EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Deep learning-based self-supervised detection and localization of magnetic resonance imaging anomalies in artificial intelligence-supported radiotherapy workflows

Mustafa Kadhim, Viktor Rogowski, Emilia Persson, Camila Gonzalez, André Haraldsson, Sofie Ceberg, Mikael Nilsson, Malin Kügele, Sven Bäck, Christian Jamtheim Gustafsson

Physics and Imaging in Radiation Oncology · 2026-11 · 42 · 101114

DOI: 10.1016/j.phro.2026.101114 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Physics and Imaging in Radiation Oncology, 2026-11. Verfasst von Mustafa Kadhim und 9 weiteren Autor:innen. Beteiligte Einrichtung: Medical Radiation Physics, Lund University, Lund, Sweden. Die Arbeit nennt 45 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Background and purpose: Artificial intelligence is increasingly integrated into radiotherapy workflows, yet it remains vulnerable to out-of-distribution image data that can introduce unexpected behavior in clinical tasks. Deep learning (DL)-based anomaly detection for pelvic magnetic resonance imaging (MRI) remains limited and transparent evaluation for full automation is lacking. This study aimed to develop two self-supervised anomaly-detection framework variations in radiotherapy for pelvic and brain MRI. Materials and methods: A two-stage self-supervised DL-framework was trained separately for pelvic and brain MRI utilizing three public datasets. The first stage compressed normal image data into discrete representations, and a second stage modeled the representation distribution. Anomaly-evidence was obtained by combining perceptual image differences with token-surprisal scores via negative log-likelihood estimation. Automated detection was evaluated for two anatomical cohorts: pelvis, encompassing both synthetic global anomalies and real clinical anomalies, and brain, comprising clinically annotated anomalies. Performance was assessed using sensitivity, specificity, and area under the receiver operating characteristic curve (AUC). For each cohort, a subset of normal images was reserved for analysis.

Results

Promising exploratory performance across evaluation cohorts was obtained. AUCs were 0.97 (95% CI 0.96-0.99) for synthetic pelvic anomalies, 0.95 (95% CI 0.90-1.00) for clinical pelvic anomalies, and 0.81 (95% CI 0.74-0.87) for brain anomalies. Qualitative heatmap analysis showed spatial agreement between detected anomalies and reference annotations.

Conclusions

The framework variations showed promising anomaly detection performance across pelvic and brain MRI, supporting its potential for automated MRI quality control in radiotherapy. Heatmap-based visualization further improved interpretability by spatially localizing detected anomalies.

Schlagwörter

Stichwörter der Autor:innen

Anomaly detectionAutomationBrain MRILUND-PROBEOutlier detectionPelvis MRIfastMRIquality control

Autor:innen und Institutionen

  • Medical Radiation Physics, Lund University, Lund, Sweden.
  • Radiation Physics, Department of Hematology, Oncology, and Radiation Physics, Skåne University Hospital, Lund, Sweden.
  • Medical Radiation Physics, Department of Translational Medicine, Lund University, Malmö, Sweden.
  • Medical University of Vienna Department of Anesthesia, Intensive Care Medicine, and Pain Therapy, Austria.
  • Comprehensive Center for Artificial Intelligence in Medicine (CAIM), Sweden.
  • Centre for Mathematical Sciences, Lund University, Lund, Sweden.

Zitieren

APA

Kadhim, M., Rogowski, V., Persson, E., Gonzalez, C., Haraldsson, A., Ceberg, S., Nilsson, M., Kügele, M., Bäck, S., & Gustafsson, C. J. (2026). Deep learning-based self-supervised detection and localization of magnetic resonance imaging anomalies in artificial intelligence-supported radiotherapy workflows. Physics and Imaging in Radiation Oncology, 42, 101114. https://doi.org/10.1016/j.phro.2026.101114

Vancouver

Kadhim M, Rogowski V, Persson E, Gonzalez C, Haraldsson A, Ceberg S, et al. Deep learning-based self-supervised detection and localization of magnetic resonance imaging anomalies in artificial intelligence-supported radiotherapy workflows. Physics and Imaging in Radiation Oncology. 2026;42:101114. doi:10.1016/j.phro.2026.101114

BibTeX
@article{Kadhim2026,
  author = {Kadhim, Mustafa and Rogowski, Viktor and Persson, Emilia and Gonzalez, Camila and Haraldsson, André and Ceberg, Sofie and Nilsson, Mikael and Kügele, Malin and Bäck, Sven and Gustafsson, Christian Jamtheim},
  title = {Deep learning-based self-supervised detection and localization of magnetic resonance imaging anomalies in artificial intelligence-supported radiotherapy workflows},
  journal = {Physics and Imaging in Radiation Oncology},
  year = {2026},
  volume = {42},
  pages = {101114},
  doi = {10.1016/j.phro.2026.101114}
}
Kostenlos starten →