Ditte Rask Tornby
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Ditte Rask Tornby has a total of 38 co-authors. Here are the top co-authors by number of shared publications:
- Thomas Emil Andersen
Department of Clinical Microbiology, Odense University Hospital and Research Unit of Clinical Microbiology, University of Southern Denmark, 5000 Odense, Denmark
5 shared publications Isik S. Johansen
University of Southern Denmark Odense Denmark
3 shared publicationsRune M. Pedersen
University of Southern Denmark Odense Denmark
3 shared publications- Carsten Thirstrup
Danish Fundamental Metrology, Kogle Allé 5, 2970 Hørsholm, Denmark
2 shared publications Claus Bistrup
Department of Nephrology, Odense University Hospital and the Nephrology Research Unit, University of Southern Denmark, Odense, Denmark
2 shared publications- Danylo Komisar
Lightnovo ApS, Blokken 11, 1., 3460 Birkerød, Denmark
2 shared publications
Latest Publications
- Open Access
- 4 August 2026
- Published by MDPI AG in Chemosensors
- Vol. 14 (8) , 179
Interpretable Spectral Transformer for Raman-Based Bacterial Identification Across Species and Strains
Raman spectroscopy combined with machine learning offers a rapid, label-free approach for bacterial identification, but robust translation remains challenged by spectral variability, biological heterogeneity, and limited model interpretability. Here, we present an integrated evaluation of an optimized Spectral Transformer (ST) framework for Raman-based bacterial classification benchmarked against a systematically optimized one-dimensional convolutional neural network (1D-CNN). The comparison was performed using a curated 36-class dataset comprising 15 Gram-negative bacterial entries, 15 Gram-positive bacterial entries, one non-bacterial microorganism, and five background/reference classes, enabling evaluation of both species-level and fine-grained bacterial classification. Under 15 dB noise-augmented evaluation, the ST achieved 80.6% ± 0.3% accuracy and a Matthews correlation coefficient (MCC) of 0.801 ± 0.003, outperforming the 1D-CNN baseline with 72.9% ± 0.3% accuracy and an MCC of 0.721 ± 0.003. Integrated Gradients analysis combined with attention map visualization enabled multi-level model interpretation, revealing that the ST’s improved robustness correlates with more bounded attribution patterns during misclassification, whereas the 1D-CNN’s feature attribution becomes scattered under noise perturbation. Importantly, this interpretability-driven analysis identified model-specific failure modes in the baseline architecture, including an over-reliance on non-specific spectral regions under noise, which can inform future data collection strategies and guide refinements to experimental protocols. These results demonstrate that attention-based spectral modeling improves Raman-based bacterial classification under noise-perturbed conditions while enabling multi-level interpretability that bridges model understanding with actionable feedback on experimental design and data quality requirements.
- 4 August 2026
- Published by MDPI AG in Chemosensors
- Vol. 14 (8) , 179
Open AccessInterpretable Spectral Transformer for Raman-Based Bacterial Identification Across Species and Strains
Raman spectroscopy combined with machine learning offers a rapid, label-free approach for bacterial identification, but robust translation remains challenged by spectral variability, biological heterogeneity, and limited model interpretability. Here, we present an integrated evaluation of an optimized Spectral Transformer (ST) framework for Raman-based bacterial classification benchmarked against a systematically optimized one-dimensional convolutional neural network (1D-CNN). The comparison was performed using a curated 36-class dataset comprising 15 Gram-negative bacterial entries, 15 Gram-positive bacterial entries, one non-bacterial microorganism, and five background/reference classes, enabling evaluation of both species-level and fine-grained bacterial classification. Under 15 dB noise-augmented evaluation, the ST achieved 80.6% ± 0.3% accuracy and a Matthews correlation coefficient (MCC) of 0.801 ± 0.003, outperforming the 1D-CNN baseline with 72.9% ± 0.3% accuracy and an MCC of 0.721 ± 0.003. Integrated Gradients analysis combined with attention map visualization enabled multi-level model interpretation, revealing that the ST’s improved robustness correlates with more bounded attribution patterns during misclassification, whereas the 1D-CNN’s feature attribution becomes scattered under noise perturbation. Importantly, this interpretability-driven analysis identified model-specific failure modes in the baseline architecture, including an over-reliance on non-specific spectral regions under noise, which can inform future data collection strategies and guide refinements to experimental protocols. These results demonstrate that attention-based spectral modeling improves Raman-based bacterial classification under noise-perturbed conditions while enabling multi-level interpretability that bridges model understanding with actionable feedback on experimental design and data quality requirements.
- Open Access
- 9 June 2026
- Published by MDPI AG in Preprints.org
Interpretable Spectral Transformer for Raman-Based Bacterial Identification Across Species and Strains
Raman spectroscopy combined with machine learning offers a rapid, label-free approach for bacterial identification, but robust translation remains challenged by spectral variability, biological heterogeneity, and limited model interpretability. Here, we present an integrated evaluation of an optimized Spectral Transformer (ST) framework for Raman-based bacterial classification benchmarked against a systematically optimized one-dimensional convolutional neural network (1D-CNN). The comparison was performed using a curated 36-class dataset comprising 15 Gram-negative bacterial entries, 15 Gram-positive bacterial entries, one non-bacterial microorganism, and five background/reference classes, enabling evaluation of both species-level and fine-grained bacterial classification. Under 15 dB noise-augmented evaluation, the ST achieved 80.6% ± 0.3% accuracy and a Matthews correlation coefficient (MCC) of 0.801 ± 0.003, outperforming the 1D-CNN baseline with 72.9% ± 0.3% accuracy andanMCCof0.721±0.003. Integrated Gradients analysis combined with attention map visualization enabled multi-level model interpretation, revealing that the ST’s improved robustness correlates with more bounded attribution patterns during misclassification, whereas the 1D-CNN’s feature attribution becomes scattered under noise perturbation. Importantly, this interpretability-driven analysis identified model-specific failure modes in the baseline architecture, including an over-reliance on non-specific spectral regions under noise, which can inform future data collection strategies and guide refinements to experimental protocols. These results demonstrate that attention-based spectral modeling improves Raman-based bacterial classification under noise-perturbed conditions while enabling multi-level interpretability that bridges model understanding with actionable feedback on experimental design and data quality requirements.
- 9 June 2026
- Published by MDPI AG in Preprints.org
Open AccessInterpretable Spectral Transformer for Raman-Based Bacterial Identification Across Species and Strains
Raman spectroscopy combined with machine learning offers a rapid, label-free approach for bacterial identification, but robust translation remains challenged by spectral variability, biological heterogeneity, and limited model interpretability. Here, we present an integrated evaluation of an optimized Spectral Transformer (ST) framework for Raman-based bacterial classification benchmarked against a systematically optimized one-dimensional convolutional neural network (1D-CNN). The comparison was performed using a curated 36-class dataset comprising 15 Gram-negative bacterial entries, 15 Gram-positive bacterial entries, one non-bacterial microorganism, and five background/reference classes, enabling evaluation of both species-level and fine-grained bacterial classification. Under 15 dB noise-augmented evaluation, the ST achieved 80.6% ± 0.3% accuracy and a Matthews correlation coefficient (MCC) of 0.801 ± 0.003, outperforming the 1D-CNN baseline with 72.9% ± 0.3% accuracy andanMCCof0.721±0.003. Integrated Gradients analysis combined with attention map visualization enabled multi-level model interpretation, revealing that the ST’s improved robustness correlates with more bounded attribution patterns during misclassification, whereas the 1D-CNN’s feature attribution becomes scattered under noise perturbation. Importantly, this interpretability-driven analysis identified model-specific failure modes in the baseline architecture, including an over-reliance on non-specific spectral regions under noise, which can inform future data collection strategies and guide refinements to experimental protocols. These results demonstrate that attention-based spectral modeling improves Raman-based bacterial classification under noise-perturbed conditions while enabling multi-level interpretability that bridges model understanding with actionable feedback on experimental design and data quality requirements.
- Open Access
An Antibiotic-Loaded Silicone–Hydrogel Interpenetrating Polymer Network for the Prevention of Surgical Site Infections
Surgical site infections (SSIs) are among the most frequent healthcare-associated infections, resulting in high morbidity, mortality, and cost. While correct hygiene measures and prophylactic antibiotics are effective in preventing SSIs, even in modern healthcare settings where recommended guidelines are strictly followed, SSIs persist as a considerable problem that has proven hard to solve. Surgical procedures involving the implantation of foreign bodies are particularly problematic due to the ability of microorganisms to adhere to and colonize the implanted material and form resilient biofilms. In these cases, SSIs may develop even months after implantation and can be difficult to treat once established. Locally applied antibiotics or specifically engineered implant materials with built-in antibiotic-release properties may prevent these complications and, ultimately, require fewer antibiotics compared to those that are systemically administered. In this study, we demonstrated an antimicrobial material concept with intended use in artificial vascular grafts. The material is a silicone–hydrogel interpenetrating polymer network developed earlier for drug-release catheters. In this study, we designed the material for permanent implantation and tested the drug-loading and drug-release properties of the material to prevent the growth of a typical causative pathogen of SSIs, Staphylococcus aureus. The novelty of this study is demonstrated through the antimicrobial properties of the material in vitro after loading it with an advantageous combination, minocycline and rifampicin, which subsequently showed superiority over the state-of-the-art (Propaten) artificial graft material in a large-animal study, using a novel porcine tissue-implantation model.
Open AccessAn Antibiotic-Loaded Silicone–Hydrogel Interpenetrating Polymer Network for the Prevention of Surgical Site Infections
Surgical site infections (SSIs) are among the most frequent healthcare-associated infections, resulting in high morbidity, mortality, and cost. While correct hygiene measures and prophylactic antibiotics are effective in preventing SSIs, even in modern healthcare settings where recommended guidelines are strictly followed, SSIs persist as a considerable problem that has proven hard to solve. Surgical procedures involving the implantation of foreign bodies are particularly problematic due to the ability of microorganisms to adhere to and colonize the implanted material and form resilient biofilms. In these cases, SSIs may develop even months after implantation and can be difficult to treat once established. Locally applied antibiotics or specifically engineered implant materials with built-in antibiotic-release properties may prevent these complications and, ultimately, require fewer antibiotics compared to those that are systemically administered. In this study, we demonstrated an antimicrobial material concept with intended use in artificial vascular grafts. The material is a silicone–hydrogel interpenetrating polymer network developed earlier for drug-release catheters. In this study, we designed the material for permanent implantation and tested the drug-loading and drug-release properties of the material to prevent the growth of a typical causative pathogen of SSIs, Staphylococcus aureus. The novelty of this study is demonstrated through the antimicrobial properties of the material in vitro after loading it with an advantageous combination, minocycline and rifampicin, which subsequently showed superiority over the state-of-the-art (Propaten) artificial graft material in a large-animal study, using a novel porcine tissue-implantation model.
- Open Access
- 1 December 2022
- Published by Centers for Disease Control and Prevention (CDC) in Emerging Infectious Diseases
- Vol. 28 (12) , 2575-2577
Omicron BA.5 Neutralization among Vaccine-Boosted Persons with Prior Omicron BA.1/BA.2 Infections
Omicron BA.5 Neutralization among Vaccine-Boosted Persons with Prior Omicron BA.1/BA.2 Infections
- 1 December 2022
- Published by Centers for Disease Control and Prevention (CDC) in Emerging Infectious Diseases
- Vol. 28 (12) , 2575-2577
Open AccessOmicron BA.5 Neutralization among Vaccine-Boosted Persons with Prior Omicron BA.1/BA.2 Infections
Omicron BA.5 Neutralization among Vaccine-Boosted Persons with Prior Omicron BA.1/BA.2 Infections
- Open Access
- 26 October 2022
- Published by American Society for Microbiology in Microbiology Spectrum
- Vol. 10 (5) , e0131422
Neutralization of SARS-CoV-2 Omicron and Delta Variants in Relation to Vaccine-Induced Antibody Levels in Kidney Transplant Recipients and Healthy Controls
- 26 October 2022
- Published by American Society for Microbiology in Microbiology Spectrum
- Vol. 10 (5) , e0131422
Open AccessNeutralization of SARS-CoV-2 Omicron and Delta Variants in Relation to Vaccine-Induced Antibody Levels in Kidney Transplant Recipients and Healthy Controls
- Open Access
- 19 January 2022
- Published by American Chemical Society (ACS) in Organic Letters
- Vol. 24 (3) , 804-808
Biosynthesis of Calipyridone A Represents a Fungal 2-Pyridone Formation without Ring Expansion in Aspergillus californicus
A chemical investigation of the filamentous fungus Aspergillus californicus led to the isolation of a polyketide–nonribosomal peptide hybrid, calipyridone A (1). A putative biosynthetic gene cluster cpd for production of 1 was next identified by genome mining. The role of the cpd cluster in the production of 1 was confirmed by multiple gene deletion experiments in the host strain as well as by heterologous expression of the hybrid gene cpdA inAspergillus oryzae. Moreover, chemical analyses of the mutant strains allowed the biosynthesis of 1 to be elucidated. The results indicate that the generation of the 2-pyridone moiety of 1 via nucleophilic attack of the iminol nitrogen to the carbonyl carbon is different from the biosynthesis of other fungal 2-pyridone products through P450-catalyzed tetramic acid ring expansions. In addition, two biogenetic intermediates, calipyridones B and C, showed modest inhibition effects on the plaque-forming ability of SARS-CoV-2.
- 19 January 2022
- Published by American Chemical Society (ACS) in Organic Letters
- Vol. 24 (3) , 804-808
Open AccessBiosynthesis of Calipyridone A Represents a Fungal 2-Pyridone Formation without Ring Expansion in Aspergillus californicus
A chemical investigation of the filamentous fungus Aspergillus californicus led to the isolation of a polyketide–nonribosomal peptide hybrid, calipyridone A (1). A putative biosynthetic gene cluster cpd for production of 1 was next identified by genome mining. The role of the cpd cluster in the production of 1 was confirmed by multiple gene deletion experiments in the host strain as well as by heterologous expression of the hybrid gene cpdA inAspergillus oryzae. Moreover, chemical analyses of the mutant strains allowed the biosynthesis of 1 to be elucidated. The results indicate that the generation of the 2-pyridone moiety of 1 via nucleophilic attack of the iminol nitrogen to the carbonyl carbon is different from the biosynthesis of other fungal 2-pyridone products through P450-catalyzed tetramic acid ring expansions. In addition, two biogenetic intermediates, calipyridones B and C, showed modest inhibition effects on the plaque-forming ability of SARS-CoV-2.
- 1 September 2021
- Published by Elsevier in Clinical Microbiology and Infection
- Vol. 27 (9) , 1371-1373
Negative SARS-CoV-2 antibodies, T-cell response and virus neutralization following full vaccination in a renal transplant recipient: a call for vigilance
- 1 September 2021
- Published by Elsevier in Clinical Microbiology and Infection
- Vol. 27 (9) , 1371-1373
Negative SARS-CoV-2 antibodies, T-cell response and virus neutralization following full vaccination in a renal transplant recipient: a call for vigilance