Filip Ciepiela
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
Research Subjects
Filip Ciepiela has a total of 18 co-authors. Here are the top co-authors by number of shared publications:
- Małgorzata Jakubowska
Faculty of Materials Science and Ceramics, AGH University of Krakow, Al. Mickiewicza 30, 30-059 Krakow, Poland
16 shared publications - Bogusław Baś
Department of Analytical Chemistry and Biochemistry, Faculty of Materials Science and Ceramics, AGH University of Krakow, 30-059 Krakow, Poland
5 shared publications - Wladyslaw W. Kubiak
Faculty of Materials Science and Ceramics, AGH University of Science and Technology, al. Mickiewicza 30, 30-059 Krakow, Poland
5 shared publications - Szymon Wójcik
Faculty of Materials Science and Ceramic, AGH University of Krakow, 30 Mickiewicza Av., 30-059 Kraków, POLAND
4 shared publications Wanda Sordoń
Faculty of Materials Science and Ceramics, AGH University of Science and Technology, Mickiewicza 30, 30-059 Kraków, Poland
3 shared publications- Łukasz Górski
Faculty of Materials Science and Ceramics, AGH University of Krakow, Al. Mickiewicza 30, 30-059 Krakow, Poland
3 shared publications
Latest Publications
- Open Access
- 10 July 2025
- Published by MDPI AG in Electronics
- Vol. 14 (14) , 2773
Color-Guided Mixture-of-Experts Conditional GAN for Realistic Biomedical Image Synthesis in Data-Scarce Diagnostics
Background: Limited availability of high-quality labeled biomedical image datasets presents a significant challenge for training deep learning models in medical diagnostics. This study proposes a novel image generation framework combining conditional generative adversarial networks (cGANs) with a Mixture-of-Experts (MoE) architecture and color histogram-aware loss functions to enhance synthetic blood cell image quality. Methods: RGB microscopic images from the BloodMNIST dataset (eight blood cell types, resolution 3 × 128 × 128) underwent preprocessing with k-means clustering to extract the dominant colors and UMAP for visualizing class similarity. Spearman correlation-based distance matrices were used to evaluate the discriminative power of each RGB channel. A MoE–cGAN architecture was developed with residual blocks and LeakyReLU activations. Expert generators were conditioned on cell type, and the generator’s loss was augmented with a Wasserstein distance-based term comparing red and green channel histograms, which were found most relevant for class separation. Results: The red and green channels contributed most to class discrimination; the blue channel had minimal impact. The proposed model achieved 0.97 classification accuracy on generated images (ResNet50), with 0.96 precision, 0.97 recall, and a 0.96 F1-score. The best Fréchet Inception Distance (FID) was 52.1. Misclassifications occurred mainly among visually similar cell types. Conclusions: Integrating histogram alignment into the MoE–cGAN training significantly improves the realism and class-specific variability of synthetic images, supporting robust model development under data scarcity in hematological imaging.
- 10 July 2025
- Published by MDPI AG in Electronics
- Vol. 14 (14) , 2773
Open AccessColor-Guided Mixture-of-Experts Conditional GAN for Realistic Biomedical Image Synthesis in Data-Scarce Diagnostics
Background: Limited availability of high-quality labeled biomedical image datasets presents a significant challenge for training deep learning models in medical diagnostics. This study proposes a novel image generation framework combining conditional generative adversarial networks (cGANs) with a Mixture-of-Experts (MoE) architecture and color histogram-aware loss functions to enhance synthetic blood cell image quality. Methods: RGB microscopic images from the BloodMNIST dataset (eight blood cell types, resolution 3 × 128 × 128) underwent preprocessing with k-means clustering to extract the dominant colors and UMAP for visualizing class similarity. Spearman correlation-based distance matrices were used to evaluate the discriminative power of each RGB channel. A MoE–cGAN architecture was developed with residual blocks and LeakyReLU activations. Expert generators were conditioned on cell type, and the generator’s loss was augmented with a Wasserstein distance-based term comparing red and green channel histograms, which were found most relevant for class separation. Results: The red and green channels contributed most to class discrimination; the blue channel had minimal impact. The proposed model achieved 0.97 classification accuracy on generated images (ResNet50), with 0.96 precision, 0.97 recall, and a 0.96 F1-score. The best Fréchet Inception Distance (FID) was 52.1. Misclassifications occurred mainly among visually similar cell types. Conclusions: Integrating histogram alignment into the MoE–cGAN training significantly improves the realism and class-specific variability of synthetic images, supporting robust model development under data scarcity in hematological imaging.
- 13 March 2025
- Published by Polish Chemical Society in Wiadomości Chemiczne
- Vol. 79 (1) , 39-58
ZASTOSOWANIE METOD CHEMOMETRYCZNYCH DO WOLTAMPEROMETRYCZNEGO PROFILOWANIA ŻYWNOŚCI
The development of society and the rise in consumer awareness generate the need for rapid and precise analysis of food products to ensure their high quality. Consumers are increasingly paying attention to the origin, composition, and production processes of food. Heightened competition among producers and rising production costs lead to serious challenges, such as food product adulteration to lower prices and maximize profits. This study presents results of analysis of a wide range of food products, including wines, whiskies, apple juices, honey, isotonic drinks, and plant-based milks. The research focused on profiling these samples and identifying potential issues, such as the presence of undesirable additives. The registration of voltammetric profiles was made possible through the use of modern working electrodes. As a result of the conducted research, several innovative methods were developed, combining chemical analysis and machine learning strategies that effectively address the identified research problems. Research plans were created to detect adulteration in apple juices and honeys, and predictive models regarding the aging of young wines were defined. Procedures for profiling products such as wines, whiskies, honeys, isotonic drinks, and plant-based milks were also designed. For each of these procedures, a multi-stage optimization of voltammetric profile registration parameters was carried out, and the obtained data underwent advanced signal processing to ensure the highest possible quality of results.
- 13 March 2025
- Published by Polish Chemical Society in Wiadomości Chemiczne
- Vol. 79 (1) , 39-58
ZASTOSOWANIE METOD CHEMOMETRYCZNYCH DO WOLTAMPEROMETRYCZNEGO PROFILOWANIA ŻYWNOŚCI
The development of society and the rise in consumer awareness generate the need for rapid and precise analysis of food products to ensure their high quality. Consumers are increasingly paying attention to the origin, composition, and production processes of food. Heightened competition among producers and rising production costs lead to serious challenges, such as food product adulteration to lower prices and maximize profits. This study presents results of analysis of a wide range of food products, including wines, whiskies, apple juices, honey, isotonic drinks, and plant-based milks. The research focused on profiling these samples and identifying potential issues, such as the presence of undesirable additives. The registration of voltammetric profiles was made possible through the use of modern working electrodes. As a result of the conducted research, several innovative methods were developed, combining chemical analysis and machine learning strategies that effectively address the identified research problems. Research plans were created to detect adulteration in apple juices and honeys, and predictive models regarding the aging of young wines were defined. Procedures for profiling products such as wines, whiskies, honeys, isotonic drinks, and plant-based milks were also designed. For each of these procedures, a multi-stage optimization of voltammetric profile registration parameters was carried out, and the obtained data underwent advanced signal processing to ensure the highest possible quality of results.
- 4 March 2025
- Published by Elsevier in Expert Systems with Applications
- Vol. 276
Deep neural networks for separation of overlapping voltammetric signals
- 4 March 2025
- Published by Elsevier in Expert Systems with Applications
- Vol. 276
Deep neural networks for separation of overlapping voltammetric signals
- Open Access
- 28 September 2023
- Published by MDPI AG in Chemosensors
- Vol. 11 (10) , 513
The Application of Picein Wax Carbon Composite Electrode for Plant-Based Milk Profiling
This work demonstrates the successful application of the picein wax carbon composite electrode (PWCCE) for profiling both commercial and homemade plant milks. Picein wax was utilized as an unconventional binder. The resulting electrode paste exhibited a solidified and hard texture, enabling its use in a manner analogous to that of the glassy carbon electrode. Differential pulse voltammetry (DPV) with an automated measurement and recording procedure was employed to obtain plant-based milk profiles. The utilization of operator-independent measurement procedures yielded high-quality electrochemical fingerprints suitable for subsequent calculations. To interpret the data, unsupervised machine learning methods were implemented, such as principal component analysis (PCA) and cluster analysis. These chemometric techniques confirmed the electrode effectiveness of the construction for this type of research. Moreover, they proved valuable in distinguishing between plant-based milk and cow’s milk, including two different variants: whole milk and lactose-free milk.
- 28 September 2023
- Published by MDPI AG in Chemosensors
- Vol. 11 (10) , 513
Open AccessThe Application of Picein Wax Carbon Composite Electrode for Plant-Based Milk Profiling
This work demonstrates the successful application of the picein wax carbon composite electrode (PWCCE) for profiling both commercial and homemade plant milks. Picein wax was utilized as an unconventional binder. The resulting electrode paste exhibited a solidified and hard texture, enabling its use in a manner analogous to that of the glassy carbon electrode. Differential pulse voltammetry (DPV) with an automated measurement and recording procedure was employed to obtain plant-based milk profiles. The utilization of operator-independent measurement procedures yielded high-quality electrochemical fingerprints suitable for subsequent calculations. To interpret the data, unsupervised machine learning methods were implemented, such as principal component analysis (PCA) and cluster analysis. These chemometric techniques confirmed the electrode effectiveness of the construction for this type of research. Moreover, they proved valuable in distinguishing between plant-based milk and cow’s milk, including two different variants: whole milk and lactose-free milk.
- 5 August 2017
- Published by The Electrochemical Society in Journal of the Electrochemical Society
- Vol. 164 (12) , H760-H769
Faradaic and Capacitive Current Estimation by DPV-ATLD
In this work we for the first time proposed an approach based on ATLD which enabled exploration of the faradaic and double-layer charging current components in voltammetric experiments in the presence of surface-active substances. For simulated data ability to successful decomposition of the signals was confirmed and theoretic shape of two original components was restored. Next the proposed approach was applied in determination of Pb(II) in the presence of the increasing CTAB concentration. Considering calibration set of Pb(II), sampled with frequency of minimum 1 kHz, it was proved that total current may be effectively separated into two components. Capacitive current in the case without CTAB addition was ca. 15% higher in comparison with the case when 2 or 4 mgL−1 of CTAB was present, but decreased faster, and after 20 ms it reached the same value. Faradaic current decreased significantly with CTAB addition but the linear relation with depolarizer concentration was preserved. The calibration models were defined omitting the separated capacitive current component, what improved analytical parameters of the method. Using sampling period of 3 to 6 ms from the potential change, the sensitivity increased 2.7-3 times (depending on the CTAB concentration) in comparison with the typical interpretation, detection limit was also lower and reached 0.2–0.4 μgL−1 of Pb(II). For each concentration of CTAB the correlation coefficient r in lead determination was greater than 0.9970, while the repeatability expressed by the coefficient of variability was lower than 3%.
- 5 August 2017
- Published by The Electrochemical Society in Journal of the Electrochemical Society
- Vol. 164 (12) , H760-H769
Faradaic and Capacitive Current Estimation by DPV-ATLD
In this work we for the first time proposed an approach based on ATLD which enabled exploration of the faradaic and double-layer charging current components in voltammetric experiments in the presence of surface-active substances. For simulated data ability to successful decomposition of the signals was confirmed and theoretic shape of two original components was restored. Next the proposed approach was applied in determination of Pb(II) in the presence of the increasing CTAB concentration. Considering calibration set of Pb(II), sampled with frequency of minimum 1 kHz, it was proved that total current may be effectively separated into two components. Capacitive current in the case without CTAB addition was ca. 15% higher in comparison with the case when 2 or 4 mgL−1 of CTAB was present, but decreased faster, and after 20 ms it reached the same value. Faradaic current decreased significantly with CTAB addition but the linear relation with depolarizer concentration was preserved. The calibration models were defined omitting the separated capacitive current component, what improved analytical parameters of the method. Using sampling period of 3 to 6 ms from the potential change, the sensitivity increased 2.7-3 times (depending on the CTAB concentration) in comparison with the typical interpretation, detection limit was also lower and reached 0.2–0.4 μgL−1 of Pb(II). For each concentration of CTAB the correlation coefficient r in lead determination was greater than 0.9970, while the repeatability expressed by the coefficient of variability was lower than 3%.
- 22 March 2017
- Published by Royal Society of Chemistry (RSC) in The Analyst
- Vol. 142 (10) , 1729-1734
Novel method for standard addition signal analysis in voltammetry
A new procedure for standard addition analysis in voltammetry based on the slope of the peak at the inflection point (SSAA) is proposed.
- 22 March 2017
- Published by Royal Society of Chemistry (RSC) in The Analyst
- Vol. 142 (10) , 1729-1734
Novel method for standard addition signal analysis in voltammetry
A new procedure for standard addition analysis in voltammetry based on the slope of the peak at the inflection point (SSAA) is proposed.
- 1 July 2016
- Published by Elsevier in Food Chemistry
- Vol. 203, 476-482
Unsupervised pattern recognition methods in ciders profiling based on GCE voltammetric signals
- 1 July 2016
- Published by Elsevier in Food Chemistry
- Vol. 203, 476-482
Unsupervised pattern recognition methods in ciders profiling based on GCE voltammetric signals