RESUMO
Digital whole slides images contain an enormous amount of information providing a strong motivation for the development of automated image analysis tools. Particularly deep neural networks show high potential with respect to various tasks in the field of digital pathology. However, a limitation is given by the fact that typical deep learning algorithms require (manual) annotations in addition to the large amounts of image data, to enable effective training. Multiple instance learning exhibits a powerful tool for training deep neural networks in a scenario without fully annotated data. These methods are particularly effective in the domain of digital pathology, due to the fact that labels for whole slide images are often captured routinely, whereas labels for patches, regions, or pixels are not. This potential resulted in a considerable number of publications, with the vast majority published in the last four years. Besides the availability of digitized data and a high motivation from the medical perspective, the availability of powerful graphics processing units exhibits an accelerator in this field. In this paper, we provide an overview of widely and effectively used concepts of (deep) multiple instance learning approaches and recent advancements. We also critically discuss remaining challenges as well as future potential.
Assuntos
Algoritmos , Redes Neurais de Computação , Processamento de Imagem Assistida por Computador/métodosRESUMO
Early-life immune development is critical to long-term host health. However, the mechanisms that determine the pace of postnatal immune maturation are not fully resolved. Here, we analyzed mononuclear phagocytes (MNPs) in small intestinal Peyer's patches (PPs), the primary inductive site of intestinal immunity. Conventional type 1 and 2 dendritic cells (cDC1 and cDC2) and RORgt+ antigen-presenting cells (RORgt+ APC) exhibited significant age-dependent changes in subset composition, tissue distribution, and reduced cell maturation, subsequently resulting in a lack in CD4+ T cell priming during the postnatal period. Microbial cues contributed but could not fully explain the discrepancies in MNP maturation. Type I interferon (IFN) accelerated MNP maturation but IFN signaling did not represent the physiological stimulus. Instead, follicle-associated epithelium (FAE) M cell differentiation was required and sufficient to drive postweaning PP MNP maturation. Together, our results highlight the role of FAE M cell differentiation and MNP maturation in postnatal immune development.
Assuntos
Células M , Nódulos Linfáticos Agregados , Intestinos , Intestino Delgado , Diferenciação Celular , Mucosa IntestinalRESUMO
Whole slide images contain a magnitude of quantitative information that may not be fully explored in qualitative visual assessments. We propose: (1) a novel pipeline for extracting a comprehensive set of visual features, which are detectable by a pathologist, as well as sub-visual features, which are not discernible by human experts and (2) perform detailed analyses on renal images from mice with experimental unilateral ureteral obstruction. An important criterion for these features is that they are easy to interpret, as opposed to features obtained from neural networks. We extract and compare features from pathological and healthy control kidneys to learn how the compartments (glomerulus, Bowman's capsule, tubule, interstitium, artery, and arterial lumen) are affected by the pathology. We define feature selection methods to extract the most informative and discriminative features. We perform statistical analyses to understand the relation of the extracted features, both individually, and in combinations, with tissue morphology and pathology. Particularly for the presented case-study, we highlight features that are affected in each compartment. With this, prior biological knowledge, such as the increase in interstitial nuclei, is confirmed and presented in a quantitative way, alongside with novel findings, like color and intensity changes in glomeruli and Bowman's capsule. The proposed approach is therefore an important step towards quantitative, reproducible, and rater-independent analysis in histopathology.
RESUMO
BACKGROUND: The fast acquisition process of frozen sections allows surgeons to wait for histological findings during the interventions to base intrasurgical decisions on the outcome of the histology. Compared with paraffin sections, however, the quality of frozen sections is often strongly reduced, leading to a lower diagnostic accuracy. Deep neural networks are capable of modifying specific characteristics of digital histological images. Particularly, generative adversarial networks proved to be effective tools to learn about translation between two modalities, based on two unconnected data sets only. The positive effects of such deep learning-based image optimization on computer-aided diagnosis have already been shown. However, since fully automated diagnosis is controversial, the application of enhanced images for visual clinical assessment is currently probably of even higher relevance. METHODS: Three different deep learning-based generative adversarial networks were investigated. The methods were used to translate frozen sections into virtual paraffin sections. Overall, 40 frozen sections were processed. For training, 40 further paraffin sections were available. We investigated how pathologists assess the quality of the different image translation approaches and whether experts are able to distinguish between virtual and real digital pathology. RESULTS: Pathologists' detection accuracy of virtual paraffin sections (from pairs consisting of a frozen and a paraffin section) was between 0.62 and 0.97. Overall, in 59% of images, the virtual section was assessed as more appropriate for a diagnosis. In 53% of images, the deep learning approach was preferred to conventional stain normalization (SN). CONCLUSION: Overall, expert assessment indicated slightly improved visual properties of converted images and a high similarity to real paraffin sections. The observed high variability showed clear differences in personal preferences.
RESUMO
PURPOSE: The psoas major muscle (PMM) volume serves as an opportunistic imaging marker in cross-sectional imaging datasets for various clinical applications. Since manual segmentation is time consuming, two different automated segmentation methods, a generative adversarial network architecture (GAN) and a multi-atlas segmentation (MAS), as well as a combined approach of both, were investigated in terms of accuracy of automated volumetrics in given CT datasets. MATERIALS AND METHODS: The bilateral PMM was manually segmented by a radiologist in 34 abdominal CT scans, resulting in 68 single 3D muscle segmentations as training data. Three different methods were tested for their ability to generate automated image segmentations: a GAN- and MAS-based approach and a combined approach of both methods (COM). Bilateral PMM volume (PMMV) was calculated in cm3 by each algorithm for every CT. Results were compared to the corresponding ground truth using the Dice similarity coefficient (DSC), Spearman's correlation coefficient and Wilcoxon signed-rank test. RESULTS: Mean PMMV was 239 ± 7.0 cm3 and 308 ± 9.6 cm3, 306 ± 9.5 cm3 and 243 ± 7.3 cm3 for the CNN, MAS and COM, respectively. Compared to the ground truth the CNN and MAS overestimated the PMMV significantly (+ 28.9% and + 28.0%, p < 0.001), while results of the COM were quite accurate (+ 0.7%, p = 0.33). Spearman's correlation coefficients were 0.38, 0.62 and 0.73, and the DSCs were 0.75 [95%CI: 0.56-0.88], 0.73 [95%CI: 0.54-0.85] and 0.82 [95%CI: 0.65-0.90] for the CNN, MAS and COM, respectively. CONCLUSION: The combined approach was able to efficiently exploit the advantages of both methods (GAN and MAS), resulting in a significantly higher accuracy in PMMV predictions compared to the isolated implementations of both methods. Even with the relatively small set of training data, the segmentation accuracy of this hybrid approach was relatively close to that of the radiologist.
Assuntos
Processamento de Imagem Assistida por Computador , Músculos Psoas , Algoritmos , Humanos , Aprendizado de Máquina , Músculos Psoas/diagnóstico por imagem , Tomografia Computadorizada por Raios XRESUMO
Computed tomography (CT) scans and magnetic resonance imaging (MRI) of spines are state-of-the-art for the evaluation of spinal cord lesions. This paper analyses micro-CT scans of rat spinal cords with the aim of generating lesion progression through the aggregation of anomaly-based scores. Since reliable labelling in spinal cords is only reasonable for the healthy class in the form of untreated spines, semi-supervised deviation-based anomaly detection algorithms are identified as powerful approaches. The main contribution of this paper is a large evaluation of different autoencoders and variational autoencoders for aggregated lesion quantification and a resulting spinal cord lesion quantification method that generates highly correlating quantifications. The conducted experiments showed that several models were able to generate 3D lesion quantifications of the data. These quantifications correlated with the weakly labelled true data with one model, reaching an average correlation of 0.83. We also introduced an area-based model, which correlated with a mean of 0.84. The possibility of the complementary use of the autoencoder-based method and the area feature were also discussed. Additionally to improving medical diagnostics, we anticipate features built on these quantifications to be useful for further applications like clustering into different lesions.
Assuntos
Algoritmos , Imageamento por Ressonância Magnética , Animais , Análise por Conglomerados , Ratos , Medula Espinal/diagnóstico por imagemRESUMO
With emerging treatment approaches, it is crucial to correctly diagnose and monitor hereditary and acquired polyneuropathies. This study aimed to assess the validity and accuracy of magnet resonance imaging (MRI)-based muscle volumetry.Using semi-automatic segmentations of upper- and lower leg muscles based on whole-body MRI and axial T1-weighted turbo spin-echo sequences, we compared and correlated muscle volumes, and clinical and neurophysiological parameters in demyelinating Charcot-Marie-Tooth disease (CMT) (n = 13), chronic inflammatory demyelinating polyneuropathy (CIDP) (n = 27), and other neuropathy (n = 17) patients.The muscle volumes of lower legs correlated with foot dorsiflexion strength (p < 0.0001), CMT Neuropathy Score 2 (p < 0.0001), early gait disorders (p = 0.0486), and in CIDP patients with tibial nerve conduction velocities (p = 0.0092). Lower (p = 0.0218) and upper (p = 0.0342) leg muscles were significantly larger in CIDP compared to CMT patients. At one-year follow-up (n = 15), leg muscle volumes showed no significant decrease.MRI muscle volumetry is a promising method to differentiate and characterize neuropathies in clinical practice.
RESUMO
Deep neural networks recently showed high performance and gained popularity in the field of radiology. However, the fact that large amounts of labeled data are required for training these architectures inhibits practical applications. We take advantage of an unpaired image-to-image translation approach in combination with a novel domain specific loss formulation to create an "easier-to-segment" intermediate image representation without requiring any label data. The requirement here is that the task can be translated from a hard to a related but simplified task for which unlabeled data are available. In the experimental evaluation, we investigate fully automated approaches for segmentation of pathological muscle tissue in T1-weighted magnetic resonance (MR) images of human thighs. The results show clearly improved performance in case of supervised segmentation techniques. Even more impressively, we obtain similar results with a basic completely unsupervised segmentation approach.
RESUMO
[This corrects the article DOI: 10.1016/j.patter.2020.100089.].
RESUMO
Image analysis in the field of digital pathology has recently gained increased popularity. The use of high-quality whole-slide scanners enables the fast acquisition of large amounts of image data, showing extensive context and microscopic detail at the same time. Simultaneously, novel machine-learning algorithms have boosted the performance of image analysis approaches. In this paper, we focus on a particularly powerful class of architectures, the so-called generative adversarial networks (GANs) applied to histological image data. Besides improving performance, GANs also enable previously intractable application scenarios in this field. However, GANs could exhibit a potential for introducing bias. Hereby, we summarize the recent state-of-the-art developments in a generalizing notation, present the main applications of GANs, and give an outlook of some chosen promising approaches and their possible future applications. In addition, we identify currently unavailable methods with potential for future applications.
RESUMO
BACKGROUND: Muscle MRI is of increasing importance for neuromuscular patients to detect changes in muscle volume, fat-infiltration, and edema. We developed a method for semi-automated segmentation of muscle MRI datasets. METHODS: An active contour-evolution algorithm implemented within the ITK-SNAP software was used to segment T1-weighted MRI, and to quantify muscle volumes of neuromuscular patients (n = 65). RESULTS: Semi-automated compared with manual segmentation was shown to be accurate and time-efficient. Muscle volumes and ratios of thigh/lower leg volume were lower in myopathy patients than in controls (P < .0001; P < .05). We found a decrease of lower leg muscle volume in neuropathy patients compared with controls (P < .01), which correlated with clinical parameters. In myopathy patients, muscle volume showed a positive correlation with muscle strength (rleft = 0.79, pleft < .0001). Muscle volumes were independent of body mass index and age. CONCLUSIONS: Our method allows for exact and time-efficient quantification of muscle volumes with possible use as a biomarker in neuromuscular patients.
Assuntos
Imageamento por Ressonância Magnética , Músculo Esquelético/diagnóstico por imagem , Doenças Musculares/diagnóstico por imagem , Doenças do Sistema Nervoso Periférico/diagnóstico por imagem , Software , Adulto , Idoso , Automação , Estudos de Casos e Controles , Doença de Charcot-Marie-Tooth/diagnóstico por imagem , Doença de Charcot-Marie-Tooth/patologia , Neuropatias Diabéticas/diagnóstico por imagem , Neuropatias Diabéticas/patologia , Feminino , Humanos , Processamento de Imagem Assistida por Computador , Masculino , Pessoa de Meia-Idade , Músculo Esquelético/patologia , Doenças Musculares/patologia , Distrofia Muscular do Cíngulo dos Membros/diagnóstico por imagem , Distrofia Muscular do Cíngulo dos Membros/patologia , Miosite/diagnóstico por imagem , Miosite/patologia , Miosite de Corpos de Inclusão/diagnóstico por imagem , Miosite de Corpos de Inclusão/patologia , Tamanho do Órgão , Doenças do Sistema Nervoso Periférico/patologia , Polimiosite/diagnóstico por imagem , Polimiosite/patologia , Polirradiculoneuropatia Desmielinizante Inflamatória Crônica/diagnóstico por imagem , Polirradiculoneuropatia Desmielinizante Inflamatória Crônica/patologia , Estudos RetrospectivosRESUMO
BACKGROUND: It was shown in previous studies that high definition endoscopy, high magnification endoscopy and image enhancement technologies, such as chromoendoscopy and digital chromoendoscopy [narrow-band imaging (NBI), i-Scan] facilitate the detection and classification of colonic polyps during endoscopic sessions. However, there are no comprehensive studies so far that analyze which endoscopic imaging modalities facilitate the automated classification of colonic polyps. In this work, we investigate the impact of endoscopic imaging modalities on the results of computer-assisted diagnosis systems for colonic polyp staging. AIM: To assess which endoscopic imaging modalities are best suited for the computer-assisted staging of colonic polyps. METHODS: In our experiments, we apply twelve state-of-the-art feature extraction methods for the classification of colonic polyps to five endoscopic image databases of colonic lesions. For this purpose, we employ a specifically designed experimental setup to avoid biases in the outcomes caused by differing numbers of images per image database. The image databases were obtained using different imaging modalities. Two databases were obtained by high-definition endoscopy in combination with i-Scan technology (one with chromoendoscopy and one without chromoendoscopy). Three databases were obtained by high-magnification endoscopy (two databases using narrow band imaging and one using chromoendoscopy). The lesions are categorized into non-neoplastic and neoplastic according to the histological diagnosis. RESULTS: Generally, it is feature-dependent which imaging modalities achieve high results and which do not. For the high-definition image databases, we achieved overall classification rates of up to 79.2% with chromoendoscopy and 88.9% without chromoendoscopy. In the case of the database obtained by high-magnification chromoendoscopy, the classification rates were up to 81.4%. For the combination of high-magnification endoscopy with NBI, results of up to 97.4% for one database and up to 84% for the other were achieved. Non-neoplastic lesions were classified more accurately in general than non-neoplastic lesions. It was shown that the image recording conditions highly affect the performance of automated diagnosis systems and partly contribute to a stronger effect on the staging results than the used imaging modality. CONCLUSION: Chromoendoscopy has a negative impact on the results of the methods. NBI is better suited than chromoendoscopy. High-definition and high-magnification endoscopy are equally suited.
Assuntos
Pólipos do Colo/diagnóstico por imagem , Colonoscopia/métodos , Neoplasias Colorretais/prevenção & controle , Diagnóstico por Computador/métodos , Lesões Pré-Cancerosas/diagnóstico por imagem , Pólipos do Colo/patologia , Corantes/administração & dosagem , Humanos , Aumento da Imagem/métodos , Imagem de Banda Estreita/métodos , Lesões Pré-Cancerosas/patologia , Gravação em Vídeo/métodosRESUMO
A major challenge in the field of segmentation in digital pathology is given by the high effort for manual data annotations in combination with many sources introducing variability in the image domain. This requires methods that are able to cope with variability without requiring to annotate a large amount of samples for each characteristic. In this paper, we develop approaches based on adversarial models for image-to-image translation relying on unpaired training. Specifically, we propose approaches for stain-independent supervised segmentation relying on image-to-image translation for obtaining an intermediate representation. Furthermore, we develop a fully-unsupervised segmentation approach exploiting image-to-image translation to convert from the image to the label domain. Finally, both approaches are combined to obtain optimum performance in unsupervised segmentation independent of the characteristics of the underlying stain. Experiments on patches showing kidney histology proof that stain-translation can be performed highly effectively and can be used for domain adaptation to obtain independence of the underlying stain. It is even capable of facilitating the underlying segmentation task, thereby boosting the accuracy if an appropriate intermediate stain is selected. Combining domain adaptation with unsupervised segmentation finally showed the most significant improvements.
Assuntos
Histocitoquímica/métodos , Processamento de Imagem Assistida por Computador/métodos , Rim , Aprendizado de Máquina , Curadoria de Dados , Humanos , Rim/química , Rim/diagnóstico por imagem , Rim/patologiaRESUMO
BACKGROUND: Fat-fraction has been established as a relevant marker for the assessment and diagnosis of neuromuscular diseases. For computing this metric, segmentation of muscle tissue in MR images is a first crucial step. PURPOSE: To tackle the high degree of variability in combination with the high annotation effort for training supervised segmentation models (such as fully convolutional neural networks). STUDY TYPE: Prospective. SUBJECTS: In all, 41 patients consisting of 20 patients showing fatty infiltration and 21 healthy subjects. Field Strength/Sequence: The T1 -weighted MR-pulse sequences were acquired on a 1.5T scanner. ASSESSMENT: To increase performance with limited training data, we propose a domain-specific technique for simulating fatty infiltrations (i.e., texture augmentation) in nonaffected subjects' MR images in combination with shape augmentation. For simulating the fatty infiltrations, we make use of an architecture comprising several competing networks (generative adversarial networks) that facilitate a realistic artificial conversion between healthy and infiltrated MR images. Finally, we assess the segmentation accuracy (Dice similarity coefficient). STATISTICAL TESTS: A Wilcoxon signed rank test was performed to assess whether differences in segmentation accuracy are significant. RESULTS: The mean Dice similarity coefficients significantly increase from 0.84-0.88 (P < 0.01) using data augmentation if training is performed with mixed data and from 0.59-0.87 (P < 0.001) if training is conducted with healthy subjects only. DATA CONCLUSION: Domain-specific data adaptation is highly suitable for facilitating neural network-based segmentation of thighs with feasible manual effort for creating training data. The results even suggest an approach completely bypassing manual annotations. LEVEL OF EVIDENCE: 4 Technical Efficacy: Stage 3.
Assuntos
Processamento de Imagem Assistida por Computador/métodos , Imageamento por Ressonância Magnética , Doenças Neuromusculares/diagnóstico por imagem , Gordura Subcutânea/diagnóstico por imagem , Coxa da Perna/diagnóstico por imagem , Algoritmos , Simulação por Computador , Bases de Dados Factuais , Feminino , Voluntários Saudáveis , Humanos , Masculino , Redes Neurais de Computação , Estudos Prospectivos , Reprodutibilidade dos TestesRESUMO
Due to the increasing availability of whole slide scanners facilitating digitization of histopathological tissue, large amounts of digital image data are being generated. Accordingly, there is a strong demand for the development of computer based image analysis systems. Here, we address application scenarios in histopathology consisting of sparse, small objects-of-interest occurring in the large gigapixel images. To tackle the thereby arising challenges, we propose two different CNN cascade approaches which are subsequently applied to segment the glomeruli in whole slide images of the kidney and compared with conventional fully-convolutional networks. To facilitate unbiased evaluation, eight-fold cross-validation is performed and finally means and standard deviations are reported. Overall, with the best performing cascade approach, single CNNs are outperformed and a pixel-level Dice similarity coefficient of 0.90 is obtained (precision: 0.89, recall: 0.92). Combined with qualitative and further object-level analyses the obtained results are assessed as excellent also compared to previous approaches. We can state that especially one of the proposed cascade networks proved to be a highly powerful tool providing the best segmentation accuracies and also keeping the computing time at the lowest level. This work facilitates accurate automated segmentation of renal whole slide images which consequently allows fully-automated big data analyses for the assessment of medical treatments. Furthermore, this approach can also easily be adapted to other similar biomedical application scenarios.
Assuntos
Processamento de Imagem Assistida por Computador/métodos , Rim/diagnóstico por imagem , Redes Neurais de Computação , Animais , CamundongosRESUMO
BACKGROUND: In medical image data sets, the number of images is usually quite small. The small number of training samples does not allow to properly train classifiers which leads to massive overfitting to the training data. In this work, we investigate whether increasing the number of training samples by merging datasets from different imaging modalities can be effectively applied to improve predictive performance. Further, we investigate if the extracted features from the employed image representations differ between different imaging modalities and if domain adaption helps to overcome these differences. METHOD: We employ twelve feature extraction methods to differentiate between non-neoplastic and neoplastic lesions. Experiments are performed using four different classifier training strategies, each with a different combination of training data. The specifically designed setup for these experiments enables a fair comparison between the four training strategies. RESULTS: Combining high definition with high magnification training data and chromoscopic with non-chromoscopic training data partly improved the results. The usage of domain adaptation has only a small effect on the results compared to just using non-adapted training data. CONCLUSION: Merging datasets from different imaging modalities turned out to be partially beneficial for the case of combining high definition endoscopic data with high magnification endoscopic data and for combining chromoscopic with non-chromoscopic data. NBI and chromoendoscopy on the other hand are mostly too different with respect to the extracted features to combine images of these two modalities for classifier training.
Assuntos
Pólipos do Colo/diagnóstico por imagem , Diagnóstico por Computador/métodos , Reconhecimento Automatizado de Padrão , Algoritmos , Colonoscopia , Neoplasias Colorretais/diagnóstico por imagem , Diagnóstico por Imagem/métodos , Endoscopia , Humanos , Aumento da Imagem/métodosRESUMO
Severity and progression of degenerative neuromuscular diseases can be sensitively captured by evaluating the fat infiltration of muscle tissue in T1-weighted MRI scans of human limbs. For computing the fat fraction, the original muscle needs to be first separated from other tissue. Five conceptionally different approaches were investigated and evaluated with respect to the segmentation of muscles of human thighs. Besides a rather basic thresholding approach, local (level set) as well as global (graph cut) energy-minimizing segmentation approaches with and without a shape prior energy term were examined. For experimental evaluations, a dataset containing 37 subjects was divided into four classes according to the degree of fat infiltration. Results show that the choice of the best method depends on the severity of fat infiltration. In severe cases, the best results were obtained with shape prior based graph cuts, whereas in marginal cases thresholding was sufficient. With the best approach, the worst-case error in fat fraction computation was always below 11% and on average between 2% for tissue showing no fat infiltrations and 6% for heavily infiltrated tissue. The obtained Dice similarity coefficients, measuring the segmentation quality, were on average between 0.85 and 0.92. Although segmentation of heavily infiltrated muscle tissue is extremely difficult, an approach for reasonably segmenting these image data was identified. Especially the negative impact on the calculated fat fraction can be reduced significantly.
Assuntos
Tecido Adiposo/diagnóstico por imagem , Tecido Adiposo/patologia , Processamento de Imagem Assistida por Computador/métodos , Imageamento por Ressonância Magnética/métodos , Doenças Neuromusculares/diagnóstico por imagem , Doenças Neuromusculares/patologia , Adulto , Algoritmos , Humanos , Pessoa de Meia-Idade , Músculos/diagnóstico por imagem , Músculos/patologia , Índice de Gravidade de Doença , Coxa da Perna/diagnóstico por imagem , Coxa da Perna/patologiaRESUMO
Digital pathology is a field of increasing interest and requires automated systems for processing huge amounts of digital data. The development of supervised-learning based automated systems is aggravated by the fact that image properties can change from slide to slide. In this work, the focus is on the segmentation of the glomeruli constituting the most important regions-of-interest in renal histopathology. We propose and investigate a two-stage pipeline consisting of a weakly supervised patch-based detection and a precise segmentation. The proposed methods do not need any previously obtained training data. For adapting and optimizing this model, a kernel two-sample test is applied. For the segmentation stage, unsupervised segmentation methods including level-set and polygon-fitting approaches are adapted, combined and evaluated. Overall, with the best performing polygon-fitting segmentation method, 51% of glomeruli were segmented with sufficient accuracy (DSC > 0.8). 42% of the detections were false positives. Due to the difficult application scenario in combination with the small required training corpus, the obtained performance is assessed as good. Strategies for increasing the segmentation performance even further are discussed in detail.
Assuntos
Processamento de Imagem Assistida por Computador/métodos , Glomérulos Renais/diagnóstico por imagem , Modelos Teóricos , Feminino , Humanos , MasculinoRESUMO
AIM: To further improve the endoscopic detection of intestinal mucosa alterations due to celiac disease (CD). METHODS: We assessed a hybrid approach based on the integration of expert knowledge into the computer-based classification pipeline. A total of 2835 endoscopic images from the duodenum were recorded in 290 children using the modified immersion technique (MIT). These children underwent routine upper endoscopy for suspected CD or non-celiac upper abdominal symptoms between August 2008 and December 2014. Blinded to the clinical data and biopsy results, three medical experts visually classified each image as normal mucosa (Marsh-0) or villous atrophy (Marsh-3). The experts' decisions were further integrated into state-of-the-art texture recognition systems. Using the biopsy results as the reference standard, the classification accuracies of this hybrid approach were compared to the experts' diagnoses in 27 different settings. RESULTS: Compared to the experts' diagnoses, in 24 of 27 classification settings (consisting of three imaging modalities, three endoscopists and three classification approaches), the best overall classification accuracies were obtained with the new hybrid approach. In 17 of 24 classification settings, the improvements achieved with the hybrid approach were statistically significant (P < 0.05). Using the hybrid approach classification accuracies between 94% and 100% were obtained. Whereas the improvements are only moderate in the case of the most experienced expert, the results of the less experienced expert could be improved significantly in 17 out of 18 classification settings. Furthermore, the lowest classification accuracy, based on the combination of one database and one specific expert, could be improved from 80% to 95% (P < 0.001). CONCLUSION: The overall classification performance of medical experts, especially less experienced experts, can be boosted significantly by integrating expert knowledge into computer-aided diagnosis systems.
Assuntos
Doença Celíaca/diagnóstico , Diagnóstico por Computador/métodos , Endoscopia Gastrointestinal/métodos , Humanos , ConhecimentoRESUMO
Besides a high distinctiveness, robustness (or invariance) to image degradations is very desirable for texture feature extraction methods in real-world applications. In this paper, focus is on making arbitrary texture descriptors invariant to blur which is often prevalent in real image data. From previous work, we know that most state-of-the-art texture feature extraction methods are unable to cope even with minor blur degradations if the classifier's training stage is based on idealistic data. However, if the training set suffers similarly from the degradations, the obtained accuracies are significantly higher. Exploiting that knowledge, in this approach the level of blur of each image is increased to a certain threshold, based on the estimation of a blur measure. Experiments with synthetically degraded data show that the method is able to generate a high degree of blur invariance without loosing too much distinctiveness. Finally, we show that our method is not limited to ideal Gaussian blur.