EUVIMEDEUVIMED entdecken
CrossrefInfokarte

Assessment of the accuracy of lung lesions diagnosis in adolescents with osteosarcoma using artificial intelligence: a cross-sectional study

Natalia G. Uskova, Victor A. Gombolevskiy, Valeriya Yu. Chernina, Dmitriy V. Burenchev, Dmitriy G. Akhaladze, Elena V. Panina, Aleksandr I. Karachunskiy, Galina V. Tereschenko, Mikhail Yu. Goncharov, Evgeniya A. Soboleva, Elena I. Konopleva, Oleg I. Bydanov, Sergei Yu. Plekhov, Nikolay S. Grachev

Digital Diagnostics · 2026-09-28 · 7 (3) · 296-310

DOI: 10.17816/dd704241 ↗
AbstractBibliografische Angaben

Abstract

INTRODUCTION: Pulmonary metastases in osteosarcoma are the leading cause of death. The accurate diagnosis of lesions on chest computed tomography is critical for staging disseminated disease and planning surgical treatment. The use of artificial intelligence to detect pulmonary lesions improves diagnostic accuracy and reduces the likelihood of missing metastases. AIM: to evaluate the accuracy of diagnosing pulmonary lesions on chest computed tomography scans using an artificial intelligence model in adolescents with osteosarcoma. METHODS: A retrospective assessment of computed tomography scans of adolescents with osteosarcoma was performed in a multicenter cross-sectional study. A pathological lesion of medium size of ≥ 4 mm was considered as the target finding. An artificial intelligence algorithm previously trained on a dataset of adult patients was assessed for its diagnostic accuracy. The number of false-positive and false-negative results was determined; sensitivity, specificity, accuracy, area under the receiver operating characteristic curve, positive predictive value, negative predictive value, and F1-score were also calculated. Based on the obtained results, the algorithm efficiency was assessed. RESULTS: 248 computed tomography scans of adolescents with osteosarcoma were analyzed. The following results were obtained: the artificial intelligence algorithm showed false-positive findings in 5 cases (2.02%), false-negative findings in 34 cases (13.71%), and a correct result—both true positive and true negative—in 209 cases (84.27%). The diagnostic accuracy of the algorithm was 0.843 (95% confidence interval: 0.794-0.887). Applying the artificial intelligence model in a radiologist’s practice for a specific clinical task would increase the sensitivity from 0.805 to 0.891 while reducing the absolute number of false-negative results by 8.59% and the relative decrease by 44%. CONCLUSION: The obtained results indicate the potential clinical utility of the artificial intelligence algorithm for diagnosing pulmonary metastases in adolescent patients with osteosarcoma. However, they also underscore the persisting statistical uncertainty and the need for validation in a larger and more representative cohort.

Wissen, das zu Ihrer Arbeit passt.

Vom kostenlosen Einstieg über die persönliche Recherche bis zur vollständigen Research Suite.

Tarife ansehen →

Sind alle Verlagsvolltexte enthalten?

Nein. Bibliografische Nachweise, Abstracts und Volltexte sind unterschiedliche Inhalte. Ob ein Volltext verfügbar ist, hängt von der Quelle, Open Access und Ihren vorhandenen Lizenzen ab. Ein EUVIMED-Tarif ersetzt keine zusätzlichen Verlagsrechte.

Tarife ansehen →