Tomaž Vrtovec
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Tomaž Vrtovec has a total of 97 co-authors. Here are the top co-authors by number of shared publications:
Franjo Pernuš
16 shared publicationsBoštjan Likar
University of Ljubljana
13 shared publications- Boštjan Likar
Faculty of Electrical Engineering, University of Ljubljana, Ljubljana, Slovenia
12 shared publications - Franjo Pernuš
Faculty of Electrical Engineering, University of Ljubljana, Ljubljana, Slovenia
12 shared publications - Bulat Ibragimov
University of Copenhagen
10 shared publications - Gašper Podobnik
University of Ljubljana
8 shared publications
Latest Publications
- 25 September 2026
- Published by Springer Nature
- p. 549-559
Graph Neural Networks for Morphometric Characterization of the Aortic Valve
- 25 September 2026
- Published by Springer Nature
- p. 549-559
Graph Neural Networks for Morphometric Characterization of the Aortic Valve
- Open Access
- 28 February 2026
- Published by Slovenian Medical Association in Slovenian Medical Journal
- Vol. 95 (1-2) , 3-14
Vrednotenje sintetičnih računalniškotomografskih slik za načrtovanje radioterapije glave in vratu z izključno uporabo magnetne resonance
Izhodišča: Koncept radioterapije z izključno uporabo magnetne resonance (MR) temelji na opustitvi računalniškotomografskih (CT) slik in njihovem nadomeščanju s sintetičnimi CT (sCT) slikami. V tem članku predstavljamo ustvarjanje sCT slik na podlagi MR slik glave in vratu ter pripadajočo geometrijsko in dozimetrično analizo segmentacij kritičnih organov v CT in sCT slikah. Metode: Z algoritmom globokega učenja, ki temelji na difuzijskem procesu, smo ustvarili model na 42 parih ter ga nato uporabili na 12 parih CT in MR slik za ustvarjanje pripadajočih sCT slik iz naključne slike šuma ter s pogojevanjem z MR sliko. V CT in sCT slikah smo nato z modelom globokega učenja avtomatsko segmentirali 12 različnih kritičnih organov ter opravili geometrijsko (računanje prostorskega ujemanja pridobljenih segmentacij v slikah) in dozimetrično (računanje prejetih odmerkov obsevanja pridobljenih segmentacij iz pripadajočih načrtov porazdelitve obsevalnih odmerkov) vrednotenje. Rezultati: Ustvarjene sCT slike so po kakovosti primerljive z obstoječimi metodami: srednja absolutna napaka 43,4 HU, vršno razmerje signal-šum 30,6 dB in indeks strukturne podobnosti 91,2 %. Geometrijska analiza segmentacij kritičnih organov je pokazala visoko ujemanje med CT in sCT slikami: koeficient podobnosti Dice 87,9 % in 95. percentil Hausdorffove razdalje 2,2 mm. Ravno tako je dozimetrična analiza pokazala majhne razlike med obsevalnimi odmerki za CT in sCT slike: relativna razlika povprečnega odmerka je 2,3 % oz. največjega odmerka 1,9 %. Zaključek: Ustvarjanje sCT slik na podlagi predstavljene metodologije daje spodbudne rezultate, ki kažejo, da so sCT slike ustrezna alternativa originalnim CT slikam z vidika geometrijske in dozimetrične analize pripadajočih segmentacij kritičnih organov.
- 28 February 2026
- Published by Slovenian Medical Association in Slovenian Medical Journal
- Vol. 95 (1-2) , 3-14
Open AccessVrednotenje sintetičnih računalniškotomografskih slik za načrtovanje radioterapije glave in vratu z izključno uporabo magnetne resonance
Izhodišča: Koncept radioterapije z izključno uporabo magnetne resonance (MR) temelji na opustitvi računalniškotomografskih (CT) slik in njihovem nadomeščanju s sintetičnimi CT (sCT) slikami. V tem članku predstavljamo ustvarjanje sCT slik na podlagi MR slik glave in vratu ter pripadajočo geometrijsko in dozimetrično analizo segmentacij kritičnih organov v CT in sCT slikah. Metode: Z algoritmom globokega učenja, ki temelji na difuzijskem procesu, smo ustvarili model na 42 parih ter ga nato uporabili na 12 parih CT in MR slik za ustvarjanje pripadajočih sCT slik iz naključne slike šuma ter s pogojevanjem z MR sliko. V CT in sCT slikah smo nato z modelom globokega učenja avtomatsko segmentirali 12 različnih kritičnih organov ter opravili geometrijsko (računanje prostorskega ujemanja pridobljenih segmentacij v slikah) in dozimetrično (računanje prejetih odmerkov obsevanja pridobljenih segmentacij iz pripadajočih načrtov porazdelitve obsevalnih odmerkov) vrednotenje. Rezultati: Ustvarjene sCT slike so po kakovosti primerljive z obstoječimi metodami: srednja absolutna napaka 43,4 HU, vršno razmerje signal-šum 30,6 dB in indeks strukturne podobnosti 91,2 %. Geometrijska analiza segmentacij kritičnih organov je pokazala visoko ujemanje med CT in sCT slikami: koeficient podobnosti Dice 87,9 % in 95. percentil Hausdorffove razdalje 2,2 mm. Ravno tako je dozimetrična analiza pokazala majhne razlike med obsevalnimi odmerki za CT in sCT slike: relativna razlika povprečnega odmerka je 2,3 % oz. največjega odmerka 1,9 %. Zaključek: Ustvarjanje sCT slik na podlagi predstavljene metodologije daje spodbudne rezultate, ki kažejo, da so sCT slike ustrezna alternativa originalnim CT slikam z vidika geometrijske in dozimetrične analize pripadajočih segmentacij kritičnih organov.
- 1 February 2026
- Published by SPIE-Intl Soc Optical Eng in Journal of Medical Imaging
- Vol. 13 (S1) , 1-4
Introduction to the JMI Special Issue on Generative AI for Medical Imaging
The editorial introduces the JMI Special Issue on Generative AI for Medical Imaging.
- 1 February 2026
- Published by SPIE-Intl Soc Optical Eng in Journal of Medical Imaging
- Vol. 13 (S1) , 1-4
Introduction to the JMI Special Issue on Generative AI for Medical Imaging
The editorial introduces the JMI Special Issue on Generative AI for Medical Imaging.
- Open Access
- 23 January 2026
- Published by Elsevier in Radiotherapy and Oncology
- Vol. 217, 111387
Dosimetric assessment of deep learning based organ-at-risk segmentation: insights from the HaN-Seg challenge
- 23 January 2026
- Published by Elsevier in Radiotherapy and Oncology
- Vol. 217, 111387
Open AccessDosimetric assessment of deep learning based organ-at-risk segmentation: insights from the HaN-Seg challenge
- 2 January 2026
- Published by Springer Nature
- p. 93-102
Transformer-Based Instance Detection in 3D Medical Images
- 2 January 2026
- Published by Springer Nature
- p. 93-102
Transformer-Based Instance Detection in 3D Medical Images
- Open Access
- 4 November 2025
- Published by Springer Nature in Scientific Reports
- Vol. 15 (1) , 1-3
Correction: Geometric, dosimetric and psychometric evaluation of three commercial AI software solutions for OAR auto-segmentation in head and neck radiotherapy
- 4 November 2025
- Published by Springer Nature in Scientific Reports
- Vol. 15 (1) , 1-3
Open AccessCorrection: Geometric, dosimetric and psychometric evaluation of three commercial AI software solutions for OAR auto-segmentation in head and neck radiotherapy
- 5 October 2025
- Published by Springer Nature
- p. 129-139
AnatomyGen: Generating Anatomically Plausible Human Phantoms at High Resolution
- 5 October 2025
- Published by Springer Nature
- p. 129-139
AnatomyGen: Generating Anatomically Plausible Human Phantoms at High Resolution
- Open Access
- 29 September 2025
- Published by Springer Nature in Scientific Reports
- Vol. 15 (1) , 1-16
Geometric, dosimetric and psychometric evaluation of three commercial AI software solutions for OAR auto-segmentation in head and neck radiotherapy
Contouring organs-at-risk (OARs) is a critical yet time-consuming step in head and neck (HaN) radiotherapy planning. Auto-segmentation methods have been widely studied, and commercial solutions are increasingly entering clinical use. However, their adoption warrants a comprehensive, multi-perspective evaluation. The purpose of this study is to compare three commercial artificial intelligence (AI) software solutions (Limbus, MIM and MVision) for HaN OAR auto-segmentation on a cohort of 10 computed tomography images with reference contours obtained from the public HaN-Seg dataset, from both observational (descriptive and empirical) and analytical (geometric, dosimetric and psychometric) perspectives. The observational evaluation included vendor questionnaires on technical specifications and radiographer feedback on usability. The analytical evaluation covered geometric (Dice similarity coefficient, DSC, and 95th percentile Hausdorff distance, HD95), dosimetric (dose constraint compliance, OAR priority-based analysis), and psychometric (5-point Likert scale) assessments. All software solutions covered a broad range of OARs. Overall geometric performance differences were relatively small (Limbus: 69.7% DSC, 5.0 mm HD95; MIM: 69.2% DSC, 5.6 mm HD95; MVision: 66.7% DSC, 5.3 mm HD95), however, statistically significant differences were observed for smaller structures such as the cochleae, optic chiasm, and pituitary and thyroid glands. Differences in dosimetric compliance were overall minor, with the lowest compliance observed for the oral cavity and submandibular glands. In terms of qualitative assessment, radiographers gave the highest average Likert rating to Limbus (3.9), followed by MVision (3.7) and MIM (3.5). With few exceptions, most software solutions produced good-quality AI-generated contours (Likert ratings ≥ 3), yet some editing should still be performed to reach clinical acceptability. Notable discrepancies were seen for the optic chiasm and in cases affected by mouth bites or dental artifacts. Importantly, no clear relationship emerged between geometric, dosimetric, and psychometric metrics, underscoring the need for a multi-perspective evaluation without shortcuts.
- 29 September 2025
- Published by Springer Nature in Scientific Reports
- Vol. 15 (1) , 1-16
Open AccessGeometric, dosimetric and psychometric evaluation of three commercial AI software solutions for OAR auto-segmentation in head and neck radiotherapy
Contouring organs-at-risk (OARs) is a critical yet time-consuming step in head and neck (HaN) radiotherapy planning. Auto-segmentation methods have been widely studied, and commercial solutions are increasingly entering clinical use. However, their adoption warrants a comprehensive, multi-perspective evaluation. The purpose of this study is to compare three commercial artificial intelligence (AI) software solutions (Limbus, MIM and MVision) for HaN OAR auto-segmentation on a cohort of 10 computed tomography images with reference contours obtained from the public HaN-Seg dataset, from both observational (descriptive and empirical) and analytical (geometric, dosimetric and psychometric) perspectives. The observational evaluation included vendor questionnaires on technical specifications and radiographer feedback on usability. The analytical evaluation covered geometric (Dice similarity coefficient, DSC, and 95th percentile Hausdorff distance, HD95), dosimetric (dose constraint compliance, OAR priority-based analysis), and psychometric (5-point Likert scale) assessments. All software solutions covered a broad range of OARs. Overall geometric performance differences were relatively small (Limbus: 69.7% DSC, 5.0 mm HD95; MIM: 69.2% DSC, 5.6 mm HD95; MVision: 66.7% DSC, 5.3 mm HD95), however, statistically significant differences were observed for smaller structures such as the cochleae, optic chiasm, and pituitary and thyroid glands. Differences in dosimetric compliance were overall minor, with the lowest compliance observed for the oral cavity and submandibular glands. In terms of qualitative assessment, radiographers gave the highest average Likert rating to Limbus (3.9), followed by MVision (3.7) and MIM (3.5). With few exceptions, most software solutions produced good-quality AI-generated contours (Likert ratings ≥ 3), yet some editing should still be performed to reach clinical acceptability. Notable discrepancies were seen for the optic chiasm and in cases affected by mouth bites or dental artifacts. Importantly, no clear relationship emerged between geometric, dosimetric, and psychometric metrics, underscoring the need for a multi-perspective evaluation without shortcuts.
- Open Access
- 20 March 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Biomedical and Health Informatics
- Vol. 29 (7) , 4981-4990
DentAssignNet: Assignment Network for Dental Cast Labeling in the Presence of Dental Abnormalities
This study focuses on the challenging problem of labeling a collection of objects with inherent morphological and positional dependencies, where instances may be missing or duplicated. We integrate principles of assignment theory in the design of a convolutional neural network to find the optimal label set given pairwise geometrical features extracted from the candidate objects. The objective function aims to minimize the distance between the one-hot encoded labels of the objects and the scores produced by the model, with added emphasis on the scores corresponding to the optimal assignment plan. We tested our solution in the dental domain on the task of finding the teeth labels given a set of candidate instances. The study database included 1200 dental casts of upper and lower jaws from 600 patients. The model reached identification accuracies of 0.952 and 0.968 for the lower and upper jaws, respectively. Moreover, we presented a solution for generating teeth candidates using a multi-step pipeline consisting of coarse and fine segmentations. The algorithm was tested on a database consisting of 600 dental casts, reaching an F1 score of 0.968.
- 20 March 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Biomedical and Health Informatics
- Vol. 29 (7) , 4981-4990
Open AccessDentAssignNet: Assignment Network for Dental Cast Labeling in the Presence of Dental Abnormalities
This study focuses on the challenging problem of labeling a collection of objects with inherent morphological and positional dependencies, where instances may be missing or duplicated. We integrate principles of assignment theory in the design of a convolutional neural network to find the optimal label set given pairwise geometrical features extracted from the candidate objects. The objective function aims to minimize the distance between the one-hot encoded labels of the objects and the scores produced by the model, with added emphasis on the scores corresponding to the optimal assignment plan. We tested our solution in the dental domain on the task of finding the teeth labels given a set of candidate instances. The study database included 1200 dental casts of upper and lower jaws from 600 patients. The model reached identification accuracies of 0.952 and 0.968 for the lower and upper jaws, respectively. Moreover, we presented a solution for generating teeth candidates using a multi-step pipeline consisting of coarse and fine segmentations. The algorithm was tested on a database consisting of 600 dental casts, reaching an F1 score of 0.968.
- Open Access
- 7 March 2025
- Published by University of Maribor
- p. 41-54
Generiranje sintetičnih CT slik iz MR slik področja glave in vratu z uporabo difuzijskih modelov
Računalniška tomografija (CT) je slikovna preiskava, ki se v klinični praksi standardno zajame v okviru načrtovanje radioterapije. V primeru raka območju glave in vratu (HaN) se pogosto zajame tudi magnetno resonančne (MR) slike za natančnejše orisovanje tumorjev in kritičnih organov. V zadnjem času se vse bolj uveljavlja radioterapija na podlagi MR-samostojnega pristopa, ki odstrani potrebo po zajemu CT slik in s tem izpostavljenost ionizirajočemu sevanju, vendar pa zahteva rešitev za generiranje sintetičnih CT slik na podlagi MR slik . Nedavne študije kažejo, da difuzijski modeli nudijo realistično generiranje slik z natančnimi anatomskimi podrobnostmi in manj artefakti kot generativne nasprotniške mreže. V tej študiji smo razvili difuzijski model za pretvorbo MR slik v sintetične CT slike za HaN področje. Naš pristop, ovrednoten na zbirki podatkov HaN-Seg, ki vključuje pare CT in MR slik istih bolnikov, doseže indeks strukturne podobnosti 92,2 %, vršno razmerje signal-šum 33,1 dB ter povprečno absolutno napako 35,3 HU. Model dodatno ovrednotimo tudi za segmentacijo kritičnih organov. Rezultati potrjujejo potencial uporabe difuzijskih modelov pri načrtovanju radioterapije.
- 7 March 2025
- Published by University of Maribor
- p. 41-54
Open AccessGeneriranje sintetičnih CT slik iz MR slik področja glave in vratu z uporabo difuzijskih modelov
Računalniška tomografija (CT) je slikovna preiskava, ki se v klinični praksi standardno zajame v okviru načrtovanje radioterapije. V primeru raka območju glave in vratu (HaN) se pogosto zajame tudi magnetno resonančne (MR) slike za natančnejše orisovanje tumorjev in kritičnih organov. V zadnjem času se vse bolj uveljavlja radioterapija na podlagi MR-samostojnega pristopa, ki odstrani potrebo po zajemu CT slik in s tem izpostavljenost ionizirajočemu sevanju, vendar pa zahteva rešitev za generiranje sintetičnih CT slik na podlagi MR slik . Nedavne študije kažejo, da difuzijski modeli nudijo realistično generiranje slik z natančnimi anatomskimi podrobnostmi in manj artefakti kot generativne nasprotniške mreže. V tej študiji smo razvili difuzijski model za pretvorbo MR slik v sintetične CT slike za HaN področje. Naš pristop, ovrednoten na zbirki podatkov HaN-Seg, ki vključuje pare CT in MR slik istih bolnikov, doseže indeks strukturne podobnosti 92,2 %, vršno razmerje signal-šum 33,1 dB ter povprečno absolutno napako 35,3 HU. Model dodatno ovrednotimo tudi za segmentacijo kritičnih organov. Rezultati potrjujejo potencial uporabe difuzijskih modelov pri načrtovanju radioterapije.