Various projects

Automated Tooth Instance Segmentation and Pathology Annotation

Caries detection in dental radiographs is a challenging and time consuming task even for experts in the field. Recent studies have shown the potential of tooth instance segmentation and caries detection with neural networks. We present a tooth level pathology annotation pipeline, based on automated tooth instance segmentation and numbering with a Mask-R-CNN architecture followed by the extraction of the bounding boxes of individual teeth as patches, that can be reassembled to the original image. 5-fold cross validation resulted in mean average precision (mAP) of 0.898 ± 0.02 for tooth instance segmentation. At performance levels at least similar to published data our approach provides flexibility for patch-based pathology diagnosis combined with the option to reassemble annotated patches to the original image. This will permit combining tooth-number-specific, neighborhood-based and entire image based features in future modeling along with tooth-centric review and diagnoses by clinical needs of dentists.

Read the full paper: CJ Hansen, J Conrad,  R Seidel, N Krekiehn, N Koser, M Götze, T Gehrmann, S Lauterbach, C Graetz, CDörfer & Claus C. Glüer (2024).  Automated Tooth Instance Segmentation and Pathology Annotation Pipeline for Panoramic Radiographs: Mask-R-CNN Approach with Elastic Transformations.  Bildverarbeitung in der Medizin, BVM Workshop. Wiesbaden: Springer Fachmedien Wiesbaden, 2024.

AI for Automated vBMD and Fragility Assessment of the Proximal
Femur in CT Scans

For this retrospective study, the challenge was to contact a large number of patients (#1799) via questionnaires, complying to ethical regulations and data privacy, and analyse their mail responses in an automated fashion. In order to provide and combine meta-data with clinical CT scans, I developed a pipeline to automatically analyse more than 1030 scanned response letters (the response rate was 57.2%!), alltoghether with more than 50.000 data fields. The data from 289 patients could ultimately be used in this study.

Background: Osteoporotic hip fractures are associated with high morbidity and mortality. Opportunistic screening by incidental analysis of routine clinical CT scans for fracture risk could reveal the need for prevention at an early stage. However, a freely available fully automated method for determining volumetric bone mineral density (vBMD) of the proximal femur is still lacking. Methods: The open-source AI tool TotalSegmentator was combined with two in-house AI models to segment both the proximal femur and a calibration phantom, enabling fully automated vBMD calculation. The accuracy of AI vBMD measurements was evaluated in 1070 hip QCT scans from the AGES study by comparison with the semi-automated gold standard MIAF. For an initial assessment of suitability, 289 clinical CT scans (ARTEMIS study) were analyzed regarding prediction of incident hip fractures. Results: AI HU vBMD values correlated closely with MIAF vBMD values (r=0.88–0.97). After calibration, correlation was r=0.96 with a bias of 1.6 mg/cm³ (integral) and 21.9 mg/cm³ (trabecular), and RMS errors of 15.1 mg/cm³ (integral) and 9.8 mg/cm³ (trabecular). Predictive performance for hip fractures (AUC 0.771–0.836) was significantly higher (p<0.031) than the baseline model of age and sex (AUC=0.641). Conclusions: The developed AI enables fully automated, rapid, and calibrated assessment of proximal femur vBMD directly from clinical CT scans and allows prediction of hip fracture risk. The positive results of this first prognostic study, however, need to be confirmed in independent and larger datasets. This approach offers the potential to identify at-risk patients in opportunistic screening and to initiate preventive measures at an earlier stage.

NR Krekiehn, S Bartenschlager, R Seidel, O Chaudry, S Sigurdsson, V Gudnason, JB Hövener, K Engelke, CC Glüer (2025). KI zur automatisierten vBMD-und Fragilitätsanalyse des proximalen Femurs an CT-Scans (AI for Automated vBMD and Fragility Assessment of the Proximal Femur in CT Scans) . Osteologie, 34(04), 256-264.

DLP-based bioprinting of bone constructs

Bioprinting technologies enable the integration of vital cells or active growth factors into 3D-printed constructs, while the integration of nanomaterials enables materialmediated functionalization of the bioink. To date, such bioink modifications with nanomaterials have rarely been reported for digital light processing (DLP) bioprinting technology. Furthermore, there is a notable lack of direct comparative studies on the impact of nanomaterials on cellular processes. In this study, we assessed and compared graphene oxide (GO)- and calcium phosphate (CaP)-modified bioinks for DLP bioprinting of methacrylated gelatin (GelMa)-based bone constructs. After printing, the impact of bioinks on cell distribution, viability, cell proliferation, and differentiation, as well as the mechanical and structural properties of constructs, was evaluated. In comparison to commercial bioinks, cell viability was higher in the established GelMa bioinks. Morphological data and DNA quantification indicate the highest cell vitality and proliferation over time in basic GelMa bioink. CaP-modified GelMa bioink displayed the highest differentiation of human mesenchymal stem cells (hMSCs), in terms of osteogenic gene expression and calcium deposition. Conversely, GO increased the Young’s modulus of the material, affecting cell morphology. Overall, the direct comparison of nanomaterials suggests diverse effects in functionalizing DLP-printed bone constructs containing living osteogenic cells.



J Kühl, SM Krümpelmann, L Hildebrandt, M Bruhn, Jan-B Hövener,  R Seidel, S Gorb, F Schütt, R Adelung, A Seekamp, L Siebert, S Fuchs (2024). Nanomaterial-modified bioinks for DLP-based bioprinting of bone constructs: Impact on mechanical properties and mesenchymal stem cell function. International Journal of Bioprinting 10.6 (2024): 4015.

Systematic analysis of post-mortem chemical excitability of the pupil

This is an ongoing collaboration with the institute of Legal Medicine at Kiel University Hospital. I developed a semi-automatic image analysis of time series photographs of eyes after chemical excitation with different pharmaceutical solutions. Changes of pupil size are measured over time – in more than 18.000 images.