EUVIMEDEUVIMED entdecken
QuerschnittstudieOpen Access · CC BY-NC-ND 4.0

Machine learning models for differential diagnosis of lipomas and soft-tissue liposarcomas: a multicenter cross-sectional study

Machine-Learning-Modelle zur Differentialdiagnose von Lipomen und Weichteil-Liposarkomen: multizentrische Querschnittsstudie

Alexander V. Urazov, Svetlana P. Aksenova, Nikolay V. Nudnov, Natalia V. Kharchenko, Nikolay I. Sergeev, Vladimir A. Solodkiy, Andrey D. Kaprin

Digital Diagnostics · 03.09.2026 · 7 (3) · 261-284

DOI: 10.17816/dd695849 ↗

Auf einen Blick

Mehrere Machine-Learning-Modelle wurden zur Unterscheidung von Lipomen und Liposarkomen auf MRT-Basis geprüft. Modelle mit semiotischen Merkmalen schnitten besser ab als rein radiomische Modelle.

Fragestellung

Wie gut unterscheiden Machine-Learning-Modelle mit semiotischen und radiomischen MRT-Merkmalen Lipome von Liposarkomen?

Methode

Retrospektive multizentrische Querschnittsstudie in 3 Zentren von 2012 bis 2025; histologisch verifizierte Fälle; Vergleich von 7 Machine-Learning-Algorithmen auf MRT-Merkmalen.

Ergebnisse

Modelle nur mit semiotischen Merkmalen waren am besten. LDA und Naive Bayes erreichten nach Optimierung eine ROC-AUC von 0,944 bzw. 0,944; die Gesamtgenauigkeit betrug 0,842. Ein Hybridmodell mit T2-Merkmalen erreichte eine ROC-AUC von 0,912 (95 %-KI 0,744–1,000).

Einschränkungen

Externe Validierung in größeren multizentrischen Kohorten erforderlich.

Schlussfolgerung der Autor:innen

Die Modelle bilden eine brauchbare Grundlage für klinische Decision-Support-Systeme zur präoperativen Beurteilung von Lipomen und Liposarkomen; die Übertragbarkeit muss extern validiert werden.

Kernaussagen

  • Semiotische Merkmale waren radiomischen Merkmalen überlegen.
  • Beste ROC-AUC: 0,944 nach Optimierung.
  • Hybridmodell T2: ROC-AUC 0,912 (95 %-KI 0,744–1,000).

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

Kurz gefasst

Querschnittstudie in der Zeitschrift Digital Diagnostics, online erschienen am 03.09.2026. Verfasst von Alexander V. Urazov und 6 weiteren Autor:innen. Die Arbeit nennt 27 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0).

Auf einen BlickAbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

BACKGROUND: Differences in management strategies, therapeutic approaches, and overall prognosis between soft-tissue lipomas and liposarcomas underscore the clinical necessity for accurate differentiation. However, differential diagnosis remains inherently challenging due to overlapping semiotic features on magnetic resonance imaging. Consequently, objective and automated analytical methods for magnetic resonance imaging interpretation are required. AIM: To develop machine learning models incorporating semiotic and radiomic features and compare their diagnostic efficacy in distinguishing soft-tissue lipomas from liposarcomas. METHODS: A retrospective multicenter cross-sectional study was conducted across three research centers from 2012 to 2025. Inclusion criteria were histologically verified lipomas or liposarcomas and a contrast-enhanced or non-contrast magnetic resonance imaging examination performed within one month prior to antitumor therapy. Non-inclusion criteria comprised lack of histological verification, histologically verified diagnosis other than lipoma or liposarcoma, incomplete magnetic resonance imaging acquisition protocols, and pronounced image artifacts precluding interpretation. The cohorts were categorized by diagnosis (lipoma vs. liposarcoma) verified by postoperative histopathology. Predictors included semiotic features and radiomic descriptors extracted from volumetric tumor segmentations on native T1- and T2-weighted sequences. The models were build using seven machine learning algorithms: logistic regression, linear discriminant analysis, support vector machines, random forest, naive Bayes classifier, k-nearest neighbors, and decision tree. RESULTS: Models constructed exclusively with semiotic features demonstrated superior predictive performance and robustness. Following hyperparameter optimization, linear discriminant analysis and naive Bayes models achieved an area under the receiver operating characteristic curve (ROC-AUC) of 0.944 [95% CI, 0.810–1.000] and (95% CI, 0.773–1.000) with an overall accuracy of 0.842. Conversely, models relying solely on radiomic features exhibited lower diagnostic performance. Peak pre- and post-optimization ROC-AUC values were 0.833 and 0.611 for the decision tree model based on the features extracted from T1-weighted images, 0.833 and 0.700 for the random forest model (T2-weighted images), and 0.851 and 0.724 for the random forest/logistic regression models (T1 + T2-weighted images). A hybrid model integrating semiotic and radiomic features extracted from T2-weighted images yielded a ROC-AUC of 0.912 (95% CI, 0.744–1.000). Using the full feature array (214 radiomic and 7 semiotic features) in a subset of 88 patients failed to yield sustained accuracy improvements (pre- and post-optimization ROC-AUC of 0.848 and 0.749, respectively). CONCLUSION: The models provide a viable foundation for clinical decision-support systems aimed at refining preoperative evaluation of lipomas and liposarcomas. External validation across larger multicenter cohorts is required to confirm diagnostic generalizability and establish feasibility of implementation into clinical practice.

Zitieren

APA

Urazov, A. V., Aksenova, S. P., Nudnov, N. V., Kharchenko, N. V., Sergeev, N. I., Solodkiy, V. A., & Kaprin, A. D. (2026). Machine learning models for differential diagnosis of lipomas and soft-tissue liposarcomas: a multicenter cross-sectional study. Digital Diagnostics, 7(3), 261-284. https://doi.org/10.17816/dd695849

Vancouver

Urazov AV, Aksenova SP, Nudnov NV, Kharchenko NV, Sergeev NI, Solodkiy VA, et al. Machine learning models for differential diagnosis of lipomas and soft-tissue liposarcomas: a multicenter cross-sectional study. Digital Diagnostics. 2026;7(3):261-284. doi:10.17816/dd695849

BibTeX
@article{Urazov2026,
  author = {Urazov, Alexander V. and Aksenova, Svetlana P. and Nudnov, Nikolay V. and Kharchenko, Natalia V. and Sergeev, Nikolay I. and Solodkiy, Vladimir A. and Kaprin, Andrey D.},
  title = {Machine learning models for differential diagnosis of lipomas and soft-tissue liposarcomas: a multicenter cross-sectional study},
  journal = {Digital Diagnostics},
  year = {2026},
  volume = {7},
  number = {3},
  pages = {261--284},
  doi = {10.17816/dd695849}
}
Kostenlos starten →