE images of brain tumors. Particularly, it explores the practicality of DCNNs in three main tasks. First, off-the shelf DCNNs were used to classify images into diagnostic and non-diagnostic. Further experiments showed that both ensemble modeling and transfer learning improved the classifier’s accuracy in evaluating the diagnostic quality of new images at test stage. Second, a weakly-supervised learning pipeline was developed for localizing key features of diagnostic CLE images from gliomas. Third, image style transfer was used to improve the diagnostic quality of CLE images from glioma tumors by transforming the histology patterns in CLE images of fluorescein sodium-stained tissue into the ones in conventional hematoxylin and eosin-stained tissue slides. These studies suggest that DCNNs are opted for analysis of CLE images. They may assist surgeons in sorting out the non-diagnostic images, highlighting the key regions and enhancing their appearance through pattern transformation in real time. With recent advances in deep learning such as generative adversarial networks and semi-supervised learning, new research directions need to be followed to discover more promises of DCNNs in CLE image analysis.
Dissertation/Thesis
Doctoral Dissertation Neuroscience 2019

Tese de mestrado integrado, Engenharia Biomédica e Biofísica (Biofísica Médica e Fisiologia de Sistemas), Universidade de Lisboa, Faculdade de Ciências, 2018
Doenças Intersticiais Pulmonares (DIP) é um termo que inclui mais de 200 doenças que afectam o parênquima pulmonar, partilhando manifestações clínicas, radiográficas e patológicas semelhantes. Este conjunto de doenças é bastante heterogéneo, apresentando cada tipo de DIP em diferente grau os elementos de inflamação e fibrose: enquanto a inflamação é reflectida pelo aumento de células inflamatórias e presença de nódulos ou edema, a fibrose reflecte-se pelas fibras adicionais de colagénio e elastina. Identificar o tipo de DIP de um doente é um processo difícil, sendo a Discussão Multidisciplinar o actual método de diagnóstico "gold standard": vários médicos especialistas compõem uma equipa multidisciplinar que vai ter em conta os dados clínicos, radiológicos e patológicos disponíveis para chegar a uma conclusão. Estes dados incluem imagens de tomografia computorizada de alta resolução (TCAR), a descrição da lavagem broncoalveolar e, quando possível, dados de biópsias. Apesar do esforço e competência da equipa multidisciplinar, 10% dos pacientes são categorizados como inclassificáveis devido a dados inadequados ou discrepância entre os dados existentes. A maior causa para DIP inclassificáveis é a ausência de dados histopatológicos associada aos riscos das biópsias cirúrgicas. É muito importante determinar a DIP específica de um doente, dadas as suas implicações no tratamento e gestão do mesmo. É particularmente crítica a distinção entre doentes com Fibrose Pulmonar Idiopática (FPI) e doentes sem FPI, dado que há terapias anti-fibróticas – como o Pirfenidone – indicadas para FPI que são extremamente dispendiosas, exigindo certeza no diagnóstico antes de serem prescritas. Além disso, o tratamento com agentes imunossupressores pode funcionar com o grupo dos não-FPI mas aumenta a morte e hospitalizações nos doentes com FPI. A discussão multidisciplinar pode beneficiar da informação adicional oferecida pelo Confocal Laser Endomicroscopy (CLE), uma técnica de imagiologia que torna possível visualizar os alvéolos pulmonares com resolução microscópica de forma minimamente invasiva, através de uma broncoscopia. O laser do CLE tem um comprimento de onda de 488 nm que permite observar a autofluorescência das fibras de elastina. Há evidências de que a quantidade de fibras de elastina é aumentada e a arquitectura destas fibras é alterada na presença de fibrose pulmonar, a qual está associada a algumas doenças intersticiais pulmonares incluindo a fibrose pulmonar idiopática. Até à data, os vídeos de Confocal Laser Endomicroscopy são, na maioria dos casos, analisados apenas visualmente, e pouca informação objectiva e consistente foi conseguida destes vídeos em doentes de DIP. No entanto, é possível obter informação mais relevante dos mesmos, convertendo-os em frames, pré-processando as imagens e extraindo atributos numéricos. Neste projecto, foram obtidas imagens dos alvéolos pulmonares de doentes de DIP através de CLE. O principal objectivo do projecto é melhorar a técnica de CLE e aumentar a sua usabilidade para que no futuro possa contribuir para facilitar a estratificação de doentes com DIP e eventualmente reduzir o número de biópsias pulmonares nestes doentes. Como mencionado, o instrumento de Confocal Laser Endomicroscopy emite uma luz laser azul de 488nm, a qual é reflectida no tecido e reorientada para o sistema de detecção pela mesma lente, passando por um pequeno orifício (pinhole). Isto permite que a luz focada seja recolhida e que feixes provenientes de planos fora de foco sejam excluídos, originando uma resolução microscópica que permite imagens ao nível celular. Quando o CLE é aplicado a imagem pulmonar, é possível observar as paredes alveolares pela autofluorescência natural presente nas fibras de elastina. No estudo clínico subjacente a este estudo, o protocolo de CLE foi aplicado a 20 pacientes, embora 8 tenham sido posteriormente excluídos da análise. Os vídeos de CLE obtidos sofreram duas selecções: uma com base na região onde uma biópsia (usada como referência) foi tirada e outra com base na qualidade técnica das imagens. Depois, os dados foram pré-processados: geraram-se imagens mosaico com um campo de visão alargado e, paralelamente converteram-se as sequências de vídeo em frames. A qualidade da imagem foi melhorada, filtrando o ruído electrónico para que posteriormente pudesse ser aplicada a análise de imagem. Esta análise extraiu valores numéricos que reflectem o estado do espaço alveolar, nomeadamente, variáveis de textura e medições relacionadas com as fibras de elastina. As imagens de CLE obtidas mostraram-se muito interessantes. A resolução é superior à tomografia computorizada de alta resolução e a tridimensionalidade acrescenta informação às biópsias. O facto de permitir feedback em tempo real e observar ao vivo os movimentos naturais da respiração contribui para a análise do estado do doente. A análise de textura feita às imagens serviu-se de um algoritmo de extracção de variáveis de Haralick a partir de uma Gray-Level Co-occurence Matrix (GLCM). Foram extraídas as variáveis de textura Momento Angular Secundário (Energia), Entropia, Momento de Diferença Inversa, Contraste, Variação e Correlação. O algoritmo de Ridge Detection (detecção de linhas) identificou a maior parte das fibras de elastina detectáveis por um observador humano e mediu o Número de Fibras, o seu Comprimento e Largura e o Número de Junções entre fibras, permitindo também calcular a Soma dos Comprimentos de todas as fibras. Estes algoritmos devolveram valores consistentes num processo mais eficiente comparado com um observador humano, conseguindo avaliar em poucos segundos múltiplas variáveis para todo o conjunto de dados. As medições relacionadas com as fibras de elastina pretendiam ajudar a identificar os doentes fibróticos. Era esperado que as fibras dos doentes fibróticos fossem mais largas, mas isso não se observou. Também se previa que este grupo de doentes apresentasse maior número de fibras e junções, mas não houve uma diferença significativa entre grupos. No entanto, quando o grupo fibrótico foi segregado, o número de fibras e junções parece separar a fibrose moderada da fibrose severa. Este resultado é interessante na medida em que sugere que a monitorização do número de fibras/junções com CLE pode potencialmente ser usado como medida de eficácia de medicação anti-fibrótica. Em relação às variáveis de textura, esperava-se que os doentes fibróticos apresentassem valores mais elevados de Entropia, Contraste e Variância e valores inferiores de Momento de Diferença Inversa, dado que o seu tecido pulmonar deveria corresponder a imagens mais complexas e heterogéneas com mais arestas presentes. No entanto, ainda não foi possível estabelecer diferenças significativas entre grupos. Apesar dos resultados com o conjunto de dados usado não ter demonstrado correlações fortes entre as conclusões do CLE e da TCAR/histopatologia, os valores das variáveis em si já contribuem para o estudo das DIP, nomeadamente da sua fisiologia. De facto, a amostra de doentes deste estudo era reduzida, mas com uma amostra maior, espera-se que algumas das varáveis se correlacionem com outras técnicas usadas no diagnóstico e permitam segregar os pacientes em grupos e eventualmente aplicar classificação de dados. Neste momento, é possível especular que algumas variáveis seriam melhores candidatas para um classificador, nomeadamente os Números de Fibras e Junções, a Soma dos Comprimentos das fibras e as variáveis de Haralick Entropia e Energia. O projecto apresentado nesta dissertação foi desenvolvido através de um estágio de 6 meses no departamento de Pneumologia no Academic Medical Center em Amsterdão, Países Baixos. No Academic Medical Center (AMC), fui acompanhada pelos estudantes de doutoramento Lizzy Wijmans - médica - e Paul Brinkman - engenheiro biomédico - e supervisionada pelo Dr. Jouke Annema, MD, PhD, Professor de endoscopia pulmonar. Este grupo de investigação do AMC está focado em técnicas inovadoras de imagiologia do sistema pulmonar e teve a oportunidade de reunir com a empresa MKT –que produz a tecnologia de Confocal Laser Endomicroscopy –, o que enriqueceu a discussão aqui apresentada. Do Departamento de Física da Faculdade de Ciências da Universidade de Lisboa, fui orientada pelo Prof. Nuno Matela.
Interstitial Lung Diseases (ILD) is a heterogeneous group of more than 200 diseases which affect the lung parenchyma. To identify the type of ILD a patient suffers from is a difficult process, and 10% of the patients are categorized as unclassifiable, mostly due to the absence of histopathological data associated with the risks of lung biopsies. The patient specific diagnosis is important because of its implications to the patient treatment and management, being particularly relevant to identify lung fibrosis. The Confocal Laser Endomicroscopy (CLE) can add information to this process. CLE allows to image the lung tissue with a micrometer resolution in a minimally invasive way, through a bronchoscopy. The elastin fibers from the lung alveoli are visible with this technique due to their autofluorescence. Since there is evidence that the amount of elastin fibers increases, and their architecture is altered in lung fibrosis, CLE should be used to extract values reflecting this condition. Thus, the main goal of this project was to improve the CLE technique and increase its usability, by extracting numerical values from the images which would reflect the state of the alveolar space, particularly the elastin fibers. The ILD patients recruited for the study had their lung alveoli imaged with CLE. The CLE movies were selected, pre-processed – were converted into frames, had their image quality enhanced and some mosaics were obtained – and then analyzed. The ridge detection algorithm detected most fibers recognized by a human observer. It allowed the measurement of the Number of Detected Fibers, their Length and Width, the Number of Junctions between fibers and to calculate the Sum from all Fibers’ Lengths. The Gray-Level Co-occurrence Matrix allowed the extraction of the Haralick texture features: Angular Second Moment (Energy), Entropy, Inverse Difference Moment, Contrast, Variance and Correlation. These algorithms produced consistent and unbiased numerical features, in an efficient process which can analyze the entire data set in a few seconds. Regarding the fiber related measurements, it was expected for the fibrotic patients to have wider fibers and a higher number of fibers and junctions. In terms of texture variables, it was expected from the fibrotic patients to present higher values of Entropy, Contrast and Variance, and lower values of Inverse Difference Moment, given their lung tissue should correspond to more complex and heterogeneous images with more ridges present. Due to the small sample size, it was still not possible to stratify patients with this data set. Nevertheless, the measurements presented here already contribute to the study of ILD, helping to understand the disease physiology. It is hoped that in the future, these measurements will aid the diagnosis process specially in those cases when patients cannot undergo a surgical biopsy. Additionally, CLE could potentially be used as an anti-fibrotic medication efficiency measurement tool.