EUVIMEDEUVIMED entdecken
Open Access · CC BY-SA 4.0

Segmentation of Acute Ischemic Stroke Infarct using Multi-scale U-Net Architecture on Non-contrast CT Scan Images

Segmentierung des akuten ischämischen Schlaganfallinfarkts mit einer Multi-Scale-U-Net-Architektur auf Nativ-CT-Bildern

Osmalina Rahma, Aisyah Widayani, Nimas Maharani Putri, Rifai Chai

Journal of Electronics, Electromedical Engineering, and Medical Informatics · 31.10.2026 · 8 (4) · 1796-1818

DOI: 10.35882/jeeemi.v8i4.1980 ↗

Auf einen Blick

Die Studie verglich U-Net mit einer Multi-Scale-U-Net-Variante zur Segmentierung akuter ischämischer Infarkte in NCCT. Die Multi-Scale-Variante verbesserte die Überlappung nicht, verschob aber Sensitivität und Präzision.

Fragestellung

Verbessert ein Multi-Scale-Block in U-Net die Segmentierung akuter ischämischer Infarkte auf NCCT im Vergleich zu U-Net?

Methode

Vergleich von U-Net und Multi-Scale-U-Net; Training auf dem öffentlichen Acute Ischemic Stroke Dataset mit 375 Studien; Test auf 51 Patient:innen; vier Konfigurationen mit und ohne Augmentation.

Ergebnisse

Auf 51 Testpatient:innen lag der Dice-Score bei 0,6372 für Multi-Scale-U-Net und 0,6632 für U-Net mit Augmentation. Die gepaarte Differenz betrug 0,0233 zugunsten von U-Net (95 %-KI −0,060 bis 0,013; p = 0,19). Sensitivität +0,060, Präzision −0,100.

Schlussfolgerung der Autor:innen

Der Multi-Scale-Block verbesserte die Segmentierung nicht, sondern änderte nur das Verhältnis von Sensitivität und Präzision. Augmentation verbesserte beide Netzwerke; kleine Infarkte blieben schwer segmentierbar.

Kernaussagen

  • Dice: 0,6372 vs. 0,6632 mit Augmentation
  • Sensitivität +0,060, Präzision −0,100
  • Augmentation senkte Fehlpositiven von 15,5 auf 5,6 pro Patient

Deutsche Zusammenfassung: KI-gestützt aus dem Original-Abstract erstellt; alle Zahlen wurden automatisch mit dem Original abgeglichen. Maßgeblich ist der Original-Abstract.

Kurz gefasst

Fachartikel in der Zeitschrift Journal of Electronics, Electromedical Engineering, and Medical Informatics, online erschienen am 31.10.2026. Verfasst von Osmalina Rahma und 3 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY-SA 4.0).

Auf einen BlickAbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Acute ischemic stroke must be treated within a few hours, and the first scan is usually a non-contrast CT (NCCT), on which an early infarction is often hard to see. Automatic segmentation assist image interpretation, and multi-scale convolution is a commonly proposed approach. This study measured how much such a block improves U-Net performance for this task, and what it costs. Every convolutional block of U-Net, in the encoder and the decoder, was replaced by a block with three parallel branches using 1×1, 3×3 and 5×5 kernels. Both networks were trained on the public Acute Ischemic Stroke Dataset, on 375 studies with supratentorial infarcts. The data were divided by study; the scans of the dataset’s own test split were kept for testing, and augmentation was applied to the training subset alone. Four configurations were compared: each architecture with and without augmentation. On the 51 test patients, Multi-scale U-Net did not improve overlap. Its Dice score was 0.6372 against 0.6632 for U-Net with augmentation, and the paired difference per patient was 0.0233 in favour of U-Net (95 % CI −0.060 to 0.013, p = 0.19). The multi-scale block instead altered the balance between the two types error: sensitivity increased by 0.060 (p = 0.002), and precision decreased by 0.100 (p < 0.001). Augmentation improved both networks (+0.106 and +0.048 Dice per patient, p < 0.001), mainly by cutting false-positive lesions from 15.5 to 5.6 per patient. The multi-scale block costs 2.31 times as many parameters and 3.37 times as much inference time. An ablation over the target definition favoured the dataset authors' labels: adding the label for infarct invisible on NCCT cost 0.040 Dice and tripled false-positive lesions. Both networks segment small infarcts poorly, with Dice below 0.16 under 100 pixels, which remains the main obstacle.

Zitieren

APA

Rahma, O., Widayani, A., Putri, N. M., & Chai, R. (2026). Segmentation of Acute Ischemic Stroke Infarct using Multi-scale U-Net Architecture on Non-contrast CT Scan Images. Journal of Electronics, Electromedical Engineering, and Medical Informatics, 8(4), 1796-1818. https://doi.org/10.35882/jeeemi.v8i4.1980

Vancouver

Rahma O, Widayani A, Putri NM, Chai R. Segmentation of Acute Ischemic Stroke Infarct using Multi-scale U-Net Architecture on Non-contrast CT Scan Images. Journal of Electronics, Electromedical Engineering, and Medical Informatics. 2026;8(4):1796-1818. doi:10.35882/jeeemi.v8i4.1980

BibTeX
@article{Rahma2026,
  author = {Rahma, Osmalina and Widayani, Aisyah and Putri, Nimas Maharani and Chai, Rifai},
  title = {Segmentation of Acute Ischemic Stroke Infarct using Multi-scale U-Net Architecture on Non-contrast CT Scan Images},
  journal = {Journal of Electronics, Electromedical Engineering, and Medical Informatics},
  year = {2026},
  volume = {8},
  number = {4},
  pages = {1796--1818},
  doi = {10.35882/jeeemi.v8i4.1980}
}
Kostenlos starten →