Advertisement

Künstliche Intelligenz und maschinelles Lernen in der onkologischen Bildgebung

  • Jens KleesiekEmail author
  • Jacob M. Murray
  • Georgios Kaissis
  • Rickmer Braren
Leitthema
  • 45 Downloads

Zusammenfassung

Hintergrund

Maschinelles Lernen (ML) hält gegenwärtig Einzug in vielen Bereichen der Gesellschaft, so auch in der Medizin. Diese Transformation birgt das Potenzial, das Berufsbild und den Berufsalltag drastisch zu verändern, auch wenn diese Neuerungen bis jetzt nur vereinzelt die klinische Praxis beeinflussen und mit Risiken verbunden sein können. In den Stadien und der Interaktion zwischen den Disziplinen und Modalitäten der onkologischen Patientenversorgung wird dies besonders deutlich. Computer erbringen in mehreren Forschungsarbeiten in Kollaboration mit Menschen oder allein bereits bessere Ergebnisse als Menschen in der Tumoridentifikation, ihrer Klassifikation sowie beim Erstellen von Prognosen und der Evaluation von Therapien. Zudem können Algorithmen – z. B. künstliche neuronale Netze (KNN), welche für viele der gegenwärtigen Errungenschaften im ML-Feld verantwortlich sind – dies reproduzierbar, schnell und kostengünstig erbringen.

Ziel der Arbeit

In dieser Übersichtsarbeit wird der gegenwärtige Forschungsstand beispielhaft anhand von ausgewählten Tumorentitäten beleuchtet und in die Entwicklung des Forschungsgebiets und der Medizin eingeordnet.

Material und Methoden

Diese Arbeit basiert auf einer selektiven Literaturrecherche in den Datenbanken PubMed und arXiv.

Schlussfolgerungen

Zukünftig werden KI-Anwendungen sich zu einem integralen Bestandteil des ärztlichen Handels entwickeln und Vorteile für die onkologische Diagnostik und Therapie bieten.

Schlüsselwörter

Maschinelles Lernen Computergestützte Bildverarbeitung Diagnostische Bildgebung Deep Learning Neuronale Netze (Computer) 

Artificial intelligence and machine learning in oncologic imaging

Abstract

Background

Machine learning (ML) is finding entry into many areas of society, including medicine. This transformation has the potential to drastically change the perception of medicine and medical practice. While these advances currently only influence clinical routine in isolated cases, they also come with risks. These aspects become particularly clear when considering the different stages of oncologic patient care and the involved interdisciplinary and intermodality interactions. In recent publications, computers—in collaboration with humans or alone—have been outperforming humans. This pertains to tumor identification, tumor classification, creation of prognoses, and evaluation of treatments. Additionally, ML algorithms, e.g., artificial neural networks (ANNs), which constitute the drivers behind many of the latest achievements in ML, can deliver this level of performance in a reproducible, fast, and cheap manner.

Objective

This review elucidates the current state of research on ML in oncology by focusing on selected tumor entities, and relates this to the development of research and medicine as a whole.

Materials and methods

This work is based on a selective literature search in the databases PubMed and arXiv.

Conclusion

In the future, AI applications will develop into an integral part of the medical profession and offer advantages for oncologic diagnostics and treatment.

Keywords

Machine learning Computer-assisted image processing Diagnostic imaging Deep Learning Neural networks (computer) 

Notes

Einhaltung ethischer Richtlinien

Interessenkonflikt

J. Kleesiek, J.M. Murray, G. Kaissis und R. Braren geben an, dass kein Interessenkonflikt besteht.

Für diesen Beitrag wurden von den Autoren keine Studien an Menschen oder Tieren durchgeführt. Für die aufgeführten Studien gelten die jeweils dort angegebenen ethischen Richtlinien.

Literatur

  1. 1.
    Ardila D, Kiraly AP, Bharadwaj S et al (2019) End-to-end lung cancer screening with three-dimensional deep learning on low-dose chest computed tomography. Nat Med 25:954–961.  https://doi.org/10.1038/s41591-019-0447-x CrossRefPubMedGoogle Scholar
  2. 2.
    Bejnordi BE, Veta M, van Diest PJ et al (2017) Diagnostic assessment of deep learning algorithms for detection of lymph node metastases in women with breast cancer. JAMA 318:2199–2210.  https://doi.org/10.1001/jama.2017.14585 CrossRefGoogle Scholar
  3. 3.
    Bickelhaupt S, Jaeger PF, Laun FB et al (2018) Radiomics based on adapted diffusion kurtosis imaging helps to clarify most mammographic findings suspicious for cancer. Radiology.  https://doi.org/10.1148/radiol.2017170273 CrossRefPubMedGoogle Scholar
  4. 4.
    Bonekamp D, Kohl S, Wiesenfarth M et al (2018) Radiomic machine learning for characterization of prostate lesions with MRI: comparison to ADC values. Radiology 289:128–137.  https://doi.org/10.1148/radiol.2018173064 CrossRefPubMedGoogle Scholar
  5. 5.
    Campanella G, Hanna MG, Geneslaw L et al (2019) Clinical-grade computational pathology using weakly supervised deep learning on whole slide images. Nat Med.  https://doi.org/10.1038/s41591-019-0508-1 CrossRefPubMedGoogle Scholar
  6. 6.
    Case N (2018) How to become a centaur. JoDS.  https://doi.org/10.21428/61b2215c CrossRefGoogle Scholar
  7. 7.
    Coudray N, Ocampo PS, Sakellaropoulos T et al (2018) Classification and mutation prediction from non–small cell lung cancer histopathology images using deep learning. Nat Med 24:1559–1567.  https://doi.org/10.1038/s41591-018-0177-5 CrossRefPubMedGoogle Scholar
  8. 8.
    Dou TH, Coroller TP, van Griethuysen JJM et al (2018) Peritumoral radiomics features predict distant metastasis in locally advanced NSCLC. PLoS ONE 13:e206108.  https://doi.org/10.1371/journal.pone.0206108 CrossRefPubMedPubMedCentralGoogle Scholar
  9. 9.
    Ellingson BM, Wen PY, Cloughesy TF (2017) Modified criteria for radiographic response assessment in glioblastoma clinical trials. Neurotherapeutics 14:307–320.  https://doi.org/10.1007/s13311-016-0507-6 CrossRefPubMedPubMedCentralGoogle Scholar
  10. 10.
    Esteva A, Kuprel B, Novoa RA et al (2017) Dermatologist-level classification of skin cancer with deep neural networks. Nature 542:115–118.  https://doi.org/10.1038/nature21056 CrossRefPubMedGoogle Scholar
  11. 11.
    Finlayson SG, Chung HW, Kohane IS, Beam AL (2018) Adversarial attacks against medical deep learning systemsGoogle Scholar
  12. 12.
    Gong E, Pauly JM, Wintermark M, Zaharchuk G (2018) Deep learning enables reduced gadolinium dose for contrast-enhanced brain MRI. J Magn Reson Imaging 48:330–340.  https://doi.org/10.1002/jmri.25970 CrossRefPubMedGoogle Scholar
  13. 13.
    Haenssle HA, Fink C, Schneiderbauer R et al (2018) Man against machine: diagnostic performance of a deep learning convolutional neural network for dermoscopic melanoma recognition in comparison to 58 dermatologists. Ann Oncol 29:1836–1842.  https://doi.org/10.1093/annonc/mdy166 CrossRefPubMedGoogle Scholar
  14. 14.
    Han SS, Kim MS, Lim W et al (2018) Classification of the clinical images for benign and malignant cutaneous tumors using a deep learning algorithm. J Invest Dermatol 138:1529–1538.  https://doi.org/10.1016/j.jid.2018.01.028 CrossRefPubMedGoogle Scholar
  15. 15.
    Hekler A, Utikal JS, Enk AH et al (2019) Deep learning outperformed 11 pathologists in the classification of histopathological melanoma images. Eur J Cancer 118:91–96.  https://doi.org/10.1016/j.ejca.2019.06.012 CrossRefPubMedGoogle Scholar
  16. 16.
    Isensee F, Kickingereder P, Wick W et al (2018) No New-NetGoogle Scholar
  17. 17.
    Jha S, Topol EJ (2016) Adapting to artificial intelligence: radiologists and pathologists as information specialists. JAMA 316:2353–2354.  https://doi.org/10.1001/jama.2016.17438 CrossRefPubMedGoogle Scholar
  18. 18.
    Kaissis G, Ziegelmayer S, Lohöfer F et al (2019) A prospectively validated machine learning model for the prediction of survival and tumor subtype in pancreatic ductal adenocarcinoma. Bioinformatics.  https://doi.org/10.1101/643809 CrossRefGoogle Scholar
  19. 19.
    Kaissis G, Ziegelmayer S, Lohöfer F et al (2019) A machine learning algorithm predicts molecular subtypes in pancreatic ductal adenocarcinoma with differential response to gemcitabine-based versus FOLFIRINOX chemotherapy. PLoS ONE 14(10):e218642.  https://doi.org/10.1371/journal.pone.0218642 CrossRefPubMedPubMedCentralGoogle Scholar
  20. 20.
  21. 21.
    Kleesiek J, Morshuis JN, Isensee F et al (2019) Can virtual contrast enhancement in brain MRI replace gadolinium?: a feasibility study. Invest Radiol.  https://doi.org/10.1097/RLI.0000000000000583 CrossRefPubMedGoogle Scholar
  22. 22.
    Kleesiek J, Petersen J, Döring M et al (2016) Virtual raters for reproducible and objective assessments in radiology. Sci Rep.  https://doi.org/10.1038/srep25007 CrossRefPubMedPubMedCentralGoogle Scholar
  23. 23.
    Kleppe A, Albregtsen F, Vlatkovic L et al (2018) Chromatin organisation and cancer prognosis: a pan-cancer study. Lancet Oncol 19:356–369.  https://doi.org/10.1016/S1470-2045(17)30899-9 CrossRefPubMedPubMedCentralGoogle Scholar
  24. 24.
    Kudo S, Mori Y, Misawa M et al (2019) Artificial intelligence and colonoscopy: current status and future perspectives. Dig Endosc 31:363–371.  https://doi.org/10.1111/den.13340 CrossRefPubMedGoogle Scholar
  25. 25.
    Liu Y, Kohlberger T, Norouzi M et al (2018) Artificial intelligence—based breast cancer nodal metastasis detection: insights into the black box for pathologists. Arch Pathol Lab Med 143:859–868.  https://doi.org/10.5858/arpa.2018-0147-OA CrossRefPubMedGoogle Scholar
  26. 26.
    Menze BH, Jakab A, Bauer S et al (2015) The multimodal brain tumor image segmentation benchmark (BRATS). IEEE Trans Med Imaging 34:1993–2024.  https://doi.org/10.1109/TMI.2014.2377694 CrossRefPubMedGoogle Scholar
  27. 27.
    Mirsky Y, Mahler T, Shelef I, Elovici Y (2019) CT-GAN: Malicious Tampering of 3D Medical Imagery using Deep LearningGoogle Scholar
  28. 28.
    Mukherjee S (2017) A.I. versus M.D. https://www.newyorker.com/magazine/2017/04/03/ai-versus-md. Zugegriffen: 29. Aug. 2019Google Scholar
  29. 29.
    Nikolov S, Blackwell S, Mendes R et al (2018) Deep learning to achieve clinically applicable segmentation of head and neck anatomy for radiotherapyGoogle Scholar
  30. 30.
    Oxipit Equipping radiologists to reach their goals faster. https://oxipit.com. Zugegriffen: 28. Aug. 2019Google Scholar
  31. 31.
    Petersen J, Jäger PF, Isensee F et al (2019) Deep probabilistic modeling of glioma growthCrossRefGoogle Scholar
  32. 32.
    Piraud M, Wennmann M, Kintzelé L et al (2019) Towards quantitative imaging biomarkers of tumor dissemination: a multi-scale parametric modeling of multiple myeloma. Med Image Anal 57:214–225.  https://doi.org/10.1016/j.media.2019.07.001 CrossRefPubMedGoogle Scholar
  33. 33.
    Rajpurkar P, Irvin J, Ball RL et al (2018) Deep learning for chest radiograph diagnosis: a retrospective comparison of the CheXneXt algorithm to practicing radiologists. PLoS Med 15:e1002686.  https://doi.org/10.1371/journal.pmed.1002686 CrossRefPubMedPubMedCentralGoogle Scholar
  34. 34.
    Rodríguez-Ruiz A, Krupinski E, Mordang J‑J et al (2018) Detection of breast cancer with mammography: effect of an artificial intelligence support system. Radiology 290:305–314.  https://doi.org/10.1148/radiol.2018181371 CrossRefPubMedGoogle Scholar
  35. 35.
    Schelb P, Kohl S, Radtke JP, Wiesenfarth M, Kickingereder P, Bickelhaupt S et al (2019) Classification of Cancer at Prostate MRI: Deep Learning versus Clinical PI-RADS Assessment. Radiology.  https://doi.org/10.1148/radiol.2019190938 CrossRefPubMedGoogle Scholar
  36. 36.
    Shamai G, Binenbaum Y, Slossberg R et al (2019) Artificial intelligence algorithms to assess hormonal status from tissue microarrays in patients with breast cancer. JAMA Netw Open 2:e197700–e197700.  https://doi.org/10.1001/jamanetworkopen.2019.7700 CrossRefPubMedPubMedCentralGoogle Scholar
  37. 37.
    Springer S, Masica DL, Molin MD et al (2019) A multimodality test to guide the management of patients with a pancreatic cyst. Sci Transl Med 11:eaav4772.  https://doi.org/10.1126/scitranslmed.aav4772 CrossRefPubMedGoogle Scholar
  38. 38.
    Steiner D, MacDonald R, Liu Y et al (2018) Impact of deep learning assistance on the histopathologic review of lymph nodes for metastatic breast cancer. Am J Surg Pathol 42:1636–1646.  https://doi.org/10.1097/PAS.0000000000001151 CrossRefPubMedPubMedCentralGoogle Scholar
  39. 39.
    Shead S (2017) Facebook’s AI boss: „In terms of general intelligence, we’re not even close to a rat“. https://www.businessinsider.de/facebooks-ai-boss-in-terms-of-general-intelligence-were-not-even-close-to-a-rat-2017-10. Zugegriffen: 29. Aug. 2019Google Scholar
  40. 40.
    US Preventive Services Task Force (2013) Lung cancer: screening. https://www.uspreventiveservicestaskforce.org/Page/Document/UpdateSummaryFinal/lung-cancer-screening. Zugegriffen: 29. Aug. 2019Google Scholar
  41. 41.
    Wang J, Wu C‑J, Bao M‑L et al (2017) Machine learning-based analysis of MR radiomics can help to improve the diagnostic performance of PI-RADS v2 in clinically relevant prostate cancer. Eur Radiol 27:4082–4090.  https://doi.org/10.1007/s00330-017-4800-5 CrossRefPubMedGoogle Scholar
  42. 42.
    Wu J, Zhou B, Peck D et al (2018) DeepMiner: discovering interpretable representations for mammogram classification and explanationGoogle Scholar
  43. 43.
    Yue Y, Osipov A, Fraass B et al (2017) Identifying prognostic intratumor heterogeneity using pre- and post-radiotherapy 18F-FDG PET images for pancreatic cancer patients. J Gastrointest Oncol 8(138):127–138CrossRefGoogle Scholar
  44. 44.
    Zhang Z, Chen P, McGough M et al (2019) Pathologist-level interpretable whole-slide cancer diagnosis with deep learning. Nat Mach Intell 1:236–245.  https://doi.org/10.1038/s42256-019-0052-1 CrossRefGoogle Scholar

Copyright information

© Springer Medizin Verlag GmbH, ein Teil von Springer Nature 2019

Authors and Affiliations

  • Jens Kleesiek
    • 1
    • 2
    Email author
  • Jacob M. Murray
    • 1
  • Georgios Kaissis
    • 3
  • Rickmer Braren
    • 2
    • 3
  1. 1.AG Computational Radiology, Department of RadiologyGerman Cancer Research Center (DKFZ)HeidelbergDeutschland
  2. 2.German Cancer Consortium (DKTK)HeidelbergDeutschland
  3. 3.Department of Diagnostic and Interventional Radiology, School of MedicineTechnical University of MunichMünchenDeutschland

Personalised recommendations