Master’s thesis · FAU, AIBE / IDEA Lab · submitted November 2026
From Scan to Report: Detecting, Segmenting and Reporting Findings in Chest CT
Reading a chest CT means three things: detecting the abnormalities, locating them, and writing them up in a report. The thesis takes each step in turn and measures where the current state of the art actually stands, on CT-RATE and ReXGroundingCT. Across all three, task design and the choice of metric change how pretrained encoders should be used, and judging clinical accuracy still needs unseen data and expert review.
- Detecting8 frozen CT encoders compared; attention over FlexiCT tokens reaches 0.8695 AUROC
- Segmentingadapting VoxTell’s text-to-mask head raises Dice by 37.8%
- ReportingCRG can’t tell whether a report reflects the scan; GREEN and CRIMSON can
Supervised by Prof. Dr. Bernhard Kainz and Dr. Johanna Müller.
Most of this is collaborative research, and several repositories are private. Links are given where code or models are public.
3× 1st
plus a 2nd, VLM3D at MICCAI 2026
VLM3D: 3D CT and MR classification and report generation
Collaborative research, FAU · PyTorch, Slurm, MIL, VLMs
Part of the FAU teams that won CT abnormality classification, CT report generation and MR report generation, and placed second in MR abnormality classification (certificates). I built a benchmarking framework comparing 10 3D image-encoder families on 50,183 chest CT volumes, cutting storage from a projected 13 TB to 260 GB. For reports, I showed that end-to-end generators were barely grounded in the scan, and that a classify-then-assemble pipeline beat the baseline by +0.165 CRG.
1st
of 37 teams, CLEF CheckThat! 2026
Collaborative research, TU Dresden · RAG, reranking, LLM verification
Multilingual retrieval of the scientific paper behind a claim made on social media. I implemented the LLM-based evidence verification that delivered the decisive score gain; the final system reached MRR@5 = 0.7628 across English, German and French and improved on the baseline by 75%. Co-authored two CLEF papers.
11/53
teams, ReXGroundingCT challenge
Text-conditioned 3D grounding in chest CT
Collaborative research, FAU · VoxTell, DDP, HPC
Locating findings from a radiology report inside the 3D scan. Freezing the VoxTell backbone and training only the grounding head lifted leaderboard Dice from 0.280 to 0.332. Analysis showed missed lesions, not mask quality, were the bottleneck. Distributed training with NVMe staging raised GPU utilisation from 10% to 95%, a 7× speedup.
45%
reports corrupted by the top-BLEU system
Research project · NLP, MT evaluation, clinical metrics
Benchmarked 13 translation systems on 296 German–English radiology reports. Fluent output can still flip a negation or drop a measurement: BLEU mis-ranks clinical safety, while claim-entailment and GREEN checks catch factual corruption far better (AUC 0.84–0.92).
956
clinical CT scans, University Hospital Erlangen
Medical VLMs on real German CT reports
University Hospital Erlangen & FAU · VLMs, quantization, evaluation
Evaluated four state-of-the-art medical vision-language models (MedM-VL, MedGemma-27B, CT-CHAT, COLIPRI) on 956 thoracic CT volumes with German free-text reports. Models built specifically for thoracic CT clearly beat broader medical VLMs, but scores stayed low overall. Presented at BAIOSPHERE MEDICAL 2026 (poster).
3D
life-size AI patient, browser and Apple Vision Pro
Hackathon, OTH Amberg-Weiden · made for Apple Vision Pro · three.js, WebXR, ElevenLabs, Claude on Bedrock
A virtual psychiatric patient for practising history-taking. Students build a case in three clicks, talk to the patient out loud in a 3D consultation room, and get turn-by-turn feedback afterwards. The patient can drift off topic or get angry, and its body language follows the conversation.