| 1. |
- Andersson, Axel, et al.
(författare)
-
End-to-end Multiple Instance Learning with Gradient Accumulation
- 2022
-
Ingår i: 2022 IEEE International Conference on Big Data (Big Data). - : Institute of Electrical and Electronics Engineers (IEEE). - 9781665480451 - 9781665480468 ; , s. 2742-2746
-
Konferensbidrag (refereegranskat)abstract
- Being able to learn on weakly labeled data and provide interpretability are two of the main reasons why attention-based deep multiple instance learning (ABMIL) methods have become particularly popular for classification of histopathological images. Such image data usually come in the form of gigapixel-sized whole-slide-images (WSI) that are cropped into smaller patches (instances). However, the sheer volume of the data poses a practical big data challenge: All the instances from one WSI cannot fit the GPU memory of conventional deep-learning models. Existing solutions compromise training by relying on pre-trained models, strategic selection of instances, sub-sampling, or self-supervised pre-training. We propose a training strategy based on gradient accumulation that enables direct end-to-end training of ABMIL models without being limited by GPU memory. We conduct experiments on both QMNIST and Imagenette to investigate the performance and training time and compare with the conventional memory-expensive baseline as well as a recent sampled-based approach. This memory-efficient approach, although slower, reaches performance indistinguishable from the memory-expensive baseline.
|
|
| 2. |
|
|
| 3. |
|
|
| 4. |
- Bajic, Buda, et al.
(författare)
-
Blind restoration of images degraded with mixed poisson-Gaussian noise with application in transmission electron microscopy
- 2016
-
Ingår i: 2016 Ieee 13Th International Symposium On Biomedical Imaging (ISBI). - : IEEE. - 9781479923496 - 9781479923502 ; , s. 123-127
-
Konferensbidrag (refereegranskat)abstract
- Noise and blur, present in images after acquisition, negatively affect their further analysis. For image enhancement when the Point Spread Function (PSF) is unknown, blind deblurring is suitable, where both the PSF and the original image are simultaneously reconstructed. In many realistic imaging conditions, noise is modelled as a mixture of Poisson (signal-dependent) and Gaussian (signal independent) noise. In this paper we propose a blind deconvolution method for images degraded by such mixed noise. The method is based on regularized energy minimization. We evaluate its performance on synthetic images, for different blur kernels and different levels of noise, and compare with non-blind restoration. We illustrate the performance of the method on Transmission Electron Microscopy images of cilia, used in clinical practice for diagnosis of a particular type of genetic disorders.
|
|
| 5. |
|
|
| 6. |
- Bajic, Buda, et al.
(författare)
-
Generalised deep learning framework for HEp-2 cell recognition using local binary pattern maps
- 2020
-
Ingår i: IET Image Processing. - : INST ENGINEERING TECHNOLOGY-IET. - 1751-9659 .- 1751-9667. ; 14:6, s. 1201-1208
-
Tidskriftsartikel (refereegranskat)abstract
- The authors propose a novel HEp-2 cell image classifier to improve the automation process of patients' serum evaluation. The authors' solution builds on the recent progress in deep learning based image classification. They propose an ensemble approach using multiple state-of-the-art architectures. They incorporate additional texture information extracted by an improved version of local binary patterns maps, $\alpha $alpha LBP-maps, which enables to create a very effective cell image classifier. This innovative combination is trained on three publicly available datasets and its general applicability is demonstrated through the evaluation on three independent test sets. The presented results show that their approach leads to a general improvement of performance on average on the three public datasets.
|
|
| 7. |
- Bajic, Buda, et al.
(författare)
-
Restoration of images degraded by signal-dependent noise based on energy minimization : an empirical study
- 2016
-
Ingår i: Journal of Electronic Imaging (JEI). - 1017-9909 .- 1560-229X. ; 25:4
-
Tidskriftsartikel (refereegranskat)abstract
- Most energy minimization-based restoration methods are developed for signal-independent Gaussian noise. The assumption of Gaussian noise distribution leads to a quadratic data fidelity term, which is appealing in optimization. When an image is acquired with a photon counting device, it contains signal-dependent Poisson or mixed Poisson–Gaussian noise. We quantify the loss in performance that occurs when a restoration method suited for Gaussian noise is utilized for mixed noise. Signal-dependent noise can be treated by methods based on either classical maximum a posteriori (MAP) probability approach or on a variance stabilization approach (VST). We compare performances of these approaches on a large image material and observe that VST-based methods outperform those based on MAP in both quality of restoration and in computational efficiency. We quantify improvement achieved by utilizing Huber regularization instead of classical total variation regularization. The conclusion from our study is a recommendation to utilize a VST-based approach combined with regularization by Huber potential for restoration of images degraded by blur and signal-dependent noise. This combination provides a robust and flexible method with good performance and high speed.
|
|
| 8. |
- Bajic, Buda, et al.
(författare)
-
Single image super-resolution reconstruction in presence of mixed Poisson-Gaussian noise
- 2016
-
Ingår i: 2016 SIXTH INTERNATIONAL CONFERENCE ON IMAGE PROCESSING THEORY, TOOLS AND APPLICATIONS (IPTA). - : IEEE. - 9781467389105
-
Konferensbidrag (refereegranskat)abstract
- Single image super-resolution (SR) reconstructionaims to estimate a noise-free and blur-free high resolution imagefrom a single blurred and noisy lower resolution observation.Most existing SR reconstruction methods assume that noise in theimage is white Gaussian. Noise resulting from photon countingdevices, as commonly used in image acquisition, is, however,better modelled with a mixed Poisson-Gaussian distribution. Inthis study we propose a single image SR reconstruction methodbased on energy minimization for images degraded by mixedPoisson-Gaussian noise.We evaluate performance of the proposedmethod on synthetic images, for different levels of blur andnoise, and compare it with recent methods for non-Gaussiannoise. Analysis shows that the appropriate treatment of signaldependentnoise, provided by our proposed method, leads tosignificant improvement in reconstruction performance.
|
|
| 9. |
- Bajic, Buda, et al.
(författare)
-
Sparsity promoting super-resolution coverage segmentation by linear unmixing in presence of blur and noise
- 2019
-
Ingår i: Journal of Electronic Imaging (JEI). - : IS&T & SPIE. - 1017-9909 .- 1560-229X. ; 28:1
-
Tidskriftsartikel (refereegranskat)abstract
- We present a segmentation method that estimates the relative coverage of each pixel in a sensed image by each image component. The proposed super-resolution blur-aware model (utilizes a priori knowledge of the image blur) for linear unmixing of image intensities relies on a sparsity promoting approach expressed by two main requirements: (i) minimization of Huberized total variation, providing smooth object boundaries and noise removal, and (ii) minimization of nonedge image fuzziness, responding to an assumption that imaged objects are crisp and that fuzziness is mainly due to the imaging and digitization process. Edge fuzziness due to partial coverage is allowed, enabling subpixel precise feature estimates. The segmentation is formulated as an energy minimization problem and solved by the spectral projected gradient method, utilizing a graduated nonconvexity scheme. Quantitative and qualitative evaluation on synthetic and real multichannel images confirms good performance, particularly relevant when subpixel precision in segmentation and subsequent analysis is a requirement. (C) 2019 SPIE and IS&T
|
|
| 10. |
-
Bioimage Data Analysis Workflows
- 2019
-
Samlingsverk (redaktörskap) (refereegranskat)abstract
- This Open Access textbook provides students and researchers in the life sciences with essential practical information on how to quantitatively analyze data images. It refrains from focusing on theory, and instead uses practical examples and step-by step protocols to familiarize readers with the most commonly used image processing and analysis platforms such as ImageJ, MatLab and Python. Besides gaining knowhow on algorithm usage, readers will learn how to create an analysis pipeline by scripting language; these skills are important in order to document reproducible image analysis workflows.The textbook is chiefly intended for advanced undergraduates in the life sciences and biomedicine without a theoretical background in data analysis, as well as for postdocs, staff scientists and faculty members who need to perform regular quantitative analyses of microscopy images.
|
|
| 11. |
|
|
| 12. |
|
|
| 13. |
|
|
| 14. |
- Buga, Roxana M., et al.
(författare)
-
HISTOBREAST : a collection of brightfield microscopy images of Haematoxylin and Eosin stained breast tissue
- 2020
-
Ingår i: Scientific Data. - : Springer Science and Business Media LLC. - 2052-4463. ; 7:1
-
Tidskriftsartikel (refereegranskat)abstract
- Modern histopathology workflows rely on the digitization of histology slides. The quality of the resulting digital representations, in the form of histology slide image mosaics, depends on various specific acquisition conditions and on the image processing steps that underlie the generation of the final mosaic, e.g. registration and blending of the contained image tiles. We introduce HISTOBREAST, an extensive collection of brightfield microscopy images that we collected in a principled manner under different acquisition conditions on Haematoxylin - Eosin (H&E) stained breast tissue. HISTOBREAST is comprised of neighbour image tiles and ensemble of mosaics composed from different combinations of the available image tiles, exhibiting progressively degraded quality levels. HISTOBREAST can be used to benchmark image processing and computer vision techniques with respect to their robustness to image modifications specific to brightfield microscopy of H&E stained tissues. Furthermore, HISTOBREAST can serve in the development of new image processing methods, with the purpose of ensuring robustness to typical image artefacts that raise interpretation problems for expert histopathologists and affect the results of computerized image analysis.
|
|
| 15. |
- Chanussot, Jocelyn, et al.
(författare)
-
Shape signaturs of fuzzy star-shaped sets based on distance from the centroid
- 2005
-
Ingår i: Pattern Recognition Letters. - : Elsevier. - 0167-8655. ; 26:6, s. 735-746
-
Tidskriftsartikel (refereegranskat)abstract
- We extend the shape signature based on the distance of the boundary points from the shape centroid, to the case of fuzzy sets. The analysis of the transition from crisp to fuzzy shape descriptor is first given in the continuous case. This is followed by a study of the specific issues induced by the discrete representation of the objects in a computer.We analyze two methods for calculating the signature of a fuzzy shape, derived from two ways of defining a fuzzy set: first, by its membership function, and second, as a stack of its α-cuts. The first approach is based on measuring the length of a fuzzy straight line by integration of the fuzzy membership function, while in the second one we use averaging of the shape signatures obtained for the individual α-cuts of the fuzzy set. The two methods, equivalent in the continuous case for the studied class of fuzzy shapes, produce different results when adjusted to the discrete case. A statistical study, aiming at characterizing the performances of each method in the discrete case, is done. Both methods are shown to provide more precise descriptions than their corresponding crisp versions. The second method (based on averaged Euclidean distance over the α-cuts) outperforms the others.
|
|
| 16. |
- Chatterjee, Swarnadip, 1990-, et al.
(författare)
-
DCNN based Oral Cancer Screening using Whole Slide Cytology Images : Effect of Increased Patch Size
- 2022
-
Konferensbidrag (övrigt vetenskapligt/konstnärligt)abstract
- Cases of Oral Cancer are increasing around the world. Oral Squamous Cell Carcinomas constitute majority of all Oral Cancer cases and arise from the oral epithelium. Although this type of Oral Cancer is highly accessible to clinicians as they are superficial, they are often discovered late. To improve early detection in order to increase the chances of survival, we propose a Deep Convolutional Neural Network based framework on whole slide cytology images. In this ongoing work, we have shown that increasing the size of the patch centered at the detected nuclei, increases the accuracy of classification of the pathological condition of the nuclei using only slide-level labels.
|
|
| 17. |
- Chatterjee, Swarnadip, 1990-, et al.
(författare)
-
Investigating the Relevance of Contextual Information Towards Improving Deep CNN Based Oral Cancer Screening on Whole Slide Cytology Samples
- 2023
-
Konferensbidrag (övrigt vetenskapligt/konstnärligt)abstract
- Cases of Oral Cancer are increasing around the world. Oral Squamous Cell Carcinomas constitute the majority of all oral cancer cases and arise from the oral epithelium. Although this type of cancer is superficial and highly accessible to clinicians, it is often discovered late. To improve early detection in order to increase the chances of survival, we propose a Deep Convolutional Neural Network based framework on whole slide cytology images. In this ongoing work, we investigate the relevance of contextual information towards improving accuracy of classification of the pathological condition of cells from brush samples using only slide-level labels. For this, we consider nuclei centered patches of sizes 80 × 80, 160 × 160, 240 × 240, and 320 × 320 pixels and observe an increasing trend in the classification performances with respect to increasing patch sizes for three deep CNN architectures: ResNet50, DenseNet201 and SEResNet50.
|
|
| 18. |
- Cimini, Beth A, et al.
(författare)
-
The NEUBIAS Gateway : A hub for bioimage analysis methods and materials.
- 2020
-
Ingår i: F1000Research. - : F1000 Research Ltd. - 2046-1402.
-
Annan publikation (övrigt vetenskapligt/konstnärligt)abstract
- We introduce the NEUBIAS Gateway, a new platform for publishing materials related to bioimage analysis, an interdisciplinary field bridging computer science and life sciences. This emerging field has been lacking a central place to share the efforts of the growing group of scientists addressing biological questions using image data. The Gateway welcomes a wide range of publication formats including articles, reviews, reports and training materials. We hope the Gateway further supports this important field to grow and helps more biologists and computational scientists learn about and contribute to these efforts.
|
|
| 19. |
|
|
| 20. |
|
|
| 21. |
- Curic, Vladimir, 1981-, et al.
(författare)
-
Distance measures between digital fuzzy objects and their applicability in image processing
- 2011
-
Ingår i: Combinatorial Image Analysis. - Berlin, Heidelberg : Springer Berlin/Heidelberg. - 9783642210723 ; 6636, s. 385-397
-
Konferensbidrag (refereegranskat)abstract
- We present two different extensions of the Sum of minimal distances and the Complement weighted sum of minimal distances to distances between fuzzy sets. We evaluate to what extent the proposed distances show monotonic behavior with respect to increasing translation and rotation of digital objects, in noise free, as well as in noisy conditions. Tests show that one of the extension approaches leads to distances exhibiting very good performance. Furthermore, we evaluate distance based classification of crisp and fuzzy representations of objects at a range of resolutions. We conclude that the proposed distances are able to utilize the additional information available in a fuzzy representation, thereby leading to improved performance of related image processing tasks.
|
|
| 22. |
- Curic, Vladimir, 1981-, et al.
(författare)
-
The Sum of minimal distances as a useful distance measure for image registration
- 2010
-
Ingår i: Proceedings SSBA 2010. - Uppsala : Centre for Image Analysis. ; , s. 55-58
-
Konferensbidrag (övrigt vetenskapligt/konstnärligt)abstract
- In this paper we study set distances which are used in image registration related problems. We introduced a new distance as a Sum of minimal distances with added linear weights. Linear weights are added in a way to reduce the impact of single outliers. An evaluation of observed distances with respect to applicability to image object registration is performed. A comparative study of set distances with respect to noise sensitivity as well as with respect to translation and rotation of objects in image is presented. Based on our experiments on synthetic images containing various types of noise, we determine that the proposed weighted sum of minimal distances has a good performances for object registration.
|
|
| 23. |
- Delic, Marija, et al.
(författare)
-
αLBP – a novel member of the Local Binary Pattern family based on α-cutting
- 2015
-
Ingår i: Proc. 9th International Symposium on Image and Signal Processing and Analysis. - : IEEE. - 9781467380324 ; , s. 13-18
-
Konferensbidrag (refereegranskat)abstract
- Local binary pattern (LBP) descriptors have been popular in texture classification in recent years. They were introduced as descriptors of local image texture and their histograms are shown to be well performing texture features. In this paper we introduce two new LBP descriptors, αLBP and its improved variant IαLBP. We evaluate their performance in classification by comparing them with some of the existing LBP descriptors - LBP, ILBP, shift LBP (SLBP) and with one ternary descriptor - LTP. The texture descriptors are evaluated on three datasets - KTH-TIPS2b, UIUC and Virus texture dataset. The novel descriptor outperforms the other descriptors on two datasets, KTH-TIPS2b and Virus, and is tied for first place with ILBP on the UIUC dataset.
|
|
| 24. |
-
Discrete Geometry and Mathematical Morphology : First International Joint Conference, DGMM 2021, Uppsala, Sweden, May 24–27, 2021, Proceedings
- 2021
-
Samlingsverk (redaktörskap) (refereegranskat)abstract
- This book constitutes the proceedings of the First IAPR International Conference on Discrete Geometry and Mathematical Morphology, DGMM 2021, which was held during May 24-27, 2021, in Uppsala, Sweden.The conference was created by joining the International Conference on Discrete Geometry for computer Imagery, DGCI, with the International Symposium on Mathematical Morphology, ISMM.The 36 papers included in this volume were carefully reviewed and selected from 59 submissions. They were organized in topical sections as follows: applications in image processing, computer vision, and pattern recognition; discrete and combinatorial topology; discrete geometry - models, transforms, visualization; discrete tomography and inverse problems; hierarchical and graph-based models, analysis and segmentation; learning-based approaches to mathematical morphology; multivariate and PDE-based mathematical morphology, morphological filtering.The book also contains 3 invited keynote papers.
|
|
| 25. |
|
|
| 26. |
- Dražić, Slobodan, et al.
(författare)
-
Precise Estimation of the Projection of a Shape from a Pixel Coverage Representation
- 2011
-
Ingår i: Proceedings of the 7th IEEE International Symposium on Image and Signal Processing and Analysis (ISPA). - : IEEE Computer Society. - 9781457708411 ; , s. 569-574, s. 569-574
-
Konferensbidrag (refereegranskat)abstract
- Measuring width and diameter of a shape areproblems well studied in the literature. A pixel coverage repre-sentation is one specific type of digital fuzzy representation of acontinuous image object, where the (membership) value of eachpixel is (approximately) equal to the relative area of the pixelwhich is covered by the continuous object. Lately a number ofmethods for shape analysis use pixel coverage for reducing errorof estimation. We introduce a novel method for estimating theprojection of a shape in a given direction. The method is based onutilizing pixel coverage representation of a shape. Performance ofthe method is evaluated by a number of tests on synthetic objects,confirming high precision and applicability for calculation ofdiameter and elongation of a shape.
|
|
| 27. |
- Fraile, Marc, et al.
(författare)
-
Automatic Analysis of Infant Engagement during Play : An End-to-end Learning and Explainable AI Pilot Experiment
- 2021
-
Ingår i: Companion Publication of the 2021 International Conference on Multimodal Interaction. - New York, NY, USA : Association for Computing Machinery (ACM).
-
Konferensbidrag (refereegranskat)abstract
- Infant engagement during play is an active area of research, relatedto the development of cognition. Automatic detection of engagementcould benefit the research process, but existing techniquesused for automatic affect detection are unsuitable for this scenario,since they rely on the automatic extraction of facial and posturalfeatures trained on clear video capture of adults. This study showsthat end-to-end Deep Learning methods can successfully detectengagement of infants, without the need of clear facial video, whentrained for a specific interaction task. It further shows that attentionmapping techniques can provide explainability, thereby enablingtrust and insight into a model’s reasoning process.
|
|
| 28. |
- Fraile, Marc, PhD Candidate, 1989-, et al.
(författare)
-
End-to-End Learning and Analysis of Infant Engagement During Guided Play : Prediction and Explainability
- 2022
-
Ingår i: ICMI '22. - New York, NY, USA : Association for Computing Machinery (ACM). - 9781450393904 ; , s. 444-454
-
Konferensbidrag (refereegranskat)abstract
- Infant engagement during guided play is a reliable indicator of early learning outcomes, psychiatric issues and familial wellbeing. An obstacle to using such information in real-world scenarios is the need for a domain expert to assess the data. We show that an end-to-end Deep Learning approach can perform well in automatic infant engagement detection from a single video source, without requiring a clear view of the face or the whole body. To tackle the problem of explainability in learning methods, we evaluate how four common attention mapping techniques can be used to perform subjective evaluation of the network’s decision process and identify multimodal cues used by the network to discriminate engagement levels. We further propose a quantitative comparison approach, by collecting a human attention baseline and evaluating its similarity to each technique.
|
|
| 29. |
|
|
| 30. |
- Gay, Jo, et al.
(författare)
-
Texture-based oral cancer detection: A performance analysis of deep learning approaches.
- 2019
-
Ingår i: 3rd NEUBIAS Conference. - Luxembourg.
-
Konferensbidrag (övrigt vetenskapligt/konstnärligt)abstract
- Early stage cancer detection is essential for reducing cancer mortality. Screening programs such as that for cervical cancer are highly effective in preventing advanced stage cancers. One obstacle to the introduction of screening for other cancer types is the cost associated with manual inspection of the resulting cell samples. Computer assisted image analysis of cytology slides may offer a significant reduction of these costs. We are particularly interested in detection of cancer of the oral cavity, being one of the most common malignancies in the world, with an increasing tendency of incidence among young people. Due to the non-invasive accessibility of the oral cavity, automated detection may enable screening programs leading to early diagnosis and treatment.It is well known that variations in the chromatin texture of the cell nucleus are an important diagnostic feature. With an aim to maximize reliability of an automated cancer detection system for oral cancer detection, we evaluate three state of the art deep convolutional neural network (DCNN) approaches which are specialized for texture analysis. A powerful tool for texture description are local binary patterns (LBPs); they describe the pattern of variations in intensity between a pixel and its neighbours, instead of using the image intensity values directly. A neural network can be trained to recognize the range of patterns found in different types of images. Many methods have been proposed which either use LBPs directly, or are inspired by them, and show promising results on a range of different image classification tasks where texture is an important discriminative feature.We evaluate multiple recently published deep learning-based texture classification approaches: two of them (referred to as Model 1, by Juefei-Xu et al. (CVPR 2017); Model 2, by Li et al. (2018)) are inspired by LBP texture descriptors, while the third (Model 3, by Marcos et al. (ICCV 2017)), based on Rotation Equivariant Vector Field Networks, aims at preserving fine textural details under rotations, thus enabling a reduced model size. Performances are compared with state-of-the-art results on the same dataset, by Wieslander et al. (CVPR 2017), which are based on ResNet and VGG architectures. Furthermore a fusion of DCNN with LBP maps as in Wetzer et al. (Bioimg. Comp. 2018) is evaluated for comparison. Our aim is to explore if focus on texture can improve CNN performance.Both of the methods based on LBPs exhibit higher performances (F1-score for Model 1: 0.85; Model 2: 0.83) than what is obtained by using CNNs directly on the greyscale data (VGG: 0.78, ResNet: 0.76). This clearly demonstrates the effectiveness of LBPs for this type of image classification task. The approach based on rotation equivariant networks stays behind in performance (F1-score for Model 3: 0.72), indicating that this method may be less appropriate for classifying single-cell images.
|
|
| 31. |
- Gupta, Anindya, et al.
(författare)
-
Convolutional neural networks for false positive reduction of automatically detected cilia in low magnification TEM images
- 2017
-
Ingår i: Image Analysis. - Cham : Springer. - 9783319591254 ; , s. 407-418
-
Konferensbidrag (refereegranskat)abstract
- Automated detection of cilia in low magnification transmission electron microscopy images is a central task in the quest to relieve the pathologists in the manual, time consuming and subjective diagnostic procedure. However, automation of the process, specifically in low magnification, is challenging due to the similar characteristics of non-cilia candidates. In this paper, a convolutional neural network classifier is proposed to further reduce the false positives detected by a previously presented template matching method. Adding the proposed convolutional neural network increases the area under Precision-Recall curve from 0.42 to 0.71, and significantly reduces the number of false positive objects.
|
|
| 32. |
|
|
| 33. |
|
|
| 34. |
- Ilic, Vladimir, et al.
(författare)
-
Precise Euclidean distance transforms in 3D from voxel coverage representation
- 2015
-
Ingår i: Pattern Recognition Letters. - : Elsevier BV. - 0167-8655 .- 1872-7344. ; 65, s. 184-191
-
Tidskriftsartikel (refereegranskat)abstract
- Distance transforms (DTs) are, usually, defined on a binary image as a mapping from each background element to the distance between its centre and the centre of the closest object element. However, due to discretization effects, such DTs have limited precision, including reduced rotational and translational invariance. We show in this paper that a significant improvement in performance of Euclidean DTs can be achieved if voxel coverage values are utilized and the position of an object boundary is estimated with sub-voxel precision. We propose two algorithms of linear time complexity for estimating Euclidean DT with sub-voxel precision. The evaluation confirms that both algorithms provide 4-14 times increased accuracy compared to what is achievable from a binary object representation.
|
|
| 35. |
- Ilic, Vladimir, et al.
(författare)
-
Signature of a Shape Based on Its Pixel Coverage Representation
- 2016
-
Ingår i: Discrete Geometry for Computer Imagery, DGCI 2016. Lecture Notes in Computer Science, Vol. 9647, pp. 181-193, Springer 2016. - Cham : Springer Berlin/Heidelberg. ; , s. 181-193
-
Konferensbidrag (refereegranskat)abstract
- Distance from the boundary of a shape to its centroid, a.k.a. signature of a shape, is a frequently used shape descriptor. Commonly, the observed shape results from a crisp (binary) segmentation of an image. The loss of information associated with binarization leads to a significant decrease in accuracy and precision of the signature, as well as its reduced invariance w.r.t. translation and rotation. Coverage information enables better estimation of edge position within a pixel. In this paper, we propose an iterative method for computing the signature of a shape utilizing its pixel coverage representation. The proposed method iteratively improves the accuracy of the computed signature, starting from a good initial estimate. A statistical study indicates considerable improvements in both accuracy and precision, compared to a crisp approach and a previously proposed approach based on averaging signatures over α-cuts of a fuzzy representation. We observe improved performance of the proposed descriptor in the presence of noise and reduced variation under translation and rotation.
|
|
| 36. |
- Kahraman, Ali T., et al.
(författare)
-
Automated detection, segmentation and measurement of major vessels and the trachea in CT pulmonary angiography
- 2023
-
Ingår i: Scientific Reports. - : Springer Nature. - 2045-2322. ; 13
-
Tidskriftsartikel (refereegranskat)abstract
- Mediastinal structure measurements are important for the radiologist's review of computed tomography pulmonary angiography (CTPA) examinations. In the reporting process, radiologists make measurements of diameters, volumes, and organ densities for image quality assessment and risk stratification. However, manual measurement of these features is time consuming. Here, we sought to develop a time-saving automated algorithm that can accurately detect, segment and measure mediastinal structures in routine clinical CTPA examinations. In this study, 700 CTPA examinations collected and annotated. Of these, a training set of 180 examinations were used to develop a fully automated deterministic algorithm. On the test set of 520 examinations, two radiologists validated the detection and segmentation performance quantitatively, and ground truth was annotated to validate the measurement performance. External validation was performed in 47 CTPAs from two independent datasets. The system had 86-100% detection and segmentation accuracy in the different tasks. The automatic measurements correlated well to those of the radiologist (Pearson's r 0.68-0.99). Taken together, the fully automated algorithm accurately detected, segmented, and measured mediastinal structures in routine CTPA examinations having an adequate representation of common artifacts and medical conditions.
|
|
| 37. |
|
|
| 38. |
- Koriakina, Nadezhda, 1991-, et al.
(författare)
-
Deep multiple instance learning versus conventional deep single instance learning for interpretable oral cancer detection
- 2024
-
Ingår i: PLOS ONE. - : Public Library of Science (PLoS). - 1932-6203. ; 19:4 April
-
Tidskriftsartikel (refereegranskat)abstract
- The current medical standard for setting an oral cancer (OC) diagnosis is histological examination of a tissue sample taken from the oral cavity. This process is time-consuming and more invasive than an alternative approach of acquiring a brush sample followed by cytological analysis. Using a microscope, skilled cytotechnologists are able to detect changes due to malignancy; however, introducing this approach into clinical routine is associated with challenges such as a lack of resources and experts. To design a trustworthy OC detection system that can assist cytotechnologists, we are interested in deep learning based methods that can reliably detect cancer, given only per-patient labels (thereby minimizing annotation bias), and also provide information regarding which cells are most relevant for the diagnosis (thereby enabling supervision and understanding). In this study, we perform a comparison of two approaches suitable for OC detection and interpretation: (i) conventional single instance learning (SIL) approach and (ii) a modern multiple instance learning (MIL) method. To facilitate systematic evaluation of the considered approaches, we, in addition to a real OC dataset with patient-level ground truth annotations, also introduce a synthetic dataset—PAP-QMNIST. This dataset shares several properties of OC data, such as image size and large and varied number of instances per bag, and may therefore act as a proxy model of a real OC dataset, while, in contrast to OC data, it offers reliable per-instance ground truth, as defined by design. PAP-QMNIST has the additional advantage of being visually interpretable for non-experts, which simplifies analysis of the behavior of methods. For both OC and PAP-QMNIST data, we evaluate performance of the methods utilizing three different neural network architectures. Our study indicates, somewhat surprisingly, that on both synthetic and real data, the performance of the SIL approach is better or equal to the performance of the MIL approach. Visual examination by cytotechnologist indicates that the methods manage to identify cells which deviate from normality, including malignant cells as well as those suspicious for dysplasia. We share the code as open source.
|
|
| 39. |
- Koriakina, Nadezhda, 1991-, et al.
(författare)
-
The Effect of Within-Bag Sampling on End-to-End Multiple Instance Learning
- 2021
-
Ingår i: Proceedings of the 12th International Symposium on Image and Signal Processing and Analysis (ISPA). - : Institute of Electrical and Electronics Engineers (IEEE). - 9781665426404 - 9781665426398 ; , s. 183-188
-
Konferensbidrag (refereegranskat)abstract
- End-to-end multiple instance learning (MIL) is an important concept with a wide range of applications. It is gaining increased popularity in the (bio)medical imaging community since it may provide a possibility to, while relying only on weak labels assigned to large regions, obtain more fine-grained information. However, processing very large bags in end-to-end MIL is problematic due to computer memory constraints. We propose within-bag sampling as one way of applying end-to-end MIL methods on very large data. We explore how different levels of sampling affect the performance of a well-known high-performing end-to-end attention-based MIL method, to understand the conditions when sampling can be utilized. We compose two new datasets tailored for the purpose of the study, and propose a strategy for sampling during MIL inference to arrive at reliable bag labels as well as instance level attention weights. We perform experiments without and with different levels of sampling, on the two publicly available datasets, and for a range of learning settings. We observe that in most situations the proposed bag-level sampling can be applied to end-to-end MIL without performance loss, supporting its confident usage to enable end-to-end MIL also in scenarios with very large bags. We share the code as open source at https://github.com/MIDA-group/SampledABMIL
|
|
| 40. |
- Koriakina, Nadezhda, 1991-, et al.
(författare)
-
Uncovering hidden reasoning of convolutional neural networks in biomedical image classification by using attribution methods
- 2020
-
Ingår i: 4th NEUBIAS Conference, Bordeaux, France.
-
Konferensbidrag (övrigt vetenskapligt/konstnärligt)abstract
- Convolutional neural networks (CNNs) are very popular in biomedical image processing and analysis, due to their impressive performance on numerous tasks. However, the performance comes at a cost of limited interpretability, which may harm users' trust in methods and their results. Robust and trustworthy methods are particularly in demand in the medical domain due to the sensitivity of the matter. There is a limited understanding of what CNNs base their decisions on, and, in particular, how their performance is related to what they are paying attention to. In this study, we utilize popular attribution methods, with the aim to explore relations between properties of a network's attention and its accuracy and certainty in classification. An intuitive reasoning is that in order for a network to make good decisions, it has to be consistent in what to draw attention to. We take a step towards understanding CNNs' behavior by identifying a relation between the model performance and the variability of its attention map.We observe two biomedical datasets and two commonly used architectures. We train several identical models of the same architecture on the given data; these identical models differ due to stochasticity of initialization and training. We analyse the variability of the predictions from such collections of networks where we observe all the network instances and their classifications independently. We utilize Gradient-weighted Class Activation Mapping (Grad-CAM) and Layer-wise Relevance Propagation (LRP), frequently employed attribution methods, for the activation analysis. Given a collection of trained CNNs, we compute, for each image of the test set: (i) the mean and standard deviation (SD) of the accuracy, over the networks in the collection; (ii) the mean and SD of the respective attention maps. We plot these measures against each other for the different combinations of network architectures and datasets, in order to expose possible relations between them.Our results reveal that there exists a relation between the variability of accuracy for collections of identical models and the variability of corresponding attention maps and that this relation is consistent among the considered combinations of datasets and architectures. We observe that the aggregated standard deviation of attention maps has a quadratic relation to the average accuracy of the sets of models and a linear relation to the standard deviation of accuracy. Motivated by the results, we are also performing subsequent experiments to reveal the relation between the score and attention, as well as to understand the impact of different images to the prediction by using mentioned statistics for each image and clustering techniques. These constitute important steps towards improved explainability and a generally clearer picture of the decision-making process of CNNs for biomedical data.
|
|
| 41. |
- Koriakina, Nadezhda, 1991-, et al.
(författare)
-
Visualization of convolutional neural network class activations in automated oral cancer detection for interpretation of malignancy associated changes
- 2019
-
Ingår i: 3rd NEUBIAS Conference, Luxembourg, 2-8 February 2019.
-
Konferensbidrag (övrigt vetenskapligt/konstnärligt)abstract
- Introduction: Cancer of the oral cavity is one of the most common malignancies in the world. The incidence of oral cavity and oropharyngeal cancer is increasing among young people. It is noteworthy that the oral cavity can be relatively easily accessed for routine screening tests that could potentially decrease the incidence of oral cancer. Automated deep learning computer aided methods show promising ability for detection of subtle precancerous changes at a very early stage, also when visual examination is less effective. Although the biological nature of these malignancy associated changes is not fully understood, the consistency of morphology and textural changes within a cell dataset could shed light on the premalignant state. In this study, we are aiming to increase understanding of this phenomenon by exploring and visualizing what parts of cell images are considered as most important when trained deep convolutional neural networks (DCNNs) are used to differentiate cytological images into normal and abnormal classes.Materials and methods: Cell samples are collected with a brush at areas of interest in the oral cavity and stained according to standard PAP procedures. Digital images from the slides are acquired with a 0.32 micron pixel size in greyscale format (570 nm bandpass filter). Cell nuclei are manually selected in the images and a small region is cropped around each nucleus resulting in images of 80x80 pixels. Medical knowledge is not used for choosing the cells but they are just randomly selected from the glass; for the learning process we are only providing ground truth on the patient level and not on the cell level. Overall, 10274 images of cell nuclei and the surrounding region are used to train state-of-the-art DCNNs to distinguish between cells from healthy persons and persons with precancerous lesions. Data augmentation through 90 degrees rotations and mirroring is applied to the datasets. Different approaches for class activation mapping and related methods are utilized to determine what image regions and feature maps are responsible for the relevant class differentiation.Results and Discussion:The best performing of the observed deep learning architectures reaches a per cell classification accuracy surpassing 80% on the observed material. Visualizing the class activation maps confirms our expectation that the network is able to learn to focus on specific relevant parts of the sample regions. We compare and evaluate our findings related to detected discriminative regions with the subjective judgements of a trained cytotechnologist. We believe that this effort on improving understanding of decision criteria used by machine and human leads to increased understanding of malignancy associated changes and also improves robustness and reliability of the automated malignancy detection procedure.
|
|
| 42. |
- Lidayová, Kristína, et al.
(författare)
-
Coverage segmentation of 3D thin structures
- 2015
-
Ingår i: Image Processing Theory, Tools and Applications (IPTA), 2015 International Conference on. - Piscataway, NJ : IEEE conference proceedings. - 9781479986361 ; , s. 23-28
-
Konferensbidrag (refereegranskat)abstract
- We present a coverage segmentation method for extracting thin structures in three-dimensional images. The proposed method is an improved extension of our coverage segmentation method for 2D thin structures. We suggest implementation that enables low memory consumption and processing time, and by that applicability of the method on real CTA data. The method needs a reliable crisp segmentation as an input and uses information from linear unmixing and the crisp segmentation to create a high-resolution crisp reconstruction of the object, which can then be used as a final result, or down-sampled to a coverage segmentation at the starting image resolution. Performed quantitative and qualitative analysis confirm excellent performance of the proposed method, both on synthetic and on real data, in particular in terms of robustness to noise.
|
|
| 43. |
|
|
| 44. |
- Lindblad, Joakim, et al.
(författare)
-
Coverage segmentation based on linear unmixing and minimization of perimeter and boundary thickness
- 2012
-
Ingår i: Pattern Recognition Letters. - : Elsevier BV. - 0167-8655 .- 1872-7344. ; 33:6, s. 728-738
-
Tidskriftsartikel (refereegranskat)abstract
- We present a method for coverage segmentation, where the, possibly partial, coverage of each image element by each of the image components is estimated. The method combines intensity information with spatial smoothness criteria. A model for linear unmixing of image intensities is enhanced by introducing two additional conditions: (i) minimization of object perimeter, leading to smooth object boundaries, and (ii) minimization of the thickness of the fuzzy object boundary, and to some extent overall image fuzziness, to respond to a natural assumption that imaged objects are crisp, and that fuzziness is mainly due to the imaging and digitization process. The segmentation is formulated as an optimization problem and solved by the Spectral Projected Gradient method. This fast, deterministic optimization method enables practical applicability of the proposed segmentation method. Evaluation on both synthetic and real images confirms very good performance of the algorithm.
|
|
| 45. |
- Lindblad, Joakim, et al.
(författare)
-
De-noising of SRµCT Fiber Images by Total Variation Minimization
- 2010
-
Ingår i: Proceedings of the 20th International Conference on Pattern Recognition (ICPR10). - Istanbul, Turkey. - 1051-4651. - 9781424475421 ; , s. 4621-4624
-
Konferensbidrag (refereegranskat)abstract
- SRμCT images of paper and pulp fiber materials are characterized by a low signal to noise ratio. De-noising is therefore a common preprocessing step before segmentation into fiber and background components. We suggest a de-noising method based on total variation minimization using a modified Spectral Conjugate Gradient algorithm. Quantitative evaluation performed on synthetic 3D data and qualitative evaluation on real 3D paper fiber data confirm appropriateness of the suggested method for the particular application.
|
|
| 46. |
- Lindblad, Joakim, et al.
(författare)
-
Defuzzification by Feature Distance Minimization Based on DC Programming
- 2007
-
Ingår i: 5th International Symposium on Image and Signal Processing and Analysis, 2007. - 9789531841160 ; , s. 373-378
-
Konferensbidrag (refereegranskat)abstract
- We introduce the use of DC programming, in combination with convex-concave regularization, as a deterministic approach for solving the optimization problem imposed by defuzzification by feature distance minimization. We provide a DC based algorithm for finding a solution to the defuzzification problem by expressing the objective function as a difference of two convex functions and iteratively solving a family of DC programs. We compare the performance with the previously recommended method, simulated annealing, on a number of test images. Encouraging results, together with several advantages of the DC based method, approve use of this approach, and motivate its further exploration.
|
|
| 47. |
- Lindblad, Joakim, et al.
(författare)
-
Exact linear time Euclidean distance transforms of grid line sampled shapes
- 2015
-
Ingår i: Mathematical Morphology and its Applications to Signal and Image Processing. - Cham : Springer. - 9783319187198 ; , s. 645-656
-
Konferensbidrag (refereegranskat)abstract
- We propose a method for computing, in linear time, the exact Euclidean distance transform of sets of points s. t. one coordinate of a point can be assigned any real value, whereas other coordinates are restricted to discrete sets of values. The proposed distance transform is applicable to objects represented by grid line sampling, and readily provides sub-pixel precise distance values. The algorithm is easy to implement; we present complete pseudo code. The method is easy to parallelize and extend to higher dimensional data. We present two ways of obtaining approximate grid line sampled representations, and evaluate the proposed EDT on synthetic examples. The method is competitive w. r. t. state-of-the-art methods for sub-pixel precise distance evaluation.
|
|
| 48. |
- Lindblad, Joakim, et al.
(författare)
-
Feature Based Defuzzification at Increased Spatial Resolution
- 2006
-
Ingår i: Combinatorial Image Analysis. - Berlin, Heidelberg : Springer Berlin Heidelberg. - 9783540351535 ; , s. 131-143
-
Konferensbidrag (refereegranskat)abstract
- Defuzzification of fuzzy spatial sets by feature distance minimization, recently proposed as an alternative to crisp segmentation, is studied further. Fully utilizing information available in a fuzzy (discrete) representation of a continuous shape, we present an improved defuzzification method, such that the crisp discrete representation of a fuzzy set is generated at an increased spatial resolution, compared to the resolution of the fuzzy set. The correspondence between a fuzzy and a crisp set is established through a distance between their representations based on selected features, where the different resolutions of the images to compare are taken into account. The performance of the method is tested on both synthetic and real images.
|
|
| 49. |
- Lindblad, Joakim, et al.
(författare)
-
Feature Based Defuzzification in Z² and Z³ Using a Scale Space Approach
- 2006
-
Ingår i: Discrete Geometry for Computer Imagery 13th International Conference, DGCI 2006, Szeged, Hungary, October 25-27, 2006. Proceedings. - Berlin, Heidelberg : Springer Berlin Heidelberg. - 9783540476511 ; , s. 379-390
-
Konferensbidrag (refereegranskat)abstract
- A defuzzification method based on feature distance minimization is further improved by incorporating into the distance function feature values measured on object representations at different scales. It is noticed that such an approach can improve defuzzification results by better preserving the properties of a fuzzy set; area preservation at scales in-between local (pixel-size) and global (the whole object) provides that characteristics of the fuzzy object are more appropriately exhibited in the defuzzification. For the purpose of comparing sets of different resolution, we propose a feature vector representation of a (fuzzy and crisp) set, utilizing a resolution pyramid. The distance measure is accordingly adjusted. The defuzzification method is extended to the 3D case. Illustrative examples are given.
|
|
| 50. |
- Lindblad, Joakim, et al.
(författare)
-
High-resolution reconstruction by feature distance minimization from multiple views of an object
- 2015
-
Ingår i: Proc. 5th International Conference on Image Processing Theory, Tools and Applications. - Piscataway, NJ : IEEE. - 9781479986361 ; , s. 29-34
-
Konferensbidrag (refereegranskat)abstract
- We present a method which utilizes advantages of fuzzy object representations and image processing techniques adjusted to them, to further increase efficient utilization of image information. Starting from a number of low-resolution images of affine transformations of an object, we create its suitably defuzzified high-resolution reconstruction. We evaluate the proposed method on synthetic data, observing its performance w.r.t. noise sensitivity, influence of the number of used low-resolution images, sensitivity to object variation and to inaccurate registration. Our aim is to explore applicability of the method to real image data acquired by Transmission Electron Microscopy, in a biomedical application we are currently working on.
|
|
