PTPatrick Thevissen

Patrick Thevissen

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Patrick Thevissen has a total of 52 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Top Cited Papers

    Attention Maps in 3D Shape Classification for Dental Stage Estimation with Class Node Graph Attention Networks

    by Barkin Buyukcakir, Rocharles Cavalcante Fontenele, Reinhilde Jacobs, Jannick De Tobel, Patrick Thevissen, Dirk Vandermeulen, Peter Claes


  • Minimal age principle versus Bayesian approach to combine age indicators from magnetic resonance imaging for multifactorial forensic age estimation

    by Heleen Coreelman, Jannick De Tobel, Thomas Widek, Martin Urschler, Steffen Fieuws, Patrick W. Thevissen, Koenraad L. Verstraete

    Multifactorial age estimation is preferred over methods based on a single anatomical site. The main challenge of the multifactorial methods lies in calculating the overall prediction interval. This study compared the performance of two approaches to achieve this: the minimal age principle versus a Bayesian approach. MRI of the third molars, left hand/wrist, and sternal extremity of both clavicles were prospectively conducted in 335 healthy Austrian Caucasian males aged 13–24 years. Development was staged according to De Tobel et al. Multi‐factorial age estimation: A Bayesian approach combining dental and skeletal magnetic resonance imaging. Forensic Sci Int. 2020;306:110054. Applying the minimal age principle rendered a mean absolute error of 1.47 years, root mean square error of 1.81 years, mean width of the 95% prediction interval (PI) of 4.44 ± 2.49 years, and coverage of 68.7%. For the Bayesian approach, the results were 1.41, 1.80, 5.15 ± 1.94 years, and 81.5%, respectively. Higher inconsistency between the different age indicators was linked to a lower coverage probability in the minimal age principle, but not in the Bayesian approach. Moreover, higher inconsistency between age indicators was also linked to a higher probability of obtaining an impossible PI with the minimal age principle. Furthermore, applying the minimal age principle rendered 97.9%/81.0% correctly categorized adults (based on the point prediction of age/based on the PI) and 69.2%/85.6% correctly categorized minors. For the Bayesian approach, the results were 95.2%/76.2% and 81.5%/95.9%, respectively. In conclusion, the Bayesian approach outperformed the minimal age principle for multifactorial forensic age estimation, allowing the construction of more appropriate PIs and more correctly categorized minors.


  • Magnetic resonance imaging for multi-factorial age estimation using Bayes’ rule: a validation study in two independent samples

    by Heleen Coreelman, Jannick De Tobel, Thomas Widek, Martin Urschler, Steffen Fieuws, Patrick Werner Thevissen, Koenraad Luc Verstraete

    Objectives Multi-factorial age estimation (MFA) models have been developed based on Bayes’ rule, combining MRI data of the third molars (T), left wrist (W) and/or both clavicles (C). Internal cross-validated performance was reported, but external validation is needed before bringing the approach into practice. This study aimed to validate these MFA models in two independent samples. Methods In the Ghent sample, W + C MRI was prospectively conducted in 108 healthy Caucasian volunteers (52 males, 56 females) aged 16 to 21 years. In the Graz sample, T + W + C MRI was prospectively conducted in 335 healthy Caucasian males aged 13 to 24 years. Development was staged and checked for intra-observer reliability, and age estimation performances were tested. Results Staging clavicles was most prone to intra-observer variability. Applying the W + C model to Ghent males rendered a mean absolute error of 1.55 years, a root mean square error of 1.90 years, 70.6% correctly categorised adults and 94.4% correctly categorised minors. In females, the results were 1.49 years, 1.83 years, 92.1% and 66.7%, respectively. Regarding the Graz sample, the W + C results were 1.66 years, 2.08 years, 94.7% and 80.7%, respectively. For the T + W + C model, the results were 1.41 years, 1.80 years, 95.2% and 81.5%, respectively. Conclusion The T + W + C and W + C models rendered a similar accuracy of the point prediction of age in both validation samples. However, they bared a larger risk of wrongfully categorising a minor as an adult than reported for internal validation, stressing the importance of the prediction interval for age estimation in practice.


  • Artificial intelligence and dental age estimation: development and validation of an automated stage allocation technique on all mandibular tooth types in panoramic radiographs

    by Lander Matthijs, Lauren Delande, Jannick De Tobel, Barkin Büyükçakir, Peter Claes, Dirk Vandermeulen, Patrick Thevissen

    Age estimation in forensic odontology is mainly based on the development of permanent teeth. To register the developmental status of an examined tooth, staging techniques were developed. However, due to inappropriate calibration, uncertainties during stage allocation, and lack of experience, non-uniformity in stage allocation exists between expert observers. As a consequence, related age estimation results are inconsistent. An automated staging technique applicable to all tooth types can overcome this drawback. This study aimed to establish an integrated automated technique to stage the development of all mandibular tooth types and to compare their staging performances. Calibrated observers staged FDI teeth 31, 33, 34, 37 and 38 according to a ten-stage modified Demirjian staging technique. According to a standardised bounding box around each examined tooth, the retrospectively collected panoramic radiographs were cropped using Photoshop CC 2021® software (Adobe®, version 23.0). A gold standard set of 1639 radiographs were selected (n31 = 259, n33 = 282, n34 = 308, n37 = 390, n38 = 400) and input into a convolutional neural network (CNN) trained for optimal staging accuracy. The performance evaluation of the network was conducted in a five-fold cross-validation scheme. In each fold, the entire dataset was split into a training and a test set in a non-overlapping fashion between the folds (i.e., 80% and 20% of the dataset, respectively). Staging performances were calculated per tooth type and overall (accuracy, mean absolute difference, linearly weighted Cohen’s Kappa and intra-class correlation coefficient). Overall, these metrics equalled 0.53, 0.71, 0.71, and 0.89, respectively. All staging performance indices were best for 37 and worst for 31. The highest number of misclassified stages were associated to adjacent stages. Most misclassifications were observed in all available stages of 31. Our findings suggest that the developmental status of mandibular molars can be taken into account in an automated approach for age estimation, while taking incisors into account may hinder age estimation.


  • Quantifying the potential of morphological parameters for human dental identification: part 3—selecting the strongest skeletal identifiers in the mandible

    by Anca R. Iliescu, Cezar V. Capitaneanu, Debora Hürter, Steffen Fieuws, Jannick De Tobel, Patrick W. Thevissen

    The current study aimed to select the best mandibular morphological identifiers. One-hundred eighty-five panoramic radiographs were retrospectively collected, in which four landmarks were located on the mandible: the most superior point of the condyle right/left (CONR/L), of the coronoid right/left (CORR/L), of the mandibular lingula right/left (LINR/L), and the most mesial point of the mental foramen right/left (MMFR/L). Five linear measurements, 6 angles, and 10 ratios were measured bilaterally. Three groups of statistics were considered: (1) mean potential set; (2) inter-observer agreement quantified by intra-class correlation (ICC) and within-subject coefficient of variation (WSCV); and (3) Spearman correlation. Parameters were selected for a step-by-step cascade. In a univariate approach, the following parameters proved to have the best identifying capacity: ratio 3 right (between lines CONR — CORR and LINR — MMFR) with mean potential set 13%, ICC 0.90, and WSCV 4.8%; ratio 4 (between lines CONR/L — CORR/L and MMFR — MMFL) with mean potential set 13%, ICC 0.92, and WSCV 8.9%; and angle 4 left (between landmarks LINL, MMFL, and MMFR) with mean potential set of 18%, ICC 0.91, and WSCV 1.2%. The correlation coefficients ranged from 0.01 to 0.33. In a multivariate approach, the identifying capacity improved drastically, with all ratios combined as the strongest identifier (mean potential set 1.29%). In conclusion, a single ratio or a single angle already narrows down the set of potential matches, but the mean potential set remains relatively large. Combining all ratios drastically increases the certainty of the match.


  • Quantifying the potential of morphological parameters for human dental identification: part 2—selecting the strongest identifiers in mandibular permanent teeth

    by Yea Lee Shu, Jannick De Tobel, Chen Jun, Steffen Fieuws, Patrick W. Thevissen

    The current study aimed to select the best dental morphological identifiers for human identification. Sixty-two panoramic radiographs were collected retrospectively, in which six measurements were performed on all seven mandibular left permanent teeth: tooth length (TL), crown length (CL), root length (RL), crown width (CW), cervical width (CEJW), and root width (RW). Nine length–width ratios were then calculated using these measurements. Three groups of statistics were considered: (1) inter-observer reliability quantified by intra-class correlation (ICC); (2) mean “potential set”; and (3) Spearman correlation. A step-by-step cascade was then established based on selected parameters. In a univariate approach, the following parameters were the best identifiers: TL/CW for tooth 36 (ICC 0.82; mean potential set 13.7%), TL/CEJW for tooth 35 (ICC 0.87; mean potential set 15.2%), and TL/RW for tooth 32 (ICC 0.89; mean potential set 16.0%). The correlations between these three parameters ranged from 0.24 to 0.47. In a multivariate approach, the following parameters had the best identifying capacity: all parameters combined for tooth 31 (mean potential set 8.1%), for tooth 35 (mean potential set 11.9%), and for tooth 32 (mean potential set 16.3%). In conclusion, a single ratio in a specific tooth narrows down the potential set of matches, but the mean potential set remains relatively large. Combining all ratios of a single specific tooth increases the certainty of the match. In particular, tooth 31 was the strongest identifier.


  • Is third molar development affected by third molar impaction or impaction-related parameters?

    by Rosalina Intan Saputri, Jannick De Tobel, Myrthel Vranckx, Anna Ockerman, Martine Van Vlierberghe, Steffen Fieuws, Patrick Thevissen

    Objective To evaluate the effect of third molar impaction and impaction-related parameters on third molar development. Materials and methods Panoramic radiographs (N=3972) from 473 males and 558 females between 3.2 and 23.5 years old were analysed. Three parameters of impaction were examined: hindering contact between third and adjacent second molar, retromolar space availability (only in lower third molars), and angulation between the third and adjacent second molar. From the separate parameters, a definition for impaction was derived. Third molars’ development was staged according to a modified Köhler et al. staging technique. A linear model was used to compare within-stage and overall age, as a function of hindering contact, retromolar space, and impaction. Furthermore, a quadratic function was used to study the correlation between age and angulation. Results Significant differences were found in mean age as a function of hindering contact and retromolar space, depending on third molar location and stage. There was a significant relation between angulation and age, depending on the stage, with all third molars evolving to a more upright position (closer to 0°). Mean ages of subjects with impacted third molars were significantly lower in certain third molar stages, but the differences were clinically small (absolute differences ≤0.65 years). Moreover, after correction for stage differences, no significant differences in age could be demonstrated. Conclusions The development of impacted and non-impacted third molars can be considered clinically equal in our study population. Clinical relevance There is no distinction required between impacted and non-impacted third molars for dental age estimation.


  • Open Access

    Validating dental age estimation in Kenyan black children and adolescents using the Willems method

    by Maria Cadenas de Llano-Pérula, Eunice Kihara, Patrick Thevissen, Donna Nyamunga, Steffen Fieuws, Mary Kanini, Guy Willems

    This study aimed to validate the Willems Belgian Caucasian (Willems BC) age estimation model in a Kenyan sample, to develop and validate a Kenyan-specific (Willems KB) age estimation model and to compare the age prediction performances of both models. Panoramic radiographs of 1038 (523 female, 515 male) Kenyan children without missing permanent teeth and without all permanent teeth fully developed (except third molars) were retrospectively selected. Tooth development of the seven lower-left permanent teeth was staged according to Demirjian et al. The Willems BC model, performed on a Belgian Caucasian sample and a constructed Kenyan-specific model (Willems KB) were validated on the Kenyan sample. Their age prediction performances were quantified and compared using the mean error (ME), mean absolute error (MAE) and root-mean-square error (RMSE). The ME with Willems BC method equalled zero. Hence, there was no systematic under- or overestimation of the age. For males and females separately, the ME with Willems BC was significantly different from zero, but negligible in magnitude (–0.04 and 0.04, respectively). Willems KB was found not to outperform Willems BC, since the MAE and RMSE were comparable (0.98 vs 0.97 and 1.31 vs 1.29, respectively). Although Willems BC resulted in a higher percentage of subjects with predicted age within a one-year difference of the true age (63.3% vs 60.4%, p=0.018), this cannot be considered as clinically relevant. There is no reason to use a country-specific (Willems KB) model in children from Kenia instead of the original Willems (BC) model.


  • Dental and Skeletal Imaging in Forensic Age Estimation: Disparities in Current Approaches and the Continuing Search for Optimization

    by Jannick De Tobel, Christian Ottow, Thomas Widek, Isabella Klasinc, Håkan Mörnstad, Patrick Werner Thevissen, Koenraad Luc Verstraete

    Medical imaging for forensic age estimation in living adolescents and young adults continues to be controversial and a subject of discussion. Because age estimation based on medical imaging is well studied, it is the current gold standard. However, large disparities exist between the centers conducting age estimation, both between and within countries. This review provides an overview of the most common approaches applied in Europe, with case examples illustrating the differences in imaging modalities, in staging of development, and in statistical processing of the age data. Additionally, the review looks toward the future because several European research groups have intensified studies on age estimation, exploring four strategies for optimization: (1) increasing sample sizes of the reference populations, (2) combining single-site information into multifactorial information, (3) avoiding ionizing radiation, and (4) conducting a fully automated analysis.


  • The influence of motion artefacts on magnetic resonance imaging of the clavicles for age estimation

    by Jannick De Tobel, Mayonne van Wijk, Ivo Alberink, Elke Hillewig, Inès Phlypo, Rick R. van Rijn, Patrick Werner Thevissen, Koenraad Luc Verstraete, Michiel Bart de Haas

    To determine how motion affects stage allocation to the clavicle’s sternal end on MRI.