Ivica Kopriva
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Ivica Kopriva has a total of 64 co-authors. Here are the top co-authors by number of shared publications:
Ivanka Jerić
Division of Organic Chemistry and Biochemistry, Ruđer Bošković Institute, Bijenička Cesta 54, HR-10000, Zagreb, Croatia
7 shared publicationsMirko Hadžija
Division of Molecular Medicine, Rud̵er Bošković Institute, Bijenička cesta 54, HR-10000 Zagreb, Croatia
5 shared publications- Xinjian Chen
School of Electronic and Information Engineering, Soochow University, Suzhou, China
5 shared publications Lidija Brkljačić
Division of Organic Chemistry and Biochemistry, Ruđer Bošković Institute, Bijenička Cesta 54, HR-10000, Zagreb, Croatia
4 shared publicationsMarijana Popović Hadžija
Division of Molecular Medicine, Rud̵er Bošković Institute, Bijenička cesta 54, HR-10000 Zagreb, Croatia
4 shared publications- Maria Brbić
School of Computer and Communication Sciences, EPFL, Lausanne, Switzerland
3 shared publications
Latest Publications
- Open Access
- 17 February 2026
- Published by Elsevier in SSRN Electronic Journal
Subspace Clustering in Wavelet Packets Domain
Subspace clustering (SC) algorithms leverage the union of subspaces model to group data points according to the subspaces from which they originate. However, raw data is often not easily separable within these subspaces. Therefore, we propose a SC approach in the wavelet packets (WP) domain. The WP transform provides multiple complementary representations, known as subbands. Our method integrates the original data and the subbands into a unified multi-view dataset. Owing to the orthogonality of the WP transform, information about the subspace bases is preserved across WP representations-an advantage not guaranteed when using multi-view representation comprised of handcrafted or learned features such as SIFT, HOG, or features extracted by deep networks. We formulate joint affinity graph learning task as a low-rank Multi-scale Entanglement Renormalization Ansatz (MERA) tensor network approximation problem. The WP-MERA approach is formulated for both grayscale and color images, and its performance is evaluated on six image datasets encompassing faces, digits, and objects. Although WP MERA SC is a linear method, it achieved clustering performance on par with, or surpassing, state-of-the-art deep SC algorithms. For instance, on the color version of the COIL100 dataset, WP-MERA achieved an accuracy of 94.76% outperforming the best deep SC competitor by 22.06%. MATLAB code of the proposed WP domain SC method is available at https://github.com/ikopriva/WPDSC.
- 17 February 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessSubspace Clustering in Wavelet Packets Domain
Subspace clustering (SC) algorithms leverage the union of subspaces model to group data points according to the subspaces from which they originate. However, raw data is often not easily separable within these subspaces. Therefore, we propose a SC approach in the wavelet packets (WP) domain. The WP transform provides multiple complementary representations, known as subbands. Our method integrates the original data and the subbands into a unified multi-view dataset. Owing to the orthogonality of the WP transform, information about the subspace bases is preserved across WP representations-an advantage not guaranteed when using multi-view representation comprised of handcrafted or learned features such as SIFT, HOG, or features extracted by deep networks. We formulate joint affinity graph learning task as a low-rank Multi-scale Entanglement Renormalization Ansatz (MERA) tensor network approximation problem. The WP-MERA approach is formulated for both grayscale and color images, and its performance is evaluated on six image datasets encompassing faces, digits, and objects. Although WP MERA SC is a linear method, it achieved clustering performance on par with, or surpassing, state-of-the-art deep SC algorithms. For instance, on the color version of the COIL100 dataset, WP-MERA achieved an accuracy of 94.76% outperforming the best deep SC competitor by 22.06%. MATLAB code of the proposed WP domain SC method is available at https://github.com/ikopriva/WPDSC.
- 5 December 2025
- Published by Elsevier in Neurocomputing
- Vol. 666
Label-independent hyperparameter-free self-supervised deep subspace clustering
- 5 December 2025
- Published by Elsevier in Neurocomputing
- Vol. 666
Label-independent hyperparameter-free self-supervised deep subspace clustering
- 19 October 2025
- Published by Springer Nature
- p. 159-173
Multilayer Graph Approach to Deep Subspace Clustering
- 19 October 2025
- Published by Springer Nature
- p. 159-173
Multilayer Graph Approach to Deep Subspace Clustering
- Open Access
- 15 October 2025
- Published by Elsevier in Pattern Recognition
- Vol. 172
Pseudo labels approach to interpretable self-guided subspace clustering
- 15 October 2025
- Published by Elsevier in Pattern Recognition
- Vol. 172
Open AccessPseudo labels approach to interpretable self-guided subspace clustering
- Open Access
- 25 July 2025
- Published by Elsevier in Computers in Biology and Medicine
- Vol. 196 (Pt B) , 110841
A hyperspectral imaging dataset and Grassmann manifold method for intraoperative pixel-wise classification of metastatic colon cancer in the liver
- 25 July 2025
- Published by Elsevier in Computers in Biology and Medicine
- Vol. 196 (Pt B) , 110841
Open AccessA hyperspectral imaging dataset and Grassmann manifold method for intraoperative pixel-wise classification of metastatic colon cancer in the liver
- Open Access
Data-Adaptive Low-Rank Sparse Subspace Clustering
Low-rank sparse subspace clustering (LRSSC) algorithms built on self-expressive model effectively capture both the global and local structure of the data. However, existing solutions, primarily based on proximal operators associated with Sp/Lp , p e {0, 1/2, 2/3, 1}, norms are not data-adaptive. In this work, we propose an LRSSC algorithm incorporating a data-adaptive surrogate for the S0/L0 quasi-norm. We provide a numerical solution for the corresponding proximal operator in cases where an analytical expression is unavailable. The proposed LRSSC algorithm is formulated within the proximal mapping framework, and we present theoretical proof of its global convergence toward a stationary point. We evaluate the performance of the proposed method on three well known datasets, comparing it against LRSSC algorithms constrained by Sp/Lp, p e {0, 1/2, 2/3, 1}, norms.
Open AccessData-Adaptive Low-Rank Sparse Subspace Clustering
Low-rank sparse subspace clustering (LRSSC) algorithms built on self-expressive model effectively capture both the global and local structure of the data. However, existing solutions, primarily based on proximal operators associated with Sp/Lp , p e {0, 1/2, 2/3, 1}, norms are not data-adaptive. In this work, we propose an LRSSC algorithm incorporating a data-adaptive surrogate for the S0/L0 quasi-norm. We provide a numerical solution for the corresponding proximal operator in cases where an analytical expression is unavailable. The proposed LRSSC algorithm is formulated within the proximal mapping framework, and we present theoretical proof of its global convergence toward a stationary point. We evaluate the performance of the proposed method on three well known datasets, comparing it against LRSSC algorithms constrained by Sp/Lp, p e {0, 1/2, 2/3, 1}, norms.
- Open Access
A Hyperspectral Imaging Dataset and Methodology for Intraoperative Pixel-Wise Classification of Metastatic Colon Cancer in the Liver
Hyperspectral imaging (HSI) holds significant potential for transforming the field of computational pathology. However, there is currently a shortage of pixel-wise annotated HSI data necessary for training deep learning (DL) models. Additionally, the number of HSI-based research studies remains limited, and in many cases, the advantages of HSI over traditional RGB imaging have not been conclusively demonstrated, particularly for specimens collected intraoperatively. To address these challenges we present a database consisted of 27 HSIs of hematoxylin-eosin stained frozen sections, collected from 14 patients with colon adenocarcinoma metastasized to the liver. It is aimed to validate pixel-wise classification for intraoperative tumor resection. The HSIs were acquired in the spectral range of 450 to 800 nm, with a resolution of 1 nm, resulting in images of 1384x1035 pixels. Pixel-wise annotations were performed by three pathologists. To overcome challenges such as experimental variability and the lack of annotated data, we combined label-propagation-based semi-supervised learning (SSL) with spectral-spatial features extracted by: the multiscale principle of relevant information (MPRI) method and tensor singular spectrum analysis method. Using only 1% of labeled pixels per class the SSL-MPRI method achieved a micro balanced accuracy (BACC) of 0.9313 and a micro F1-score of 0.9235 on the HSI dataset. The performance on corresponding RGB images was lower, with a micro BACC of 0.8809 and a micro F1-score of 0.8688. These improvements are statistically significant. The SSL-MPRI approach outperformed six DL architectures trained with 63% of labeled pixels. Data and code are available at: https://github.com/ikopriva/ColonCancerHSI.
Open AccessA Hyperspectral Imaging Dataset and Methodology for Intraoperative Pixel-Wise Classification of Metastatic Colon Cancer in the Liver
Hyperspectral imaging (HSI) holds significant potential for transforming the field of computational pathology. However, there is currently a shortage of pixel-wise annotated HSI data necessary for training deep learning (DL) models. Additionally, the number of HSI-based research studies remains limited, and in many cases, the advantages of HSI over traditional RGB imaging have not been conclusively demonstrated, particularly for specimens collected intraoperatively. To address these challenges we present a database consisted of 27 HSIs of hematoxylin-eosin stained frozen sections, collected from 14 patients with colon adenocarcinoma metastasized to the liver. It is aimed to validate pixel-wise classification for intraoperative tumor resection. The HSIs were acquired in the spectral range of 450 to 800 nm, with a resolution of 1 nm, resulting in images of 1384x1035 pixels. Pixel-wise annotations were performed by three pathologists. To overcome challenges such as experimental variability and the lack of annotated data, we combined label-propagation-based semi-supervised learning (SSL) with spectral-spatial features extracted by: the multiscale principle of relevant information (MPRI) method and tensor singular spectrum analysis method. Using only 1% of labeled pixels per class the SSL-MPRI method achieved a micro balanced accuracy (BACC) of 0.9313 and a micro F1-score of 0.9235 on the HSI dataset. The performance on corresponding RGB images was lower, with a micro BACC of 0.8809 and a micro F1-score of 0.8688. These improvements are statistically significant. The SSL-MPRI approach outperformed six DL architectures trained with 63% of labeled pixels. Data and code are available at: https://github.com/ikopriva/ColonCancerHSI.
- 8 June 2023
- Published by IOP Publishing in Physics in Medicine & Biology
- Vol. 68 (12) , 125002
Low tensor train and low multilinear rank approximations of 3D tensors for compression and de-speckling of optical coherence tomography images
Objective . Many methods for compression and/or de-speckling of 3D optical coherence tomography (OCT) images operate on a slice-by-slice basis and, consequently, ignore spatial relations between the B-scans. Thus, we develop compression ratio (CR)-constrained low tensor train (TT)—and low multilinear (ML) rank approximations of 3D tensors for compression and de-speckling of 3D OCT images. Due to inherent denoising mechanism of low-rank approximation, compressed image is often even of better quality than the raw image it is based on. Approach . We formulate CR-constrained low rank approximations of 3D tensor as parallel non-convex non-smooth optimization problems implemented by alternating direction method of multipliers of unfolded tensors. In contrast to patch- and sparsity-based OCT image compression methods, proposed approach does not require clean images for dictionary learning, enables CR as high as 60:1, and it is fast. In contrast to deep networks based OCT image compression, proposed approach is training free and does not require any supervised data pre-processing. Main results . Proposed methodology is evaluated on twenty four images of a retina acquired on Topcon 3D OCT-1000 scanner, and twenty images of a retina acquired on Big Vision BV1000 3D OCT scanner. For the first dataset, statistical significance analysis shows that for CR ≤ 35, all low ML rank approximations and Schatten-0 ( S 0 ) norm constrained low TT rank approximation can be useful for machine learning-based diagnostics by using segmented retina layers. Also for CR ≤ 35, S 0 -constrained ML rank approximation and S 0 -constrained low TT rank approximation can be useful for visual inspection-based diagnostics. For the second dataset, statistical significance analysis shows that for CR ≤ 60 all low ML rank approximations as well as S 0 and S 1/2 low TT ranks approximations can be useful for machine learning-based diagnostics by using segmented retina layers. Also, for CR ≤ 60, low ML rank approximations constrained with S p , p ∊ {0, 1/2, 2/3} and one surrogate of S 0 can be useful for visual inspection-based diagnostics. That is also true for low TT rank approximations constrained with S p , p ∊ {0, 1/2, 2/3} for CR ≤ 20. Significance . Studies conducted on datasets acquired by two different types of scanners confirmed capabilities of proposed framework that, for a wide range of CRs, yields de-speckled 3D OCT images suitable for clinical data archiving and remote consultation, for visual inspection-based diagnosis and for machine learning-based diagnosis by using segmented retina layers.
- 8 June 2023
- Published by IOP Publishing in Physics in Medicine & Biology
- Vol. 68 (12) , 125002
Low tensor train and low multilinear rank approximations of 3D tensors for compression and de-speckling of optical coherence tomography images
Objective . Many methods for compression and/or de-speckling of 3D optical coherence tomography (OCT) images operate on a slice-by-slice basis and, consequently, ignore spatial relations between the B-scans. Thus, we develop compression ratio (CR)-constrained low tensor train (TT)—and low multilinear (ML) rank approximations of 3D tensors for compression and de-speckling of 3D OCT images. Due to inherent denoising mechanism of low-rank approximation, compressed image is often even of better quality than the raw image it is based on. Approach . We formulate CR-constrained low rank approximations of 3D tensor as parallel non-convex non-smooth optimization problems implemented by alternating direction method of multipliers of unfolded tensors. In contrast to patch- and sparsity-based OCT image compression methods, proposed approach does not require clean images for dictionary learning, enables CR as high as 60:1, and it is fast. In contrast to deep networks based OCT image compression, proposed approach is training free and does not require any supervised data pre-processing. Main results . Proposed methodology is evaluated on twenty four images of a retina acquired on Topcon 3D OCT-1000 scanner, and twenty images of a retina acquired on Big Vision BV1000 3D OCT scanner. For the first dataset, statistical significance analysis shows that for CR ≤ 35, all low ML rank approximations and Schatten-0 ( S 0 ) norm constrained low TT rank approximation can be useful for machine learning-based diagnostics by using segmented retina layers. Also for CR ≤ 35, S 0 -constrained ML rank approximation and S 0 -constrained low TT rank approximation can be useful for visual inspection-based diagnostics. For the second dataset, statistical significance analysis shows that for CR ≤ 60 all low ML rank approximations as well as S 0 and S 1/2 low TT ranks approximations can be useful for machine learning-based diagnostics by using segmented retina layers. Also, for CR ≤ 60, low ML rank approximations constrained with S p , p ∊ {0, 1/2, 2/3} and one surrogate of S 0 can be useful for visual inspection-based diagnostics. That is also true for low TT rank approximations constrained with S p , p ∊ {0, 1/2, 2/3} for CR ≤ 20. Significance . Studies conducted on datasets acquired by two different types of scanners confirmed capabilities of proposed framework that, for a wide range of CRs, yields de-speckled 3D OCT images suitable for clinical data archiving and remote consultation, for visual inspection-based diagnosis and for machine learning-based diagnosis by using segmented retina layers.
- 7 December 2020
- Published by American Chemical Society (ACS) in Analytical Chemistry
- Vol. 93 (2) , 745-751
Non-negative Least Squares Approach to Quantification of 1H Nuclear Magnetic Resonance Spectra of Human Urine
Because of its quantitative character and capability for high-throughput screening, 1H nuclear magnetic resonance (NMR) spectroscopy is used extensively in the profiling of biofluids such as urine and blood plasma. However, the narrow frequency bandwidth of 1H NMR spectroscopy leads to a severe overlap of the spectra of components present in the complex mixtures such as biofluids. Therefore, 1H NMR-based metabolomics analysis is focused on targeted studies related to concentrations of the small number of metabolites. Here, we propose a library-based approach to quantify proportions of overlapping metabolites from 1H NMR mixture spectra. The method boils down to the linear non-negative least squares (NNLS) problem, whereas proportions of the pure components contained in the library stand for the unknowns. The method is validated on an estimation of the proportions of (i) the 78 pure spectra, presumably related to type 2 diabetes mellitus (T2DM), from their synthetic linear mixture; (ii) metabolites present in 62 1H NMR spectra of urine of subjects with T2DM and 62 1H NMR spectra of urine of control subjects. In both cases, the in-house library of 210 pure component 1H NMR spectra represented the design matrix in the related NNLS problem. The proposed method pinpoints 63 metabolites that in a statistically significant way discriminate the T2DM group from the control group and 46 metabolites discriminating control from the T2DM group. For several T2DM-discriminative metabolites, we prove their presence by independent analytical determination or by pointing out the corresponding findings in the published literature.
- 7 December 2020
- Published by American Chemical Society (ACS) in Analytical Chemistry
- Vol. 93 (2) , 745-751
Non-negative Least Squares Approach to Quantification of 1H Nuclear Magnetic Resonance Spectra of Human Urine
Because of its quantitative character and capability for high-throughput screening, 1H nuclear magnetic resonance (NMR) spectroscopy is used extensively in the profiling of biofluids such as urine and blood plasma. However, the narrow frequency bandwidth of 1H NMR spectroscopy leads to a severe overlap of the spectra of components present in the complex mixtures such as biofluids. Therefore, 1H NMR-based metabolomics analysis is focused on targeted studies related to concentrations of the small number of metabolites. Here, we propose a library-based approach to quantify proportions of overlapping metabolites from 1H NMR mixture spectra. The method boils down to the linear non-negative least squares (NNLS) problem, whereas proportions of the pure components contained in the library stand for the unknowns. The method is validated on an estimation of the proportions of (i) the 78 pure spectra, presumably related to type 2 diabetes mellitus (T2DM), from their synthetic linear mixture; (ii) metabolites present in 62 1H NMR spectra of urine of subjects with T2DM and 62 1H NMR spectra of urine of control subjects. In both cases, the in-house library of 210 pure component 1H NMR spectra represented the design matrix in the related NNLS problem. The proposed method pinpoints 63 metabolites that in a statistically significant way discriminate the T2DM group from the control group and 46 metabolites discriminating control from the T2DM group. For several T2DM-discriminative metabolites, we prove their presence by independent analytical determination or by pointing out the corresponding findings in the published literature.
- Open Access
Robust Self-Supervised Convolutional Neural Network for Subspace Clustering and Classification
Insufficient capability of existing subspace clustering methods to handle data coming from nonlinear manifolds, data corruptions, and out-of-sample data hinders their applicability to address real-world clustering and classification problems. This paper proposes the robust formulation of the self-supervised convolutional subspace clustering network ($S^2$ConvSCN) that incorporates the fully connected (FC) layer and, thus, it is capable for handling out-of-sample data by classifying them using a softmax classifier. $S^2$ConvSCN clusters data coming from nonlinear manifolds by learning the linear self-representation model in the feature space. Robustness to data corruptions is achieved by using the correntropy induced metric (CIM) of the error. Furthermore, the block-diagonal (BD) structure of the representation matrix is enforced explicitly through BD regularization. In a truly unsupervised training environment, Robust $S^2$ConvSCN outperforms its baseline version by a significant amount for both seen and unseen data on four well-known datasets. Arguably, such an ablation study has not been reported before.
Open AccessRobust Self-Supervised Convolutional Neural Network for Subspace Clustering and Classification
Insufficient capability of existing subspace clustering methods to handle data coming from nonlinear manifolds, data corruptions, and out-of-sample data hinders their applicability to address real-world clustering and classification problems. This paper proposes the robust formulation of the self-supervised convolutional subspace clustering network ($S^2$ConvSCN) that incorporates the fully connected (FC) layer and, thus, it is capable for handling out-of-sample data by classifying them using a softmax classifier. $S^2$ConvSCN clusters data coming from nonlinear manifolds by learning the linear self-representation model in the feature space. Robustness to data corruptions is achieved by using the correntropy induced metric (CIM) of the error. Furthermore, the block-diagonal (BD) structure of the representation matrix is enforced explicitly through BD regularization. In a truly unsupervised training environment, Robust $S^2$ConvSCN outperforms its baseline version by a significant amount for both seen and unseen data on four well-known datasets. Arguably, such an ablation study has not been reported before.