OpenLB
Open Library of Bioscience
Advanced Search
BMC cancer
Jul 22, 2024;24 (1): 875.

Differentiation of granulomatous nodules with lobulation and spiculation signs from solid lung adenocarcinomas using a CT deep learning model.

Yanhua Wen, Wensheng Wu, Yuling Liufu, Xiaohuan Pan, Yingying Zhang, Shouliang Qi, Yubao Guan
Author Information
  1. Yanhua Wen: Department of Medical Imaging, the Fifth Affiliated Hospital of Guangzhou Medical University, 621 Gangwan Road, Guangzhou, 510700, Guangdong, China.
  2. Wensheng Wu: Department of Medical Imaging, the Fifth Affiliated Hospital of Guangzhou Medical University, 621 Gangwan Road, Guangzhou, 510700, Guangdong, China.
  3. Yuling Liufu: Department of Medical Imaging, the Fifth Affiliated Hospital of Guangzhou Medical University, 621 Gangwan Road, Guangzhou, 510700, Guangdong, China.
  4. Xiaohuan Pan: Department of Radiology, The First Affiliated Hospital of Guangzhou Medical University, Guangzhou, Guangdong, China.
  5. Yingying Zhang: Department of Medical Imaging, the Fifth Affiliated Hospital of Guangzhou Medical University, 621 Gangwan Road, Guangzhou, 510700, Guangdong, China.
  6. Shouliang Qi: Key Laboratory of Intelligent Computing in Medical Image, College of Medicine and Biological Information Engineering, Northeastern University, Shenyang, Liaoning, China.
  7. Yubao Guan: Department of Medical Imaging, the Fifth Affiliated Hospital of Guangzhou Medical University, 621 Gangwan Road, Guangzhou, 510700, Guangdong, China. yubaoguan@163.com.
PMID: 39039511 DOI: 10.1186/s12885-024-12611-0

Abstract

BACKGROUND: The diagnosis of solitary pulmonary nodules has always been a difficult and important point in clinical research, especially granulomatous nodules (GNs) with lobulation and spiculation signs, which are easily misdiagnosed as malignant tumors. Therefore, in this study, we utilised a CT deep learning (DL) model to distinguish GNs with lobulation and spiculation signs from solid lung adenocarcinomas (LADCs), to improve the diagnostic accuracy of preoperative diagnosis.
METHODS: 420 patients with pathologically confirmed GNs and LADCs from three medical institutions were retrospectively enrolled. The regions of interest in non-enhanced CT (NECT) and venous contrast-enhanced CT (VECT) were identified and labeled, and self-supervised labels were constructed. Cases from institution 1 were randomly divided into a training set (TS) and an internal validation set (IVS), and cases from institutions 2 and 3 were treated as an external validation set (EVS). Training and validation were performed using self-supervised transfer learning, and the results were compared with the radiologists' diagnoses.
RESULTS: The DL model achieved good performance in distinguishing GNs and LADCs, with area under curve (AUC) values of 0.917, 0.876, and 0.896 in the IVS and 0.889, 0.879, and 0.881 in the EVS for NECT, VECT, and non-enhanced with venous contrast-enhanced CT (NEVECT) images, respectively. The AUCs of radiologists 1, 2, 3, and 4 were, respectively, 0.739, 0.783, 0.883, and 0.901 in the (IVS) and 0.760, 0.760, 0.841, and 0.844 in the EVS.
CONCLUSIONS: A CT DL model showed great value for preoperative differentiation of GNs with lobulation and spiculation signs from solid LADCs, and its predictive performance was higher than that of radiologists.

Keywords

Artificial Intelligence Computer tomography Deep learning Granulomatous nodules Lung adenocarcinomas

References

  1. Lung Cancer. 2018 Nov;125:109-114 [PMID: 30429007]
  2. Comput Methods Programs Biomed. 2021 Sep;208:106254 [PMID: 34260970]
  3. JAMA. 2001 Feb 21;285(7):914-24 [PMID: 11180735]
  4. J Magn Reson Imaging. 2016 Mar;43(3):669-79 [PMID: 26340144]
  5. Radiographics. 2018 Sep-Oct;38(5):1337-1350 [PMID: 30207935]
  6. Arch Pathol Lab Med. 2018 Sep;142(9):1127-1133 [PMID: 29547001]
  7. Diagnostics (Basel). 2021 Nov 26;11(12): [PMID: 34943443]
  8. IEEE Trans Neural Netw Learn Syst. 2023 May 01;PP: [PMID: 37126634]
  9. Sci Rep. 2016 Apr 15;6:24454 [PMID: 27079888]
  10. Adv Exp Med Biol. 2020;1213:3-21 [PMID: 32030660]
  11. Semin Ultrasound CT MR. 2022 Jun;43(3):230-245 [PMID: 35688534]
  12. Eur Radiol. 2020 Dec;30(12):6497-6507 [PMID: 32594210]
  13. Nat Med. 2018 Oct;24(10):1559-1567 [PMID: 30224757]
  14. N Engl J Med. 2011 Aug 4;365(5):395-409 [PMID: 21714641]
  15. Chest. 2013 Mar;143(3):825-839 [PMID: 23460160]
  16. Intern Med J. 2019 Mar;49(3):306-315 [PMID: 30897667]
  17. Phys Med Biol. 2021 Apr 01;66(7): [PMID: 33647890]
  18. Brief Bioinform. 2021 Mar 22;22(2):1592-1603 [PMID: 33569575]
  19. Chest. 2013 Mar;143(3):840-846 [PMID: 23460161]
  20. IEEE Trans Pattern Anal Mach Intell. 2021 Nov;43(11):4037-4058 [PMID: 32386141]
  21. Tomography. 2022 Sep 28;8(5):2411-2425 [PMID: 36287799]
  22. Comput Intell Neurosci. 2021 Feb 23;2021:6659083 [PMID: 33679966]
  23. Cancer Imaging. 2023 Oct 5;23(1):95 [PMID: 37798797]
  24. Diagn Interv Imaging. 2016 Oct;97(10):1003-1017 [PMID: 27567555]
  25. J Magn Reson Imaging. 2019 Jun;49(7):e256-e264 [PMID: 30575193]
  26. J Thorac Imaging. 2011 May;26(2):90-105 [PMID: 21508732]
  27. Magn Reson Imaging. 2012 Nov;30(9):1301-12 [PMID: 22898682]
  28. Expert Rev Precis Med Drug Dev. 2019;4(2):59-72 [PMID: 31080889]
  29. Eur J Radiol. 2020 Aug;129:109013 [PMID: 32505895]
  30. Strahlenther Onkol. 2020 Oct;196(10):879-887 [PMID: 32367456]

MeSH Term

Humans
Deep Learning
Male
Adenocarcinoma of Lung
Female
Tomography, X-Ray Computed
Middle Aged
Lung Neoplasms
Diagnosis, Differential
Aged
Retrospective Studies
Solitary Pulmonary Nodule
Adult
Granuloma
Journal Article
Article has an altmetric score of 1

Word Cloud

Created with Highcharts 10.0.00CTGNsnoduleslobulationspiculationsignslearningmodelLADCsDLsolidadenocarcinomassetvalidationIVSEVSdiagnosisgranulomatousdeeplungpreoperativeinstitutionsnon-enhancedNECTvenouscontrast-enhancedVECTself-supervised123usingperformancerespectivelyradiologists760BACKGROUND:solitarypulmonaryalwaysdifficultimportantpointclinicalresearchespeciallyeasilymisdiagnosedmalignanttumorsThereforestudyutiliseddistinguishimprovediagnosticaccuracyMETHODS:420patientspathologicallyconfirmedthreemedicalretrospectivelyenrolledregionsinterestidentifiedlabeledlabelsconstructedCasesinstitutionrandomlydividedtrainingTSinternalcasestreatedexternalTrainingperformedtransferresultscomparedradiologists'diagnosesRESULTS:achievedgooddistinguishingareacurveAUCvalues917876896889879881NEVECTimagesAUCs4739783883901841844CONCLUSIONS:showedgreatvaluedifferentiationpredictivehigherDifferentiationArtificialIntelligenceComputertomographyDeepGranulomatousLung

Similar Articles

Cited By

No available data.